[jira] Commented: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12714209#action_12714209 ] Ashutosh Chauhan commented on PIG-796: -- Since Pig allows values in a map to be of different types caching the type may not be safe. There are two possible alternatives: a) Find type by introspection every time. This will ensure we are always correct and can handle all cases (including when values in maps are of different types). This though will incur a performance overhead for every cast call. b) Find the type first time and then cache it for subsequent calls. When encountered with different type Pig will bail out with a ClassCastException. This will avoid performance overhead but Pig will die when values in maps are of different types. In this performance Vs handling all cases trade-off wondering which route should we go ? support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-796: - Attachment: pig-796.patch This patch implements the fix as suggested by Alan. support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich Attachments: pig-796.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-796: - Attachment: 796.patch updated patch. This patch fixes the following issue: Sometimes (e.g. when values coming out of map lookup) Pig assumes type of element as ByteArray when actually it is of some other type. In such cases request for a Cast fails. This patch first finds out the actual type of element before casting it (specifically when Pig thinks its ByteArray) and then do the cast. It also caches the type. When type changes ClassCastException is raised which gets caught and cast is then tried again. Cached value of type is also updated. This ensures that type is not determined on each cast call as well as handling of casts when types changes from one call to the next. support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich Attachments: 796.patch, pig-796.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-796: - Attachment: pig-796.patch Updated patch incorporating suggested changes. support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich Attachments: 796.patch, pig-796.patch, pig-796.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-796: - Attachment: (was: pig-796.patch) support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich Attachments: 796.patch, pig-796.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-796) support conversion from numeric types to chararray
[ https://issues.apache.org/jira/browse/PIG-796?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-796: - Attachment: pig-796.patch Updated patch incorporating suggested changes. support conversion from numeric types to chararray --- Key: PIG-796 URL: https://issues.apache.org/jira/browse/PIG-796 Project: Pig Issue Type: Improvement Affects Versions: 0.2.0 Reporter: Olga Natkovich Attachments: 796.patch, pig-796.patch, pig-796.patch -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Status: Patch Available (was: Open) Submitting patch Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Priority: Minor Attachments: pig-773.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12717505#action_12717505 ] Ashutosh Chauhan commented on PIG-773: -- FindBugs is complaining for starting function name with uppercase letter. All the method names in QueryParser.jjt starts with uppercase letter. So, following that convention I am leaving that function name as it is. FindBug warning: The method name org.apache.pig.impl.logicalLayer.parser.QueryParser.UnionClause(LogicalPlan) doesn't start with a lower case letter Bug type NM_METHOD_NAMING_CONVENTION In class org.apache.pig.impl.logicalLayer.parser.QueryParser In method org.apache.pig.impl.logicalLayer.parser.QueryParser.UnionClause(LogicalPlan) At QueryParser.java:[lines 2662-2713] Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Priority: Minor Attachments: pig-773.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Fix For: 0.4.0 Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820.patch In addition to explanation above SampleOptimizer is introduced which visits the compiled MR plan to detect this pattern (MR operator containing only load-store followed by MR operator containing sampling job in map plan). If this pattern is present, SampleOptimizer deletes the unnecessary predecessor MR operator and replaces the POLoad of sampling job with RandomSampleLoader which uses the loader of its predecessor. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: pig-820.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Affects Version/s: 0.4.0 Status: Patch Available (was: Open) Submitting for both 0.3 and 0.4 branches. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: pig-820.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Attachment: pig-773_v2.patch This patch doesn't have extra EmptyConstant production but rather matches for empty content of bag / tuple / map in their respective productions. As a result it avoids the unintuitive logic as Santhosh pointed above. Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Affects Version/s: (was: 0.2.0) 0.3.0 Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Patch Info: (was: [Patch Available]) trying to make hudson pick the patch Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-859) Optimizer throw error on self-joins
Optimizer throw error on self-joins --- Key: PIG-859 URL: https://issues.apache.org/jira/browse/PIG-859 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Fix For: 0.4.0 Doing self-join results in exception thrown by Optimizer. Consider the following query {code} grunt A = load 'a'; grunt B = Join A by $0, A by $0; grunt explain B; 2009-06-20 15:51:38,303 [main] ERROR org.apache.pig.tools.grunt.Grunt - ERROR 1094: Attempt to insert between two nodes that were not connected. Details at logfile: pig_1245538027026.log {code} Relevant stack-trace from log-file: {code} Caused by: org.apache.pig.impl.plan.optimizer.OptimizerException: ERROR 2047: Internal error. Unable to introduce split operators. at org.apache.pig.impl.logicalLayer.optimizer.ImplicitSplitInserter.transform(ImplicitSplitInserter.java:163) at org.apache.pig.impl.logicalLayer.optimizer.LogicalOptimizer.optimize(LogicalOptimizer.java:163) at org.apache.pig.PigServer.compileLp(PigServer.java:844) at org.apache.pig.PigServer.compileLp(PigServer.java:781) at org.apache.pig.PigServer.getStorePlan(PigServer.java:723) at org.apache.pig.PigServer.explain(PigServer.java:566) ... 8 more Caused by: org.apache.pig.impl.plan.PlanException: ERROR 1094: Attempt to insert between two nodes that were not connected. at org.apache.pig.impl.plan.OperatorPlan.doInsertBetween(OperatorPlan.java:500) at org.apache.pig.impl.plan.OperatorPlan.insertBetween(OperatorPlan.java:480) at org.apache.pig.impl.logicalLayer.optimizer.ImplicitSplitInserter.transform(ImplicitSplitInserter.java:139) ... 13 more {code} A possible workaround is: {code} grunt A = load 'a'; grunt B = load 'a'; grunt C = join A by $0, B by $0; grunt explain C; {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12722706#action_12722706 ] Ashutosh Chauhan commented on PIG-820: -- In the patch RandomSampleLoader is marked as serializable and loader field in it is marked as transient. Since loader is initialized in constructor and is used later on findbugs is complaining : This class contains a field that is updated at multiple places in the class, thus it seems to be part of the state of the class.However, since the field is marked as transient and not set in readObject or readResolve, it will contain the default value in any deserialized instance of the class. However there is no need for RandomSampleLoader to implement Serializable anyway (and thus loader to be marked as transient) because loader is reconstructed from FunSpec later on. Because of this reason, both PigStorage and BinStorage also doesnt implement serializable. Will be submitting a new patch with the required changes. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: pig-820.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v2.patch Patch which fixes findbugs warning. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Alan Gates Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Patch Available (was: Open) Submitting to hudson PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Fix Version/s: 0.4.0 Assignee: Ashutosh Chauhan (was: Alan Gates) Status: Open (was: Patch Available) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v3.patch PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12723194#action_12723194 ] Ashutosh Chauhan commented on PIG-773: -- Santhosh, thanks for the review. 1. Will be fixing it in new patch. 2. Test passes while it should fail. Seems like there is an issue how Bag handles its schema. Will be investigating it further. 3. Will include test cases which check for existence of constants in the plan. Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12723316#action_12723316 ] Ashutosh Chauhan commented on PIG-820: -- Thanks Alan and Pradeep for the review. Will be incorporating SampleOptimizer changes. Constructor of RandomSampleLoader can only take string args since it is instantiated from FuncSpec on backend. So, cant make changes to types of RandomSampleLoader constructor argument. However, instead of String having classname of loader , String version of FuncSpec can be used so that loader with correct constructor gets instantiated. Will be uploading a new patch soon. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v4.patch PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Patch Available (was: Open) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Open (was: Patch Available) Due to change in LoadFunc interface as a part of PIG-734 commit, my patch won't apply cleanly on trunk anymore. Will merge with trunk and regenerate the patch again. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Patch Available (was: Open) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v5.patch PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v6.patch PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Patch Available (was: Open) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-865) Performance: Unnnecessary computation in FRJoin
Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-865: - Attachment: pig-865.patch Patch which fixes the issue described above. A useful side-effect of it is it removes the code duplication as function getValueTuple(POLocalRearrange lr, Tuple tuple) is also present in POPackage.java Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-865.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-865: - Status: Patch Available (was: Open) Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-865.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12724921#action_12724921 ] Ashutosh Chauhan commented on PIG-865: -- Patch contains no new unit-tests as it neither introduces new functionality nor modifies the existing one. Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-865.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Status: Open (was: Patch Available) Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch, pig-773_v3.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Attachment: pig-773_v3.patch It turned out that there was a bug in DataType.java where schema for a Bag is computed. The patch fixes the bug. Test cases are modified to match the expected behavior. Also the values generated by the parser are checked against expected values for the parsed constants. Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-773.patch, pig-773_v2.patch, pig-773_v3.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Fix Version/s: 0.4.0 Status: Patch Available (was: Open) Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-773.patch, pig-773_v2.patch, pig-773_v3.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12725368#action_12725368 ] Ashutosh Chauhan commented on PIG-865: -- Thanks for the review, Pradeep. As I was looking into code, I also found that bags used to hold replicate contents are recreated everytime, instead same bag object can be cleared and used again, thus minimizing object overhead. In the extreme case where every value of join key is different for every tuple (of replicate) but matches with tuples of fragment, we will end up creating as many bags as there are tuples where one bag would do. Will include this change and upload new patch. Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-865.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Status: Patch Available (was: Open) Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-773.patch, pig-773_v2.patch, pig-773_v3.patch, pig-773_v4.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Reopened: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan reopened PIG-820: -- Samplable interface introduced as a part of this patch enforces the contract of implementing getPosition() and next() on the loaders implementing it. An additional requirement for a loader to be a sampler is that they should correctly handle getNext() without knowing the position in the file. Current patch doesn't include this contract as a part of interface. That should be a part of the interface. Reopening the jira because of this issue. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-865: - Status: Patch Available (was: Open) Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-865.patch, pig-865_v2.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v7.patch Submitting the patch for review. Currently running tests. Will update the jira with the result. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch, pig-820_v7.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Attachment: pig-820_v8.patch Thanks Pradeep for the review. skip(1) is not required because reading a byte (by calling in.read()) would result in pointer getting advanced by 1. I updated that comment in the interface noting the fact that loader implementing the interface should not assume that current read position is at the beginning of a tuple. PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch, pig-820_v7.patch, pig-820_v8.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-820) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader
[ https://issues.apache.org/jira/browse/PIG-820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-820: - Status: Patch Available (was: Reopened) PERFORMANCE: The RandomSampleLoader should be changed to allow it subsume another loader - Key: PIG-820 URL: https://issues.apache.org/jira/browse/PIG-820 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0, 0.4.0 Reporter: Alan Gates Assignee: Ashutosh Chauhan Fix For: 0.4.0 Attachments: pig-820.patch, pig-820_v2.patch, pig-820_v3.patch, pig-820_v4.patch, pig-820_v5.patch, pig-820_v6.patch, pig-820_v7.patch, pig-820_v8.patch Currently a sampling job requires that data already be stored in BinaryStorage format, since RandomSampleLoader extends BinaryStorage. For order by this has mostly been acceptable, because users tend to use order by at the end of their script where other MR jobs have already operated on the data and thus it is already being stored in BinaryStorage. For pig scripts that just did an order by, an entire MR job is required to read the data and write it out in BinaryStorage format. As we begin work on join algorithms that will require sampling, this requirement to read the entire input and write it back out will not be acceptable. Join is often the first operation of a script, and thus is much more likely to trigger this useless up front translation job. Instead RandomSampleLoader can be changed to subsume an existing loader, using the user specified loader to read the tuples while handling the skipping between tuples itself. This will require the subsumed loader to implement a Samplable Interface, that will look something like: {code} public interface SamplableLoader extends LoadFunc { /** * Skip ahead in the input stream. * @param n number of bytes to skip * @return number of bytes actually skipped. The return semantics are * exactly the same as {...@link java.io.InpuStream#skip(long)} */ public long skip(long n) throws IOException; /** * Get the current position in the stream. * @return position in the stream. */ public long getPosition() throws IOException; } {code} The MRCompiler would then check if the loader being used to load data implemented the SamplableLoader interface. If so, rather than create an initial MR job to do the translation it would create the sampling job, having RandomSampleLoader use the user specified loader. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-773) Empty complex constants (empty bag, empty tuple and empty map) should be supported
[ https://issues.apache.org/jira/browse/PIG-773?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-773: - Attachment: pig-773_v5.patch Updated patch. Empty complex constants (empty bag, empty tuple and empty map) should be supported -- Key: PIG-773 URL: https://issues.apache.org/jira/browse/PIG-773 Project: Pig Issue Type: Bug Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.4.0 Attachments: pig-773.patch, pig-773_v2.patch, pig-773_v3.patch, pig-773_v4.patch, pig-773_v5.patch We should be able to create empty bag constant using {}, empty tuple constant using (), empty map constant using [] within a pig script -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-513) PERFORMANCE: optimize some of the code in DefaultTuple
[ https://issues.apache.org/jira/browse/PIG-513?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-513: - Status: Patch Available (was: Reopened) PERFORMANCE: optimize some of the code in DefaultTuple -- Key: PIG-513 URL: https://issues.apache.org/jira/browse/PIG-513 Project: Pig Issue Type: Bug Affects Versions: 0.2.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Attachments: PIG-513.patch, pig-513_2.patch The following areas in DefaultTuple.java can be changed: The member methods get(), set(), getType() and isNull() all call checkBounds() which is redundant call since all these 4 functions throw ExecException. Instead of doing a bounds check, we can catch the IndexOutOfBounds exception in a try-catch and throw it as an ExecException The write() method has the following unused object (d in the code below): {code} for (int i = 0; i sz; i++) { try { Object d = get(i); } catch (ExecException ee) { throw new RuntimeException(ee); } DataReaderWriter.writeDatum(out, mFields.get(i)); } {code} {noformat} The get(i) call in the try should be replaced by the writeDatum call directly since d is never used and there is an unncessary call to get() {noformat} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-845: - Attachment: merge-join-for-review.patch Initial patch for review. PERFORMANCE: Merge Join --- Key: PIG-845 URL: https://issues.apache.org/jira/browse/PIG-845 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Attachments: merge-join-for-review.patch Thsi join would work if the data for both tables is sorted on the join key. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Assigned: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan reassigned PIG-845: Assignee: Ashutosh Chauhan PERFORMANCE: Merge Join --- Key: PIG-845 URL: https://issues.apache.org/jira/browse/PIG-845 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Ashutosh Chauhan Attachments: merge-join-1.patch, merge-join-for-review.patch Thsi join would work if the data for both tables is sorted on the join key. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12741733#action_12741733 ] Ashutosh Chauhan commented on PIG-845: -- Hi Dmitriy, Thanks for review. Please find my comments inline. 1. EndOfAllInput flags - could you add comments here about what the point of this flag is? You explain what EndOfAllInputSetter does (which is actually rather self-explanatory) but not what the meaning of the flag is and how it's used. There is a bit of an explanation in PigMapBase, but it really belongs here. EndofAllInput flag is basically a flag to indicate that on close() call of map/reduce task, run the pipeline once more. Till now it was used only by POStream, but now POMergeJoin also make use of it. 2. Could you explain the relationship between EndOfAllInput and (deleted) POStream? POStream is still there, I guess you are referring to MRStreamHandler which is deleted. Its renaming of class. Now that POMergeJoin also makes use of it, its better to give it a generic name like EndOfAllInput instead of MRStreamHandler. 3. Comments in MRCompiler alternate between referring to the left MROp as LeftMROper and curMROper. Choose one. Ya, will update the comments. 4. I am curious about the decision to throw compiler exceptions if MergeJoin requirements re number of inputs, etc, aren't satisfied. It seems like a better user experience would be to log a warning and fall back to a regular join. Ya, a good suggestion. It would be straight forward to do it while parsing (e.g. when there are more then two inputs). Though its not straight forward to do at logical to physical plan and physical to MRJobs translation time. 5. Style notes for visitMergeJoin: It's a 200-line method. Any way you can break it up into smaller components? As is, it's hard to follow. I can break it up, but that will bloat the MRCompiler class size. Better idea is to have MRCompilerHelper or some such class where all the low level helper function lives, so that MRCompiler itself is small and thus easier to read. The if statements should be broken up into multiple lines to agree with the style guides. Variable naming: you've got topPrj, prj, pkg, lr, ce, nig.. one at a time they are fine, but together in a 200-line method they are undreadable. Please consider more descriptive names. Will use more descriptive names in next patch. 6. Kind of a global comment, since it applies to more than just MergeJoin: It seems to me like we need a Builder for operators to clean up some of the new, set, set, set stuff. Having the setters return this and a Plan's add() method return the plan, would let us replace this: POProject topPrj = new POProject(new OperatorKey(scope,nig.getNextNodeId(scope))); topPrj.setColumn(1); topPrj.setResultType(DataType.TUPLE); topPrj.setOverloaded(true); rightMROpr.reducePlan.add(topPrj); rightMROpr.reducePlan.connect(pkg, topPrj); with this: POProject topPrj = new POProject(new OperatorKey(scope,nig.getNextNodeId(scope))) .setColumn(1).setResultType(DataType.TUPLE) .setOverloaded(true); rightMROpr.reducePlan.add(topPrj).connect(pkg, topPrj) I agree. At many places there are too many parameters to set. Setters should be smart and should return the object instead of being void and then this idea of chaining will help to cut down the number of lines. 7. Is the change to ListListByte keyTypes in POFRJoin related to MergeJoin or just rolled in? POFRJoin can do without this change, but to avoid code duplication, I update the POFRJoin to use ListListByte keyTypes. 8. MergeJoin break getNext() into components. I dont want to do that because it already has lots of class members which are getting updated at various places. Making those variables live in multiple functions will make logic even more harder to follow. Also, I am not sure if java compiler can always inline the private methods. I don't see you supporting Left outer joins. Plans for that? At least document the planned approach. Ya, outer joins are currently not supported. Its documented in specification. Will include comment in code also. Error codes being declared deep inside classes, and documented on the wiki, is a poor practice, imo. They should be pulled out into PigErrors (as lightweight final objects that have an error code, a name, and a description..) I thought Santhosh made progress on this already, no? Not sure if I understand you completely. I am using ExecException, FrontEndException etc. Arent these are lightweight final objects you are referring to ? Could you explain the problem with splits and streams? Why can't this work for them? Streaming after the join will be supported. There was a bug which I fixed and will be a part of next patch. Streaming before Join will not be supported because in endOfAllInput case, streaming may potentially produce multiple tuples
[jira] Updated: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-845: - Attachment: (was: merge-join-1.patch) PERFORMANCE: Merge Join --- Key: PIG-845 URL: https://issues.apache.org/jira/browse/PIG-845 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Ashutosh Chauhan Thsi join would work if the data for both tables is sorted on the join key. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-845: - Status: Patch Available (was: Open) Running through hudson. Release audit warning can be ignored. PERFORMANCE: Merge Join --- Key: PIG-845 URL: https://issues.apache.org/jira/browse/PIG-845 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Ashutosh Chauhan Attachments: merge-join.patch, merge-join.patch Thsi join would work if the data for both tables is sorted on the join key. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-845) PERFORMANCE: Merge Join
[ https://issues.apache.org/jira/browse/PIG-845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-845: - Attachment: merge-join.patch {code} if(rightMROpr == null || rightMROpr.equals(curMROp)) throw new MRCompilerException(Successor of right input not ... {code} Do you also need to check rightMROpr == null here? I removed null check because that indicates that two preceding MROperator exists but one of them is null. This is highly unlikely and MRCompiler probably would have thrown exception while compiling those preceding physical operator. But I added the check back again in any case. If index is empty it could mean one of the following two things: 1) Data for right input only has null for join key(s) 2) right input is empty Are there any other reasons why the index would be empty? In both these cases, join output would be empty - currently the code throws an exception Should this change? A unit test where right side input is empty would be a good one to add. Exception thrown at that point is correct because if after reading index you get null object, its a bug. But there was problem dealing with empty right file nonetheless. I fixed that and added a test case for it as well. Additionally, fixed findbugs warning. Release audit warning is because of gold file addition for testing. Apache header cant be added in it. So, it can be ignored. PERFORMANCE: Merge Join --- Key: PIG-845 URL: https://issues.apache.org/jira/browse/PIG-845 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Ashutosh Chauhan Attachments: merge-join.patch, merge-join.patch Thsi join would work if the data for both tables is sorted on the join key. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-926) Merge-Join phase 2
Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Attachment: mj_phase2_1.patch The attached first patch runs the full pipeline of right side in indexer before sampling the tuple from block. This has following advantages: a) It addresses the concern which Pradeep pointed out in phase-1: Strictly we should not allow LOForeach since it could change sort order or position of join keys and hence invalidate the index - but we need it so that the Foreach introduced by the TypeCastInserter when there is a schema for either of the inputs remains. Now since pipeline is run before sampling the tuple, this becomes a non-issue. b) Currently type information doesn't make it to the POSort which sorts the index entries in reduce task of index job. This works due to other reasons, but this patch fixes this. c) It will improve on performance. Instead of always sampling the first record of the block, index now contains the entry of first record in the block for which join may happen, thus saving time spent in fetching right tuples over the network which couldn't be joined in any case. Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Attachment: (was: mj_phase2_1.patch) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Attachment: (was: mj_phase2_1.patch) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Status: Open (was: Patch Available) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Status: Open (was: Patch Available) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Attachment: (was: mj_phase2_1.patch) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Attachment: mj_phase2_1.patch Updated patch addressing Pradeep's comments. Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-926: - Status: Patch Available (was: Open) Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-926) Merge-Join phase 2
[ https://issues.apache.org/jira/browse/PIG-926?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12745510#action_12745510 ] Ashutosh Chauhan commented on PIG-926: -- Findbugs warning is about dummyTuple. A dummyTuple is used as an argument to call appropriate overloaded getNext() of physical operator. Since this is just a marker, it is initialized as null and never updated. Findbugs thinks that it will always be null, which is true, but it doesn't affect in any way. There is no workaround to get rid of this warning. Merge-Join phase 2 -- Key: PIG-926 URL: https://issues.apache.org/jira/browse/PIG-926 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: mj_phase2_1.patch This jira is created to keep track of phase-2 work for MergeJoin. Various limitations exist in phase-1 for Merge Join which are listed on: http://wiki.apache.org/pig/PigMergeJoin Those will be addressed here. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Assigned: (PIG-934) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index
[ https://issues.apache.org/jira/browse/PIG-934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan reassigned PIG-934: Assignee: Ashutosh Chauhan Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index -- Key: PIG-934 URL: https://issues.apache.org/jira/browse/PIG-934 Project: Pig Issue Type: Bug Affects Versions: 0.3.1 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Attachments: pig-934.patch We use POLoad to seek into right file which has the following code: {noformat} public void setUp() throws IOException{ String filename = lFile.getFileName(); loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); is = FileLocalizer.open(filename, pc); loader.bindTo(filename , new BufferedPositionedInputStream(is), this.offset, Long.MAX_VALUE); } {noformat} Between opening the stream and bindTo we do not seek to the right offset. bindTo itself does not perform any seek. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-934) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index
[ https://issues.apache.org/jira/browse/PIG-934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12749188#action_12749188 ] Ashutosh Chauhan commented on PIG-934: -- Seeking to an offset would only work for a single file - hence maybe have a separate function... Since open() returns an input stream it is not hard to conceive of usecase when one would want to seek into that stream even when filespec points to a directory or a glob. We have to define the semantics here. What does seeking in a directory/glob means? One reasonable answer is to view all the files in directory/glob as one big logical file and offset as an offset in this logical file and then seek into this file. Something along the lines of : {code} iterator = DataStreamIterator bytesSeen = 0; while(itertor.hasNext()){ open current file pointed by iterator bytesSeen += current file length if (bytesSeen offset) bind to adjusted offset in current file and return else continue; } {code} But since there is no requirement for such currently, we can catch the situation when seeking is asked for directory/glob and throw an exception (as is done in this patch). Later on, if we decide to support it instead of throwing exception, we can implement whatever semantics we decide on. If we create a new function with separate name it will be confusing to do these changes later on. Moreover, if there is a different function, user of the api needs to know about it and deal with it (e.g., need of special constructor in POLoad). Presence/absence of offset parameter in argument list I think is a sufficient indicator to tell which version of overloaded open() to call if there is a need for seek. Thoughts? Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index -- Key: PIG-934 URL: https://issues.apache.org/jira/browse/PIG-934 Project: Pig Issue Type: Bug Affects Versions: 0.3.1 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Attachments: pig-934.patch We use POLoad to seek into right file which has the following code: {noformat} public void setUp() throws IOException{ String filename = lFile.getFileName(); loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); is = FileLocalizer.open(filename, pc); loader.bindTo(filename , new BufferedPositionedInputStream(is), this.offset, Long.MAX_VALUE); } {noformat} Between opening the stream and bindTo we do not seek to the right offset. bindTo itself does not perform any seek. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-934) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index
[ https://issues.apache.org/jira/browse/PIG-934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-934: - Attachment: (was: pig-934.patch) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index -- Key: PIG-934 URL: https://issues.apache.org/jira/browse/PIG-934 Project: Pig Issue Type: Bug Affects Versions: 0.3.1 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Attachments: pig-934_2.patch We use POLoad to seek into right file which has the following code: {noformat} public void setUp() throws IOException{ String filename = lFile.getFileName(); loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); is = FileLocalizer.open(filename, pc); loader.bindTo(filename , new BufferedPositionedInputStream(is), this.offset, Long.MAX_VALUE); } {noformat} Between opening the stream and bindTo we do not seek to the right offset. bindTo itself does not perform any seek. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-934) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index
[ https://issues.apache.org/jira/browse/PIG-934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-934: - Status: Patch Available (was: Open) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index -- Key: PIG-934 URL: https://issues.apache.org/jira/browse/PIG-934 Project: Pig Issue Type: Bug Affects Versions: 0.3.1 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Attachments: pig-934_2.patch We use POLoad to seek into right file which has the following code: {noformat} public void setUp() throws IOException{ String filename = lFile.getFileName(); loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); is = FileLocalizer.open(filename, pc); loader.bindTo(filename , new BufferedPositionedInputStream(is), this.offset, Long.MAX_VALUE); } {noformat} Between opening the stream and bindTo we do not seek to the right offset. bindTo itself does not perform any seek. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-934) Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index
[ https://issues.apache.org/jira/browse/PIG-934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12749901#action_12749901 ] Ashutosh Chauhan commented on PIG-934: -- All tests passed on my local box. Not sure why they failed on hudson. Merge join implementation currently does not seek to right point on the right side input based on the offset provided by the index -- Key: PIG-934 URL: https://issues.apache.org/jira/browse/PIG-934 Project: Pig Issue Type: Bug Affects Versions: 0.3.1 Reporter: Pradeep Kamath Assignee: Ashutosh Chauhan Attachments: pig-934_2.patch We use POLoad to seek into right file which has the following code: {noformat} public void setUp() throws IOException{ String filename = lFile.getFileName(); loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); is = FileLocalizer.open(filename, pc); loader.bindTo(filename , new BufferedPositionedInputStream(is), this.offset, Long.MAX_VALUE); } {noformat} Between opening the stream and bindTo we do not seek to the right offset. bindTo itself does not perform any seek. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-948) [Usability] Relating pig script with MR jobs
[Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-948: - Attachment: pig-948.patch Attached is a patch which prints following information on grunt shell : {code} 09/09/07 15:11:48 INFO mapReduceLayer.MapReduceLauncher: Submitting job: job_200908291847_0046 to execution engine. 09/09/07 15:11:48 INFO mapReduceLayer.MapReduceLauncher: More information at: http://www.jobtracker-site:50030/jobdetails.jsp?jobid=job_200908291847_0046 09/09/07 15:11:48 INFO mapReduceLayer.MapReduceLauncher: To kill this job, use: kill job_200908291847_0046 {code} [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12753260#action_12753260 ] Ashutosh Chauhan commented on PIG-948: -- In this string, we are determining job-tracker address, port number and job-ids through apis, so thats fine. I agree that hardcoding other parts of url ( jobdetails.jsp?jobid= ) is not the best way to do it, as it will break the link if that web-url changes in later hadoop releases. But since there is no way to programatically get that url, I went ahead with this. If there is a way to get that url programatically, let me know. If not, I think its useful enough to have it like this and update it if it gets changed in later hadoop releases. [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-951) Reset parallelism to 1 for indexing job in MergeJoin
Reset parallelism to 1 for indexing job in MergeJoin Key: PIG-951 URL: https://issues.apache.org/jira/browse/PIG-951 Project: Pig Issue Type: Bug Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan After sampling one tuple from every block, one reducer is used to sort the index entries in reduce phase to produce sorted index to be used in actual join job. Thus, parallelism of index job should be explictly set to 1. Currently, its not. Currently, this is a non-issue, since we don't allow any blocking operators in pipeline before merge-join. However, later when we do allow blocking operators, then parallelism of indexing job will be that of preceding blocking operator. Even then, job will complete successfully because all tuple will go to only one reducer, because we are grouping on only one key all. However, it will waste cluster resources by starting all the extra reducers which get no data and thus do nothing. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-951) Reset parallelism to 1 for indexing job in MergeJoin
[ https://issues.apache.org/jira/browse/PIG-951?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-951: - Attachment: pig-951.patch One line patch which fixes this. Also, added test case to catch regression on this. Reset parallelism to 1 for indexing job in MergeJoin Key: PIG-951 URL: https://issues.apache.org/jira/browse/PIG-951 Project: Pig Issue Type: Bug Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Attachments: pig-951.patch After sampling one tuple from every block, one reducer is used to sort the index entries in reduce phase to produce sorted index to be used in actual join job. Thus, parallelism of index job should be explictly set to 1. Currently, its not. Currently, this is a non-issue, since we don't allow any blocking operators in pipeline before merge-join. However, later when we do allow blocking operators, then parallelism of indexing job will be that of preceding blocking operator. Even then, job will complete successfully because all tuple will go to only one reducer, because we are grouping on only one key all. However, it will waste cluster resources by starting all the extra reducers which get no data and thus do nothing. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-948: - Status: Patch Available (was: Open) [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-793) Improving memory efficiency of Tuple implementation
[ https://issues.apache.org/jira/browse/PIG-793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12754491#action_12754491 ] Ashutosh Chauhan commented on PIG-793: -- In addition to String Vs Text, Alan also mentioned using array instead of ArrayListObject. Did any took a look at that? I think that change should also help. When I benchmarked merge join, nearly 20-30% CPU time was spent in arraylist's operations, which should benefit a lot if an array is used instead. So, changing to arrays should help both in memory and CPU runtime at the cost of expensive appends. Also, some small benefits can be gained by very simple changes introduced in https://issues.apache.org/jira/browse/PIG-513 Improving memory efficiency of Tuple implementation --- Key: PIG-793 URL: https://issues.apache.org/jira/browse/PIG-793 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Alan Gates Currently, our tuple is a real pig and uses a lot of extra memory. There are several places where we can improve memory efficiency: (1) Laying out memory for the fields rather than using java objects since since each object for a numeric field takes 16 bytes (2) For the cases where we know the schema using Java arrays rather than ArrayList. There might be more. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-953) Enable merge join in pig to work with loaders and store functions which can internally index sorted data
[ https://issues.apache.org/jira/browse/PIG-953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12754671#action_12754671 ] Ashutosh Chauhan commented on PIG-953: -- 1. [Pradeep] zebra store function would basically needs to know the sort keys in order and which of them are asc/dsc. For this they would iterate over our data structure and require that the ordering of the keys match the primary/secondary order of the sort keys [Ashutosh] What about LinkedHashMap? It provides all the properties we are seeking here, one data structure, O(1) lookup and guaranteed iteration order. 2. In Utils.java {code} public static boolean checkNullAndClass(Object obj1, Object obj2) { return checkNullEquals(obj1, obj2, false) obj1.getClass() == obj2.getClass(); } {code} will result in NPE when both obj1 and obj2 are null. A minor detail: Suppose obj1 is declared of type ArrayListInteger and obj2 is declared of type ArrayListString, obj1.getClass() == obj2.getClass() will return true thanks to type erasure by java compiler at compile time. Not sure if thats OK or not for the check here. 3. In StoreConfig.java One of the scenarios in which SortInfo is returned as null is {code} * 3) the store follows an order by but the schema * of order by does not have column name(s) for the sort * column(s) {code} I understand that reason for this additional constraint is because SortInfo maintains list of column names. But even if schema contains only type information and not the column names, that still is a sufficient information to build indexes. Information about on which column data is sorted on can be recorded using column positions isn't it? Does zebra requires columns to be named? If it doesn't then SortInfo could be changed in such a way that it can provide column position instead of names to loader, if columns arent named. In POMergeJoin.java 4. {code} +currentFileName = lFile.getFileName(); +loader = (LoadFunc)PigContext.instantiateFuncFromSpec(lFile.getFuncSpec()); +is = FileLocalizer.open(currentFileName, offset, pc); +if (currentFileName.endsWith(.bz) || currentFileName.endsWith(.bz2)) { +is = new CBZip2InputStream((SeekableInputStream)is, 9); +} else if (currentFileName.endsWith(.gz)) { +is = new GZIPInputStream(is); +} + {code} Isnt this blocked on https://issues.apache.org/jira/browse/PIG-930 ? 5. {code} default: // We don't deal with ERR/NULL. just pass them down return res; {code} should be changed to {code} default: throwProcessingException(false,null); {code} because if status is Error, execution should be stopped and exception should be thrown as early as possible instead of continue doing work which will be wasted. If status is Null NPE will occur while doing join. 6. {code} InputStream is = FileLocalizer.open(rightInputFileName, pc); rightLoader.bindTo(rightInputFileName, new BufferedPositionedInputStream(is), 0, Long.MAX_VALUE); {code} I dont see any use of this code. I think its not required and can be removed. Infact, there is no need of following function too: {code} /** * @param rightInputFileName the rightInputFileName to set */ public void setRightInputFileName(String rightInputFileName) { this.rightInputFileName = rightInputFileName; } {code} file name of right side is obtained from index which is contained in index file. Index file is directly passed as a constructor argument of indexableLoadFunc, so there is no need of passing rightinputfilename from MRCompiler to POMergeJoin. And if this reasoning is correct then DefaultIndexableLoader.bindTo() should throw an IOException, because contract on DefaultIndexableLoader is that it is initialized with all the info it needs in constructor and then seekNear is called on it to seek to correct location. bindTo() shouldn't be used for this loader. Also, seekNear() doesn't sound right. How about seekToClosest() ? 7. I think introducing order preserving flag on logical operator is a good idea. First its self documenting as the information is contained within operator and not checked by doing instanceof else where in code. Second its a useful information which if present can help make optimizer smart decisions. As an example, optimizer can rewrite a symmetric hash join to merge-sort join if all the logical operators in query DAG from join inputs to the root has these flags set to true. Without this flag, doing such optimizations will be hard. Enable merge join in pig to work with loaders and store functions which can internally index sorted data - Key: PIG-953 URL: https://issues.apache.org/jira/browse/PIG-953 Project: Pig
[jira] Commented: (PIG-953) Enable merge join in pig to work with loaders and store functions which can internally index sorted data
[ https://issues.apache.org/jira/browse/PIG-953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12754811#action_12754811 ] Ashutosh Chauhan commented on PIG-953: -- And couple more: 8. bq. Findbugs complains about passing internal members as is in getters since the caller can then modifiy these internal members - hence the copy. {code} public ListBoolean getAscColumns() { return Utils.getCopy(ascColumns); } {code} Instead if we use following, we will achieve the same thing and then neither findbugs will complain, nor their is need for our own copy method. {code} public ListBoolean getAscColumns() { return new ArrayListBoolean(ascColumns); } {code} 9. In POMergeJoin.java {code} // we should never get here! return new Result(POStatus.STATUS_ERR, null); {code} could be changed to {code} // we should never get here! throw new ExecException(errMsg,2176); {code} because if we ever get there, it will result in NPE later on otherwise. Enable merge join in pig to work with loaders and store functions which can internally index sorted data - Key: PIG-953 URL: https://issues.apache.org/jira/browse/PIG-953 Project: Pig Issue Type: Improvement Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Attachments: PIG-953.patch Currently merge join implementation in pig includes construction of an index on sorted data and use of that index to seek into the right input to efficiently perform the join operation. Some loaders (notably the zebra loader) internally implement an index on sorted data and can perform this seek efficiently using their index. So the use of the index needs to be abstracted in such a way that when the loader supports indexing, pig uses it (indirectly through the loader) and does not construct an index. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Assigned: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
[ https://issues.apache.org/jira/browse/PIG-858?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan reassigned PIG-858: Assignee: Ashutosh Chauhan Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Attachments: pig-858.patch Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
[ https://issues.apache.org/jira/browse/PIG-858?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-858: - Attachment: pig-858.patch Patch as discussed in previous comment. Also included are test cases, where blocking operator (order-by, distinct) occurs before FRjoin. Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Attachments: pig-858.patch Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Assigned: (PIG-959) Merge Join fails when there is a blocking operator before it in query.
[ https://issues.apache.org/jira/browse/PIG-959?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan reassigned PIG-959: Assignee: Ashutosh Chauhan Merge Join fails when there is a blocking operator before it in query. -- Key: PIG-959 URL: https://issues.apache.org/jira/browse/PIG-959 Project: Pig Issue Type: Bug Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan If there is an order-by, distinct or any other blocking operator in query followed by Merge Join, pig fails to compile it. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-959) Merge Join fails when there is a blocking operator before it in query.
[ https://issues.apache.org/jira/browse/PIG-959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12755270#action_12755270 ] Ashutosh Chauhan commented on PIG-959: -- This issue is blocked on PIG-858 Merge Join fails when there is a blocking operator before it in query. -- Key: PIG-959 URL: https://issues.apache.org/jira/browse/PIG-959 Project: Pig Issue Type: Bug Reporter: Ashutosh Chauhan If there is an order-by, distinct or any other blocking operator in query followed by Merge Join, pig fails to compile it. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (PIG-959) Merge Join fails when there is a blocking operator before it in query.
Merge Join fails when there is a blocking operator before it in query. -- Key: PIG-959 URL: https://issues.apache.org/jira/browse/PIG-959 Project: Pig Issue Type: Bug Reporter: Ashutosh Chauhan If there is an order-by, distinct or any other blocking operator in query followed by Merge Join, pig fails to compile it. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-865) Performance: Unnnecessary computation in FRJoin
[ https://issues.apache.org/jira/browse/PIG-865?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12755808#action_12755808 ] Ashutosh Chauhan commented on PIG-865: -- Ouch... It should have been atleast at par if not better ! Reading the code, I can see there are more opportunities to optimize here. Currently, I am trying to get an access on M45, once I get it I will run few benchmarks and report back if I see improvements. Performance: Unnnecessary computation in FRJoin --- Key: PIG-865 URL: https://issues.apache.org/jira/browse/PIG-865 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.3.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Attachments: pig-865.patch, pig-865_v2.patch In POFRJoin implementation POLocalRearrange is used to extract join keys from the input tuples. If keys match then to perform actual join input tuples are fed to Foreach which does a cross on its inputs. After keys are extracted using POLocalRearrange output; function getValueTuple(POLocalRearrange lr, Tuple tuple) is called to reconstruct the input tuple. It seems that this function call is unnecessary since we already have input tuple at that time. This is not a bug, but since this function would get called for every tuple, if it is eliminated, it should certainly help to improve performance. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-953) Enable merge join in pig to work with loaders and store functions which can internally index sorted data
[ https://issues.apache.org/jira/browse/PIG-953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12760116#action_12760116 ] Ashutosh Chauhan commented on PIG-953: -- Changes look good. Couple of points: bq. I think this internal structure at this point does not need to be optimized for lookup Well, its less about optimization and more about maintainability. First the relationship between two parallel arrays is implicit. So, if someone is reading that code he needs to understand that relationship of his own. If there is only one structure relationship would be explicit. Second, there is quite a bit of code around it, which IMO will be simplified if a single data structure is instead used. That said, either approach works just as fine so I will leave it upto you. bq. Zebra needs column names and cannot work with positions That is then the limitation of Zebra which it should overcome someone point in time. There might be a good reason for it, but I fail to see what extra information names of column provides where type and position of columns should be sufficient. This also implies an additional requirement on user. If data is stored using ZebraStorage and if later is loaded back, then user has to provide the same names for columns that he gave while storing it. No such constraint exists for any other load-store like PigStorage. Enable merge join in pig to work with loaders and store functions which can internally index sorted data - Key: PIG-953 URL: https://issues.apache.org/jira/browse/PIG-953 Project: Pig Issue Type: Improvement Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Attachments: PIG-953-2.patch, PIG-953.patch Currently merge join implementation in pig includes construction of an index on sorted data and use of that index to seek into the right input to efficiently perform the join operation. Some loaders (notably the zebra loader) internally implement an index on sorted data and can perform this seek efficiently using their index. So the use of the index needs to be abstracted in such a way that when the loader supports indexing, pig uses it (indirectly through the loader) and does not construct an index. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12760121#action_12760121 ] Ashutosh Chauhan commented on PIG-948: -- @Daniel bq. Also I notice in many cases we cannot get first job id correctly (job id is null in this case). If I change sleepTime (MapReduceLauncher.java:100) from 500 to 1000 (ms), things look fine. Does anyone else also see that? Reason for that is JobControlCompiler compiles a set of inter-dependent MR jobs and generates a job-control object which is then submitted asynchronously to hadoop for execution. Since we dont block on those thread, its possible that job-ids are not yet assigned when we ask for them. Setting sleep time to higher value like 1000ms should be sufficient for most cases and should work. Note increasing this sleep time doesn't affect execution in anyway since we are sleeping in a thread which only does reporting. Another fool-proof though complicated approach is to sleep for shorter time duration, then check if id is assigned, if not sleep again in a while loop until ids are assigned. [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.6.0 Attachments: pig-948-2.patch, pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
[ https://issues.apache.org/jira/browse/PIG-858?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ashutosh Chauhan updated PIG-858: - Fix Version/s: 0.6.0 Affects Version/s: (was: 0.3.0) 0.4.0 Status: Patch Available (was: Open) Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Fix For: 0.6.0 Attachments: pig-858.patch Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-981) Merge join should restrict join key expressions to simple projects
[ https://issues.apache.org/jira/browse/PIG-981?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12760406#action_12760406 ] Ashutosh Chauhan commented on PIG-981: -- Default Merge Join implementation can handle order preserving join expressions, that is, when merge join itself builds the index and doesn't rely on underlying storage for index. When Merge Join doesn't build index itself, this can't be guaranteed, but then we don't have to limit all possible uses of merge-join because of this reason. Rather, we should check if Merge Join is building indexes of its own, if it is then allow order preserving expression, if it is not, only *then* restrict expressions to projections. Merge join should restrict join key expressions to simple projects -- Key: PIG-981 URL: https://issues.apache.org/jira/browse/PIG-981 Project: Pig Issue Type: Bug Affects Versions: 0.4.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Currently merge join allows join key expressions to be arbitrary expressions with the assumption that the expressions keep the sort order. Since currently only ascending sort order is supported, the code checks at run times for sort order and catches the case where sort order is broken because the join key expression is not order preserving. However there is a reason we should restrict the join keys to projection of columns only: PIG-953 will enable pig to perform merge join to work with loaders and store functions which can internally index sorted data. These store functions can only create an index (and hence lookup on the index) on raw data columns (and not expressions on the columns). Hopefully this does not downgrade the usability of merge join much since if the expressions can always be applied post join on the join columns and since the expressions are order preserving they do not affect the outcome of the join. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12760909#action_12760909 ] Ashutosh Chauhan commented on PIG-948: -- +1 for the patch. [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.6.0 Attachments: pig-948-2.patch, pig-948-3.patch, pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-948) [Usability] Relating pig script with MR jobs
[ https://issues.apache.org/jira/browse/PIG-948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12763169#action_12763169 ] Ashutosh Chauhan commented on PIG-948: -- +1 Change looks good. It should be log.info instead of log.error. In local hadoop mode, since its all running in one java process there is no port address of job tracker to get. [Usability] Relating pig script with MR jobs Key: PIG-948 URL: https://issues.apache.org/jira/browse/PIG-948 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Priority: Minor Fix For: 0.6.0 Attachments: pig-948-2.patch, pig-948-3.patch, PIG-948-4.patch, pig-948.patch Currently its hard to find a way to relate pig script with specific MR job. In a loaded cluster with multiple simultaneous job submissions, its not easy to figure out which specific MR jobs were launched for a given pig script. If Pig can provide this info, it will be useful to debug and monitor the jobs resulting from a pig script. At the very least, Pig should be able to provide user the following information 1) Job id of the launched job. 2) Complete web url of jobtracker running this job. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-953) Enable merge join in pig to work with loaders and store functions which can internally index sorted data
[ https://issues.apache.org/jira/browse/PIG-953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12763218#action_12763218 ] Ashutosh Chauhan commented on PIG-953: -- Changes look good. One comment I have: 1) In SortInfo.java#equals We have two lists and we want to check for their equality. I quickly looked up jdk sources and it seems that ArrayList doesn't override equals, so doing equals check on lists would result in reference equality test which would be incorrect. Correct way to do this would be to first check the sizes of two lists, if they are equal iterate through both lists and check equality of items at the same index in two list. Few nits: 1) TestMergeJoin contains a System.err.println which we can get rid of. 2) There are few unused imports in patch. 3) SortInfo.java#getSortColInfoList may result in Findbugs warning because of similar reason we discussed earlier in this jira. Enable merge join in pig to work with loaders and store functions which can internally index sorted data - Key: PIG-953 URL: https://issues.apache.org/jira/browse/PIG-953 Project: Pig Issue Type: Improvement Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Attachments: PIG-953-2.patch, PIG-953-3.patch, PIG-953.patch Currently merge join implementation in pig includes construction of an index on sorted data and use of that index to seek into the right input to efficiently perform the join operation. Some loaders (notably the zebra loader) internally implement an index on sorted data and can perform this seek efficiently using their index. So the use of the index needs to be abstracted in such a way that when the loader supports indexing, pig uses it (indirectly through the loader) and does not construct an index. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-953) Enable merge join in pig to work with loaders and store functions which can internally index sorted data
[ https://issues.apache.org/jira/browse/PIG-953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12763256#action_12763256 ] Ashutosh Chauhan commented on PIG-953: -- ..aah.. I should have had dug more in jdk sources. AbstractList , which ArrayList extends does override equals and provides correct behavior. So, my comment is a non-issue. With nits taken care of +1 for the patch. Enable merge join in pig to work with loaders and store functions which can internally index sorted data - Key: PIG-953 URL: https://issues.apache.org/jira/browse/PIG-953 Project: Pig Issue Type: Improvement Affects Versions: 0.3.0 Reporter: Pradeep Kamath Assignee: Pradeep Kamath Attachments: PIG-953-2.patch, PIG-953-3.patch, PIG-953.patch Currently merge join implementation in pig includes construction of an index on sorted data and use of that index to seek into the right input to efficiently perform the join operation. Some loaders (notably the zebra loader) internally implement an index on sorted data and can perform this seek efficiently using their index. So the use of the index needs to be abstracted in such a way that when the loader supports indexing, pig uses it (indirectly through the loader) and does not construct an index. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
[ https://issues.apache.org/jira/browse/PIG-858?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12765720#action_12765720 ] Ashutosh Chauhan commented on PIG-858: -- visitUnion has same changes as others visit functions, that is it adds MR Operator corresponding to POUnion in phyToMROpMap map. Real changes are in visitFRJoin. Earlier in visitFRJoin, it used to look in compiledInputs array of MROper one by one trying to match MROPer leaf PO with POFRJoin using operator key. Now, it doesn't need to do that it can simply lookup in the phyToMROpMap. Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Fix For: 0.6.0 Attachments: pig-858.patch Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-858) Order By followed by replicated join fails while compiling MR-plan from physical plan
[ https://issues.apache.org/jira/browse/PIG-858?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12765735#action_12765735 ] Ashutosh Chauhan commented on PIG-858: -- Its been a while since I did that patch. So, bit more clarification: We are interested in finding PO which corresponds to fragment PO input of POFRJoin. This PO is already compiled and is in one the MROper. Earlier we will iterate through compiledInputs array trying to match this PO with PO contained in each MROperator. This fails as discussed in previous comments. With this change, since we keep track of MR operator with each physical operator it need not to do that but can simply look up for MROper corresponding to fragment PO in the phyToMROpMap. Order By followed by replicated join fails while compiling MR-plan from physical plan --- Key: PIG-858 URL: https://issues.apache.org/jira/browse/PIG-858 Project: Pig Issue Type: Bug Components: impl Affects Versions: 0.4.0 Reporter: Ashutosh Chauhan Assignee: Ashutosh Chauhan Fix For: 0.6.0 Attachments: pig-858.patch Consider the query: {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0; explain C; {code} works. But if replicated join is used instead {code} A = load 'a'; B = order A by $0; C = join A by $0, B by $0 using replicated; explain C; {code} this fails with ERROR org.apache.pig.tools.grunt.Grunt - ERROR 2034: Error compiling operator POFRJoin relevant stacktrace: {code} Caused by: java.lang.RuntimeException: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:306) at org.apache.pig.PigServer.explain(PigServer.java:574) ... 8 more Caused by: org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompilerException: ERROR 2034: Error compiling operator POFRJoin at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:942) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.visit(POFRJoin.java:173) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:342) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:327) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.compile(MRCompiler.java:233) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.compile(MapReduceLauncher.java:301) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher.explain(MapReduceLauncher.java:278) at org.apache.pig.backend.hadoop.executionengine.HExecutionEngine.explain(HExecutionEngine.java:303) ... 9 more Caused by: java.lang.ArrayIndexOutOfBoundsException: -1 at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MRCompiler.visitFRJoin(MRCompiler.java:901) ... 16 more {code} -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-928) UDFs in scripting languages
[ https://issues.apache.org/jira/browse/PIG-928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12766750#action_12766750 ] Ashutosh Chauhan commented on PIG-928: -- 30x is indeed too slow. But, between BSF and direct bindings, I imagine direct bindings should have been more performant, since BSF adds an extra layer of translation. Isn't it ? UDFs in scripting languages --- Key: PIG-928 URL: https://issues.apache.org/jira/browse/PIG-928 Project: Pig Issue Type: New Feature Reporter: Alan Gates Attachments: package.zip It should be possible to write UDFs in scripting languages such as python, ruby, etc. This frees users from needing to compile Java, generate a jar, etc. It also opens Pig to programmers who prefer scripting languages over Java. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-928) UDFs in scripting languages
[ https://issues.apache.org/jira/browse/PIG-928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12766763#action_12766763 ] Ashutosh Chauhan commented on PIG-928: -- Though good learning from this test is BSF is not slower then direct bindings (need additional verifications though..) So, this feature could be implemented in lot less code and complexity using BSF as oppose to using different direct bindings for different languages. On the other hand, only useful language BSF supports currently is Ruby. Not sure how many people using Pig will also be interested in groovy, javascript etc.( other languages supported by BSF ). UDFs in scripting languages --- Key: PIG-928 URL: https://issues.apache.org/jira/browse/PIG-928 Project: Pig Issue Type: New Feature Reporter: Alan Gates Attachments: package.zip It should be possible to write UDFs in scripting languages such as python, ruby, etc. This frees users from needing to compile Java, generate a jar, etc. It also opens Pig to programmers who prefer scripting languages over Java. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-1025) Should be able to set job priority through Pig Latin
[ https://issues.apache.org/jira/browse/PIG-1025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12766771#action_12766771 ] Ashutosh Chauhan commented on PIG-1025: --- Useful feature. Patch looks straightforward. In your test case you are only testing whether it parses it correctly or not, I will suggest to also test whether priority is actually set in the jobconf or not. Should be able to set job priority through Pig Latin Key: PIG-1025 URL: https://issues.apache.org/jira/browse/PIG-1025 Project: Pig Issue Type: New Feature Components: grunt Affects Versions: 0.4.0 Reporter: Kevin Weil Priority: Minor Fix For: 0.6.0 Attachments: PIG-1025.patch Currently users can set the job name through Pig Latin by saying set job.name 'my job name' The ability to set the priority would also be nice, and the patch should be small. The goal is to be able to say set job.priority 'high' and throw a JobCreationException in the JobControlCompiler if the priority is not one of the allowed string values from the o.a.h.mapred.JobPriority enum: very_low, low, normal, high, very_high. Case insensitivity makes this a little nicer. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-928) UDFs in scripting languages
[ https://issues.apache.org/jira/browse/PIG-928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12766774#action_12766774 ] Ashutosh Chauhan commented on PIG-928: -- Right, I overlooked it. I think Ruby and Python are two most widely used scripting languages and both are supported by BSF. So, comparing BSF with direct bindings: 1) Performance : Initial test shows almost equal. 2) Support of multiple languages. 3) Ease of implementation To me, BSF seems to be the way to go for this, atleast the first-cut. Implementing this feature using BSF will allow us to expose this to users quickly and if many people are using it and finding one particular language to be slow then we can explore language bindings for that particular language. Thoughts? UDFs in scripting languages --- Key: PIG-928 URL: https://issues.apache.org/jira/browse/PIG-928 Project: Pig Issue Type: New Feature Reporter: Alan Gates Attachments: package.zip It should be possible to write UDFs in scripting languages such as python, ruby, etc. This frees users from needing to compile Java, generate a jar, etc. It also opens Pig to programmers who prefer scripting languages over Java. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-928) UDFs in scripting languages
[ https://issues.apache.org/jira/browse/PIG-928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12766984#action_12766984 ] Ashutosh Chauhan commented on PIG-928: -- I did some quick benchmarking using BSF approach for UDFs written in Ruby, Python, Groovy and native builtin in Pig. It's a standard wordcount example where udf tokenizes an input string into number of words. I used pig sources(src/org/apache/pig) as input which has more then 210K lines. Since, I haven't yet figured out type translation so to be consistent in experiment, I passed data as String argument and return type as Object[] in all languages. Following are the numbers I got averaged over 3 runs: ||Language|Time(seconds)|Factor|| ||Pig|17|1|| ||Ruby|155|9.1|| ||Python|178|10.4|| ||Groovy|1460|85|| This shows Groovy-BSF combo is super-slow and Ruby and Python is much better. These numbers must be seen as an absolute worst case. I believe type translations, compiling script in constructor and using the compiled version instead of evaluating script in every exec() call will give much better performance. Also, there might exist other optimizations. Sometime next week, I will try to repeat the same experiment with javax.script UDFs in scripting languages --- Key: PIG-928 URL: https://issues.apache.org/jira/browse/PIG-928 Project: Pig Issue Type: New Feature Reporter: Alan Gates Attachments: package.zip It should be possible to write UDFs in scripting languages such as python, ruby, etc. This frees users from needing to compile Java, generate a jar, etc. It also opens Pig to programmers who prefer scripting languages over Java. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-1012) FINDBUGS: SE_BAD_FIELD: Non-transient non-serializable instance field in serializable class
[ https://issues.apache.org/jira/browse/PIG-1012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12768353#action_12768353 ] Ashutosh Chauhan commented on PIG-1012: --- We just looked at POFRJoin, this might be happening at other places as well. FINDBUGS: SE_BAD_FIELD: Non-transient non-serializable instance field in serializable class --- Key: PIG-1012 URL: https://issues.apache.org/jira/browse/PIG-1012 Project: Pig Issue Type: Bug Reporter: Olga Natkovich Attachments: PIG-1012.patch SeClass org.apache.pig.backend.executionengine.PigSlice defines non-transient non-serializable instance field is SeClass org.apache.pig.backend.executionengine.PigSlice defines non-transient non-serializable instance field loader Sejava.util.zip.GZIPInputStream stored into non-transient field PigSlice.is Seorg.apache.pig.backend.datastorage.SeekableInputStream stored into non-transient field PigSlice.is Seorg.apache.tools.bzip2r.CBZip2InputStream stored into non-transient field PigSlice.is Seorg.apache.pig.builtin.PigStorage stored into non-transient field PigSlice.loader Seorg.apache.pig.backend.hadoop.DoubleWritable$Comparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigBagWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigCharArrayWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigDBAWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigDoubleWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigFloatWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigIntWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigLongWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigTupleWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigWritableComparator implements Comparator but not Serializable SeClass org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceOper defines non-transient non-serializable instance field nig SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.EqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.GreaterThanExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.GTOrEqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.LessThanExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.LTOrEqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.NotEqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.POCast defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.POProject defines non-transient non-serializable instance field bagIterator SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.POUserComparisonFunc defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.POUserFunc defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POCombinerPackage defines non-transient non-serializable instance field log SeClass
[jira] Commented: (PIG-1012) FINDBUGS: SE_BAD_FIELD: Non-transient non-serializable instance field in serializable class
[ https://issues.apache.org/jira/browse/PIG-1012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12768352#action_12768352 ] Ashutosh Chauhan commented on PIG-1012: --- Marking log in POFRJoin transient causes FRJoin to fail. Because at the backend it can't be deserialized and log.debug is used while building hashtables resulting in NPE. Either it shouldn't be marked transient or it should be instantiated in readObject() method. Stack Trace: Pig Stack Trace --- ERROR 2999: Unexpected internal error. null java.lang.NullPointerException at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.setUpHashMap(POFRJoin.java:293) at org.apache.pig.backend.hadoop.executionengine.physicalLayer.relationalOperators.POFRJoin.getNext(POFRJoin.java:197) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigMapBase.runPipeline(PigMapBase.java:249) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigMapBase.map(PigMapBase.java:240) at org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigMapOnly$Map.map(PigMapOnly.java:65) at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:50) at org.apache.hadoop.mapred.MapTask.runOldMapper(MapTask.java:358) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:307) Thanks to Tejal to pointing this out. FINDBUGS: SE_BAD_FIELD: Non-transient non-serializable instance field in serializable class --- Key: PIG-1012 URL: https://issues.apache.org/jira/browse/PIG-1012 Project: Pig Issue Type: Bug Reporter: Olga Natkovich Attachments: PIG-1012.patch SeClass org.apache.pig.backend.executionengine.PigSlice defines non-transient non-serializable instance field is SeClass org.apache.pig.backend.executionengine.PigSlice defines non-transient non-serializable instance field loader Sejava.util.zip.GZIPInputStream stored into non-transient field PigSlice.is Seorg.apache.pig.backend.datastorage.SeekableInputStream stored into non-transient field PigSlice.is Seorg.apache.tools.bzip2r.CBZip2InputStream stored into non-transient field PigSlice.is Seorg.apache.pig.builtin.PigStorage stored into non-transient field PigSlice.loader Seorg.apache.pig.backend.hadoop.DoubleWritable$Comparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigBagWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigCharArrayWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigDBAWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigDoubleWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigFloatWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigIntWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigLongWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigTupleWritableComparator implements Comparator but not Serializable Se org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.JobControlCompiler$PigWritableComparator implements Comparator but not Serializable SeClass org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceOper defines non-transient non-serializable instance field nig SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.EqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.GreaterThanExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.GTOrEqualToExpr defines non-transient non-serializable instance field log SeClass org.apache.pig.backend.hadoop.executionengine.physicalLayer.expressionOperators.LessThanExpr defines non-transient non-serializable instance field log SeClass
[jira] Commented: (PIG-1038) Optimize nested distinct/sort to use secondary key
[ https://issues.apache.org/jira/browse/PIG-1038?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12772404#action_12772404 ] Ashutosh Chauhan commented on PIG-1038: --- I think its a useful optimization. I presume this will be implemented as a visitor in MapReduceLauncher which visits on compiled MR plan. Design looks good. I have few questions: bq. 1.1 Discover if we use sort/distinct in nested foreach plan. How are you planning to discover ? Depending on some pattern like LR in map-plan followed by POPackage, POForeach, POSort in reduce-plan? Kind of orthogonal but related to this issue. We have rule-based optimizer framework in front-end, it seems to me that similar optimizer framework is required in backend too to refactor all the optimizer visitors we currently have and to add similar kind of optimizations easily in future. There are seven optimizations in front-end expressed through rules. On the other hand after addition of this one we will have nine optimization visitors in backend. May be we can think about it to avoid lot of rework every time such optimization is added. Optimize nested distinct/sort to use secondary key -- Key: PIG-1038 URL: https://issues.apache.org/jira/browse/PIG-1038 Project: Pig Issue Type: Improvement Components: impl Affects Versions: 0.4.0 Reporter: Olga Natkovich Assignee: Daniel Dai Fix For: 0.6.0 If nested foreach plan contains sort/distinct, it is possible to use hadoop secondary sort instead of SortedDataBag and DistinctDataBag to optimize the query. Eg1: A = load 'mydata'; B = group A by $0; C = foreach B { D = order A by $1; generate group, D; } store C into 'myresult'; We can specify a secondary sort on A.$1, and drop order A by $1. Eg2: A = load 'mydata'; B = group A by $0; C = foreach B { D = A.$1; E = distinct D; generate group, E; } store C into 'myresult'; We can specify a secondary sort key on A.$1, and simplify D=A.$1; E=distinct D to a special version of distinct, which does not do the sorting. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (PIG-1037) better memory layout and spill for sorted and distinct bags
[ https://issues.apache.org/jira/browse/PIG-1037?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12772410#action_12772410 ] Ashutosh Chauhan commented on PIG-1037: --- I am kinda late on this, but I would appreciate if someone can provide brief description of how this patch improves the memory layout and alleviates the spill problem. I took a quick look at the patch. According to my understanding, previously when memory is about to get exhausted Pig will start writing to the disk one tuple at a time. With this new patch, once the memory limit is hit whole bag is spilled to disk, at that point in-memory bag contains no tuples. If in-memory bag fills again, all of its content are spilled to disk in entirety again and so on.. So this patch ensures that we are not spilling one tuple at a time, but a full bag a time. Is this correct or am I missing something ? better memory layout and spill for sorted and distinct bags --- Key: PIG-1037 URL: https://issues.apache.org/jira/browse/PIG-1037 Project: Pig Issue Type: Improvement Reporter: Olga Natkovich Assignee: Ying He Fix For: 0.6.0 Attachments: PIG-1037.patch, PIG-1037.patch2, PIG-1037.patch3 -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.