[jira] [Commented] (CASSANDRA-14812) Multiget Thrift query processor skips records in case of digest mismatch
[
https://issues.apache.org/jira/browse/CASSANDRA-14812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16700527#comment-16700527
]
Benedict commented on CASSANDRA-14812:
--
Thanks [~sivukhin.nikita]. Now you've confirmed, I'll work to get a patch
ready for mainline; hopefully it shouldn't take too long.
At the same time, I intend to establish what effect, if any, this has on CQL
queries. I think it may (by happy coincidence) be fine, even though the same
{{concat}} bug applies, as the limits are applied differently.
> Multiget Thrift query processor skips records in case of digest mismatch
>
>
> Key: CASSANDRA-14812
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14812
> Project: Cassandra
> Issue Type: Bug
> Components: Coordination, Core
>Reporter: Sivukhin Nikita
>Assignee: Benedict
>Priority: Critical
> Labels: bug
> Attachments: repro_script.py, requirements.txt, small_repro_script.py
>
>
> It seems that in Cassandra 3.0.0 a nasty bug was introduced in {{multiget}}
> Thrift query processing logic. When one tries to read data from several
> partitions with a single {{multiget}} query and {{DigestMismatch}} exception
> is raised during this query processing, request coordinator prematurely
> terminates response stream right at the point where the first
> \{{DigestMismatch}} error is occurring. This leads to situation where clients
> "do not see" some data contained in the database.
> We managed to reproduce this bug in all versions of Cassandra starting with
> v3.0.0. The pre-release version 3.0.0-rc2 works correctly. It looks like
> [refactoring of iterator transformation
> hierarchy|https://github.com/apache/cassandra/commit/609497471441273367013c09a1e0e1c990726ec7]
> related to CASSANDRA-9975 triggers incorrect behaviour.
> When concatenated iterator is returned from the
> [StorageProxy.fetchRows(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/service/StorageProxy.java#L1770],
> Cassandra starts to consume this combined iterator. Because of
> {{DigestMismatch}} exception some elements of this combined iterator contain
> additional {{ThriftCounter}}, that was added during
> [DataResolver.resolve(...)|https://github.com/apache/cassandra/blob/ee9e06b5a75c0be954694b191ea4170456015b98/src/java/org/apache/cassandra/service/reads/DataResolver.java#L120]
> execution. While consuming iterator for many partitions Cassandra calls
> [BaseIterator.tryGetMoreContents(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/db/transform/BaseIterator.java#L115]
> method that must switch from one partition iterator to another in case of
> exhaustion of the former. In this case all Transformations contained in the
> next iterator are applied to the combined BaseIterator that enumerates
> partitions sequence which is wrong. This behaviour causes BaseIterator to
> stop enumeration after it fully consumes partition with {{DigestMismatch}}
> error, because this partition iterator has additional {{ThriftCounter}} data
> limit.
> The attachment contains the python2 script [^small_repro_script.py] that
> reproduces this bug within 3-nodes ccmlib controlled cluster. Also, there is
> an extended version of this script - [^repro_script.py] - that contains more
> logging information and provides the ability to test behavior for many
> Cassandra versions (to run all test cases from repro_script.py you can call
> {{python -m unittest2 -v repro_script.ThriftMultigetTestCase}}). All the
> necessary dependencies contained in the [^requirements.txt]
>
> This bug is critical in our production environment because we can't permit
> any data skip.
> Any ideas about a patch for this issue?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14812) Multiget Thrift query processor skips records in case of digest mismatch
[
https://issues.apache.org/jira/browse/CASSANDRA-14812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16693207#comment-16693207
]
Sivukhin Nikita commented on CASSANDRA-14812:
-
Thank you, [~benedict] . We tested your patch against provided integration
tests and also try more tricky scenarios (for example, when digest mismatch
happened in the middle of the sequence of requested keys). In all tested cases
patch is working and the bug is not reproducing. When you plan to patch the
master?
Also, I want to clarify one issue, just for my confidence. Is it true, that the
machinery of request processing for Thrift queries is totally different than
for CQL queries? I'm asking about it because the patch is very specific and if
the same bug also contained somewhere in the CQL processing pipeline then the
patch will not help. The code surrounding response transformation and
filtration is slightly entangled for me, so I can't be sure that there is no
subtly dependency between CQL and Thrift processing pipelines that can also be
under attack of a bug.
> Multiget Thrift query processor skips records in case of digest mismatch
>
>
> Key: CASSANDRA-14812
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14812
> Project: Cassandra
> Issue Type: Bug
> Components: Coordination, Core
>Reporter: Sivukhin Nikita
>Priority: Critical
> Labels: bug
> Attachments: repro_script.py, requirements.txt, small_repro_script.py
>
>
> It seems that in Cassandra 3.0.0 a nasty bug was introduced in {{multiget}}
> Thrift query processing logic. When one tries to read data from several
> partitions with a single {{multiget}} query and {{DigestMismatch}} exception
> is raised during this query processing, request coordinator prematurely
> terminates response stream right at the point where the first
> \{{DigestMismatch}} error is occurring. This leads to situation where clients
> "do not see" some data contained in the database.
> We managed to reproduce this bug in all versions of Cassandra starting with
> v3.0.0. The pre-release version 3.0.0-rc2 works correctly. It looks like
> [refactoring of iterator transformation
> hierarchy|https://github.com/apache/cassandra/commit/609497471441273367013c09a1e0e1c990726ec7]
> related to CASSANDRA-9975 triggers incorrect behaviour.
> When concatenated iterator is returned from the
> [StorageProxy.fetchRows(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/service/StorageProxy.java#L1770],
> Cassandra starts to consume this combined iterator. Because of
> {{DigestMismatch}} exception some elements of this combined iterator contain
> additional {{ThriftCounter}}, that was added during
> [DataResolver.resolve(...)|https://github.com/apache/cassandra/blob/ee9e06b5a75c0be954694b191ea4170456015b98/src/java/org/apache/cassandra/service/reads/DataResolver.java#L120]
> execution. While consuming iterator for many partitions Cassandra calls
> [BaseIterator.tryGetMoreContents(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/db/transform/BaseIterator.java#L115]
> method that must switch from one partition iterator to another in case of
> exhaustion of the former. In this case all Transformations contained in the
> next iterator are applied to the combined BaseIterator that enumerates
> partitions sequence which is wrong. This behaviour causes BaseIterator to
> stop enumeration after it fully consumes partition with {{DigestMismatch}}
> error, because this partition iterator has additional {{ThriftCounter}} data
> limit.
> The attachment contains the python2 script [^small_repro_script.py] that
> reproduces this bug within 3-nodes ccmlib controlled cluster. Also, there is
> an extended version of this script - [^repro_script.py] - that contains more
> logging information and provides the ability to test behavior for many
> Cassandra versions (to run all test cases from repro_script.py you can call
> {{python -m unittest2 -v repro_script.ThriftMultigetTestCase}}). All the
> necessary dependencies contained in the [^requirements.txt]
>
> This bug is critical in our production environment because we can't permit
> any data skip.
> Any ideas about a patch for this issue?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14812) Multiget Thrift query processor skips records in case of digest mismatch
[
https://issues.apache.org/jira/browse/CASSANDRA-14812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16660334#comment-16660334
]
Benedict commented on CASSANDRA-14812:
--
[~sivukhin.nikita], sorry for the delay - unfortunately other things cropped up
that required my attention.
I'm awaiting the results of a clean CI run, but
[here|https://github.com/belliottsmith/cassandra/commit/111392281afa41ca2b22e8f90369f9c29a18bfd2]
is a commit that should fix the problem for you.
The only necessary changes are those to {{PartitionIterators}} and
{{BasePartitions}}
> Multiget Thrift query processor skips records in case of digest mismatch
>
>
> Key: CASSANDRA-14812
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14812
> Project: Cassandra
> Issue Type: Bug
>Reporter: Sivukhin Nikita
>Priority: Critical
> Labels: bug
> Attachments: repro_script.py, requirements.txt, small_repro_script.py
>
>
> It seems that in Cassandra 3.0.0 a nasty bug was introduced in {{multiget}}
> Thrift query processing logic. When one tries to read data from several
> partitions with a single {{multiget}} query and {{DigestMismatch}} exception
> is raised during this query processing, request coordinator prematurely
> terminates response stream right at the point where the first
> \{{DigestMismatch}} error is occurring. This leads to situation where clients
> "do not see" some data contained in the database.
> We managed to reproduce this bug in all versions of Cassandra starting with
> v3.0.0. The pre-release version 3.0.0-rc2 works correctly. It looks like
> [refactoring of iterator transformation
> hierarchy|https://github.com/apache/cassandra/commit/609497471441273367013c09a1e0e1c990726ec7]
> related to CASSANDRA-9975 triggers incorrect behaviour.
> When concatenated iterator is returned from the
> [StorageProxy.fetchRows(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/service/StorageProxy.java#L1770],
> Cassandra starts to consume this combined iterator. Because of
> {{DigestMismatch}} exception some elements of this combined iterator contain
> additional {{ThriftCounter}}, that was added during
> [DataResolver.resolve(...)|https://github.com/apache/cassandra/blob/ee9e06b5a75c0be954694b191ea4170456015b98/src/java/org/apache/cassandra/service/reads/DataResolver.java#L120]
> execution. While consuming iterator for many partitions Cassandra calls
> [BaseIterator.tryGetMoreContents(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/db/transform/BaseIterator.java#L115]
> method that must switch from one partition iterator to another in case of
> exhaustion of the former. In this case all Transformations contained in the
> next iterator are applied to the combined BaseIterator that enumerates
> partitions sequence which is wrong. This behaviour causes BaseIterator to
> stop enumeration after it fully consumes partition with {{DigestMismatch}}
> error, because this partition iterator has additional {{ThriftCounter}} data
> limit.
> The attachment contains the python2 script [^small_repro_script.py] that
> reproduces this bug within 3-nodes ccmlib controlled cluster. Also, there is
> an extended version of this script - [^repro_script.py] - that contains more
> logging information and provides the ability to test behavior for many
> Cassandra versions (to run all test cases from repro_script.py you can call
> {{python -m unittest2 -v repro_script.ThriftMultigetTestCase}}). All the
> necessary dependencies contained in the [^requirements.txt]
>
> This bug is critical in our production environment because we can't permit
> any data skip.
> Any ideas about a patch for this issue?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14812) Multiget Thrift query processor skips records in case of digest mismatch
[
https://issues.apache.org/jira/browse/CASSANDRA-14812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16651165#comment-16651165
]
Sivukhin Nikita commented on CASSANDRA-14812:
-
[~benedict] , any progress in investigating the issue? I create unit-test that
reproduce the problem in the same way as my reproducing scripts do. You can
find this test
[here|https://github.com/Umqra/cassandra/commit/1de595a6d456cc78b0e3af77607814e014f6b4ed].
> Multiget Thrift query processor skips records in case of digest mismatch
>
>
> Key: CASSANDRA-14812
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14812
> Project: Cassandra
> Issue Type: Bug
>Reporter: Sivukhin Nikita
>Priority: Critical
> Labels: bug
> Attachments: repro_script.py, requirements.txt, small_repro_script.py
>
>
> It seems that in Cassandra 3.0.0 a nasty bug was introduced in {{multiget}}
> Thrift query processing logic. When one tries to read data from several
> partitions with a single {{multiget}} query and {{DigestMismatch}} exception
> is raised during this query processing, request coordinator prematurely
> terminates response stream right at the point where the first
> \{{DigestMismatch}} error is occurring. This leads to situation where clients
> "do not see" some data contained in the database.
> We managed to reproduce this bug in all versions of Cassandra starting with
> v3.0.0. The pre-release version 3.0.0-rc2 works correctly. It looks like
> [refactoring of iterator transformation
> hierarchy|https://github.com/apache/cassandra/commit/609497471441273367013c09a1e0e1c990726ec7]
> related to CASSANDRA-9975 triggers incorrect behaviour.
> When concatenated iterator is returned from the
> [StorageProxy.fetchRows(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/service/StorageProxy.java#L1770],
> Cassandra starts to consume this combined iterator. Because of
> {{DigestMismatch}} exception some elements of this combined iterator contain
> additional {{ThriftCounter}}, that was added during
> [DataResolver.resolve(...)|https://github.com/apache/cassandra/blob/ee9e06b5a75c0be954694b191ea4170456015b98/src/java/org/apache/cassandra/service/reads/DataResolver.java#L120]
> execution. While consuming iterator for many partitions Cassandra calls
> [BaseIterator.tryGetMoreContents(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/db/transform/BaseIterator.java#L115]
> method that must switch from one partition iterator to another in case of
> exhaustion of the former. In this case all Transformations contained in the
> next iterator are applied to the combined BaseIterator that enumerates
> partitions sequence which is wrong. This behaviour causes BaseIterator to
> stop enumeration after it fully consumes partition with {{DigestMismatch}}
> error, because this partition iterator has additional {{ThriftCounter}} data
> limit.
> The attachment contains the python2 script [^small_repro_script.py] that
> reproduces this bug within 3-nodes ccmlib controlled cluster. Also, there is
> an extended version of this script - [^repro_script.py] - that contains more
> logging information and provides the ability to test behavior for many
> Cassandra versions (to run all test cases from repro_script.py you can call
> {{python -m unittest2 -v repro_script.ThriftMultigetTestCase}}). All the
> necessary dependencies contained in the [^requirements.txt]
>
> This bug is critical in our production environment because we can't permit
> any data skip.
> Any ideas about a patch for this issue?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14812) Multiget Thrift query processor skips records in case of digest mismatch
[
https://issues.apache.org/jira/browse/CASSANDRA-14812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16648189#comment-16648189
]
Benedict commented on CASSANDRA-14812:
--
Hi [~sivukhin.nikita],
Thanks for the excellent bug report. I will take a closer look at this on
Monday.
> Multiget Thrift query processor skips records in case of digest mismatch
>
>
> Key: CASSANDRA-14812
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14812
> Project: Cassandra
> Issue Type: Bug
>Reporter: Sivukhin Nikita
>Priority: Critical
> Labels: bug
> Attachments: repro_script.py, requirements.txt, small_repro_script.py
>
>
> It seems that in Cassandra 3.0.0 a nasty bug was introduced in {{multiget}}
> Thrift query processing logic. When one tries to read data from several
> partitions with a single {{multiget}} query and {{DigestMismatch}} exception
> is raised during this query processing, request coordinator prematurely
> terminates response stream right at the point where the first
> \{{DigestMismatch}} error is occurring. This leads to situation where clients
> "do not see" some data contained in the database.
> We managed to reproduce this bug in all versions of Cassandra starting with
> v3.0.0. The pre-release version 3.0.0-rc2 works correctly. It looks like
> [refactoring of iterator transformation
> hierarchy|https://github.com/apache/cassandra/commit/609497471441273367013c09a1e0e1c990726ec7]
> related to CASSANDRA-9975 triggers incorrect behaviour.
> When concatenated iterator is returned from the
> [StorageProxy.fetchRows(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/service/StorageProxy.java#L1770],
> Cassandra starts to consume this combined iterator. Because of
> {{DigestMismatch}} exception some elements of this combined iterator contain
> additional {{ThriftCounter}}, that was added during
> [DataResolver.resolve(...)|https://github.com/apache/cassandra/blob/ee9e06b5a75c0be954694b191ea4170456015b98/src/java/org/apache/cassandra/service/reads/DataResolver.java#L120]
> execution. While consuming iterator for many partitions Cassandra calls
> [BaseIterator.tryGetMoreContents(...)|https://github.com/apache/cassandra/blob/a05785d82c621c9cd04d8a064c38fd2012ef981c/src/java/org/apache/cassandra/db/transform/BaseIterator.java#L115]
> method that must switch from one partition iterator to another in case of
> exhaustion of the former. In this case all Transformations contained in the
> next iterator are applied to the combined BaseIterator that enumerates
> partitions sequence which is wrong. This behaviour causes BaseIterator to
> stop enumeration after it fully consumes partition with {{DigestMismatch}}
> error, because this partition iterator has additional {{ThriftCounter}} data
> limit.
> The attachment contains the python2 script [^small_repro_script.py] that
> reproduces this bug within 3-nodes ccmlib controlled cluster. Also, there is
> an extended version of this script - [^repro_script.py] - that contains more
> logging information and provides the ability to test behavior for many
> Cassandra versions (to run all test cases from repro_script.py you can call
> {{python -m unittest2 -v repro_script.ThriftMultigetTestCase}}). All the
> necessary dependencies contained in the [^requirements.txt]
>
> This bug is critical in our production environment because we can't permit
> any data skip.
> Any ideas about a patch for this issue?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
