[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Nicolas Spiegelberg updated HBASE-2312: --- Attachment: HBASE-2312.patch forgot to attach jira with the apache license grant Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 Attachments: D99.1.patch, D99.2.patch, D99.3.patch, HBASE-2312.patch There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Phabricator updated HBASE-2312: --- Attachment: D99.2.patch nspiegelberg updated the revision HBASE-2312 [jira] Possible data loss when RS goes into GC pause while rolling HLog. Reviewers: JIRA, stack Addressed Stack Ted's change requests. Integrated some fixes that Prakash found. REVISION DETAIL https://reviews.facebook.net/D99 AFFECTED FILES src/main/java/org/apache/hadoop/hbase/master/MasterFileSystem.java src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLog.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/SequenceFileLogWriter.java src/main/java/org/apache/hadoop/hbase/replication/regionserver/ReplicationSource.java src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 Attachments: D99.1.patch, D99.2.patch There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Phabricator updated HBASE-2312: --- Attachment: D99.3.patch nspiegelberg updated the revision HBASE-2312 [jira] Possible data loss when RS goes into GC pause while rolling HLog. Reviewers: JIRA, stack Minor touchups + reinsert @Ignore since 0.20.205 doesn't have the critical patches necessary. 205.1 should however REVISION DETAIL https://reviews.facebook.net/D99 AFFECTED FILES src/main/java/org/apache/hadoop/hbase/master/MasterFileSystem.java src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLog.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/SequenceFileLogWriter.java src/main/java/org/apache/hadoop/hbase/replication/regionserver/ReplicationSource.java src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 Attachments: D99.1.patch, D99.2.patch, D99.3.patch There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Phabricator updated HBASE-2312: --- Attachment: D99.1.patch nspiegelberg requested code review of HBASE-2312 [jira] Possible data loss when RS goes into GC pause while rolling HLog. Reviewers: JIRA HBASE-2312 Possible data loss when RS goes into GC pause while rolling HLog There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. TEST PLAN EMPTY REVISION DETAIL https://reviews.facebook.net/D99 AFFECTED FILES src/main/java/org/apache/hadoop/hbase/HConstants.java src/main/java/org/apache/hadoop/hbase/master/MasterFileSystem.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLog.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java src/main/java/org/apache/hadoop/hbase/regionserver/wal/SequenceFileLogWriter.java src/main/java/org/apache/hadoop/hbase/replication/regionserver/ReplicationSource.java src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLogSplit.java MANAGE HERALD DIFFERENTIAL RULES https://reviews.facebook.net/herald/view/differential/ WHY DID I GET THIS EMAIL? https://reviews.facebook.net/herald/transcript/213/ Tip: use the X-Herald-Rules header to filter Herald messages in your client. Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 Attachments: D99.1.patch There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack updated HBASE-2312: - Fix Version/s: (was: 0.92.0) 0.94.0 We don't have the hdfs support for this (0.20.205.0 or CDH). Moving out to 0.94.0. Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.94.0 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack updated HBASE-2312: - Fix Version/s: (was: 0.94.0) 0.92.0 Yes Nicolas. Pulling it back in. Let me try and apply latest version of the patch. Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack updated HBASE-2312: - Fix Version/s: (was: 0.90.2) 0.92.0 Moving to 0.92. We don't have the API in HDFS to support Nicolas's patch. Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Todd Lipcon updated HBASE-2312: --- Fix Version/s: (was: 0.90.1) 0.90.2 Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.90.0 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.90.2 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack updated HBASE-2312: - Fix Version/s: (was: 0.90.0) 0.92.0 Moving out of 0.90 after some discussion off-list between Nicolas and Jon. Not having this issue in place means possible dataloss (Nicolas)... but Jon comes back It's data loss only in case where you have a GC pause of death (long enough that the RS is expired) and then I think there's one or two more edge cases that would have to be hit, and then could be data loss. Leaving the issue critical but not a blocker on 0.90. Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.20.3 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.92.0 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Updated: (HBASE-2312) Possible data loss when RS goes into GC pause while rolling HLog
[ https://issues.apache.org/jira/browse/HBASE-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] stack updated HBASE-2312: - Priority: Critical (was: Major) Possible data loss when RS goes into GC pause while rolling HLog Key: HBASE-2312 URL: https://issues.apache.org/jira/browse/HBASE-2312 Project: HBase Issue Type: Bug Components: master, regionserver Affects Versions: 0.20.3 Reporter: Karthik Ranganathan Assignee: Nicolas Spiegelberg Priority: Critical Fix For: 0.90.0 There is a very corner case when bad things could happen(ie data loss): 1)RS #1 is going to roll its HLog - not yet created the new one, old one will get no more writes 2)RS #1 enters GC Pause of Death 3)Master lists HLog files of RS#1 that is has to split as RS#1 is dead, starts splitting 4)RS #1 wakes up, created the new HLog (previous one was rolled) and appends an edit - which is lost The following seems like a possible solution: 1)Master detects RS#1 is dead 2)The master renames the /hbase/.logs/regionserver name directory to something else (say /hbase/.logs/regionserver name-dead) 3)Add mkdir support (as opposed to mkdirs) to HDFS - so that a file create fails if the directory doesn't exist. Dhruba tells me this is very doable. 4)RS#1 comes back up and is not able create the new hlog. It restarts itself. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.