[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[
https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14532467#comment-14532467
]
Hive QA commented on HIVE-10595:
{color:green}Overall{color}: +1 all checks pass
Here are the results of testing the latest attachment:
https://issues.apache.org/jira/secure/attachment/12730973/HIVE-10595.1.patch
{color:green}SUCCESS:{color} +1 8912 tests passed
Test results:
http://ec2-174-129-184-35.compute-1.amazonaws.com/jenkins/job/PreCommit-HIVE-TRUNK-Build/3795/testReport
Console output:
http://ec2-174-129-184-35.compute-1.amazonaws.com/jenkins/job/PreCommit-HIVE-TRUNK-Build/3795/console
Test logs:
http://ec2-174-129-184-35.compute-1.amazonaws.com/logs/PreCommit-HIVE-TRUNK-Build-3795/
Messages:
{noformat}
Executing org.apache.hive.ptest.execution.PrepPhase
Executing org.apache.hive.ptest.execution.ExecutionPhase
Executing org.apache.hive.ptest.execution.ReportingPhase
{noformat}
This message is automatically generated.
ATTACHMENT ID: 12730973 - PreCommit-HIVE-TRUNK-Build
Dropping a table can cause NPEs in the compactor
Key: HIVE-10595
URL: https://issues.apache.org/jira/browse/HIVE-10595
Project: Hive
Issue Type: Bug
Components: Metastore
Affects Versions: 0.14.0, 1.0.0, 1.1.0
Reporter: Alan Gates
Assignee: Alan Gates
Attachments: HIVE-10595.1.patch, HIVE-10595.patch
Reproduction:
# start metastore with compactor off
# insert enough entries in a table to trigger a compaction
# drop the table
# stop metastore
# restart metastore with compactor on
Result: NPE in the compactor threads. I suspect this would also happen if
the inserts and drops were done in between a run of the compactor, but I
haven't proven it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14531154#comment-14531154 ] Eugene Koifman commented on HIVE-10595: --- I filed a ticket to investigate the more general issue. +1 this patch Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14531062#comment-14531062 ] Alan Gates commented on HIVE-10595: --- bq. So then there is nothing for Worker/Cleaner to do in this case. Unless the table get's dropped between when the initiator initiates it and the worker picks it up, same for worker- cleaner. The issue with the worker can also be seen if a user initiates compactions of a deleted or non-existent table. bq. How will data from TXNS, COMPLETED_TXN_COMPONENTS, TXN_COMPONENTS which relates to these table get cleaned up? Hmm, that might be a general issue, but it isn't introduced by this. We need to explore separately whether when compactions get dropped without being done (for whatever reason) that everything in TXNS etc. get's properly cleaned up. Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14529725#comment-14529725 ] Eugene Koifman commented on HIVE-10595: --- I'm not sure I understand how this works. The Initiator (if the table/partition is no longer there) will not add anything to compaction queue. So then there is nothing for Worker/Cleaner to do in this case. How will data from TXNS, COMPLETED_TXN_COMPONENTS, TXN_COMPONENTS which relates to these table get cleaned up? Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14529230#comment-14529230 ] Alan Gates commented on HIVE-10595: --- [~sushanth], I'd like to add this to 1.2 as it can jam the compactor. It's already patch available, so I should be able to get it in quickly. [~ekoifman], would you have time to review this? Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14529235#comment-14529235 ] Sushanth Sowmyan commented on HIVE-10595: - From the description, this is an NPE in a pretty core feature, so this qualifies as an outage. Added to https://cwiki.apache.org/confluence/display/Hive/Hive+1.2+Release+Status Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14529236#comment-14529236 ] Sushanth Sowmyan commented on HIVE-10595: - And by that, I mean +1 for inclusion to branch-1.2 Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Attachments: HIVE-10595.patch Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (HIVE-10595) Dropping a table can cause NPEs in the compactor
[ https://issues.apache.org/jira/browse/HIVE-10595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14528863#comment-14528863 ] Alan Gates commented on HIVE-10595: --- More details: # Dropping a table did not impact the initiator thread, as it could already handle not finding a table. However, it was assuming that all such cases were temp tables. # Dropping a table before the worker thread or cleaner thread got to it caused bogus error messages. In the case of the worker it also prevented forward progress on compacting any other requests, as the dropped table stayed at the head of the queue. # Dropping a partition before any of the threads got to it caused the compaction do to be done against the table rather than the partition, which would result in an error once the compaction work started. Dropping a table can cause NPEs in the compactor Key: HIVE-10595 URL: https://issues.apache.org/jira/browse/HIVE-10595 Project: Hive Issue Type: Bug Components: Metastore Affects Versions: 0.14.0, 1.0.0, 1.1.0 Reporter: Alan Gates Assignee: Alan Gates Reproduction: # start metastore with compactor off # insert enough entries in a table to trigger a compaction # drop the table # stop metastore # restart metastore with compactor on Result: NPE in the compactor threads. I suspect this would also happen if the inserts and drops were done in between a run of the compactor, but I haven't proven it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
