[jira] [Created] (OAK-4865) Test failure: HeavyWriteIT

["Michael Dürig (JIRA)" <ji...@apache.org>]

Michael Dürig created OAK-4865:
----------------------------------

             Summary: Test failure: HeavyWriteIT
                 Key: OAK-4865
                 URL: https://issues.apache.org/jira/browse/OAK-4865
             Project: Jackrabbit Oak
          Issue Type: Bug
          Components: segment-tar
            Reporter: Michael Dürig
            Assignee: Michael Dürig
            Priority: Minor
             Fix For: Segment Tar 0.0.14


Said test sometimes fails with the following stack trace:

{noformat}
09:46:40.900 ERROR [main] SegmentId.java:127                Segment not found: 8399230c-9338-47e3-acf5-b92d326cf171. SegmentId age=7473ms,gc-count=32,gc-status=success,store-generation=29,reclaim-predicate=(generation<=27),segment-generation27
org.apache.jackrabbit.oak.segment.SegmentNotFoundException: Segment 8399230c-9338-47e3-acf5-b92d326cf171 not found
at org.apache.jackrabbit.oak.segment.file.FileStore$14.call(FileStore.java:1345) ~[oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.file.FileStore$14.call(FileStore.java:1285) ~[oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.cache.CacheLIRS$Segment.load(CacheLIRS.java:1013) ~[oak-core-1.5.8.jar:1.5.8]
at org.apache.jackrabbit.oak.cache.CacheLIRS$Segment.get(CacheLIRS.java:974) ~[oak-core-1.5.8.jar:1.5.8]
at org.apache.jackrabbit.oak.cache.CacheLIRS.get(CacheLIRS.java:285) ~[oak-core-1.5.8.jar:1.5.8]
at org.apache.jackrabbit.oak.segment.SegmentCache.getSegment(SegmentCache.java:92) ~[oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.file.FileStore.readSegment(FileStore.java:1285) ~[oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentId.getSegment(SegmentId.java:123) ~[oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.Record.getSegment(Record.java:70) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentNodeState.getStableIdBytes(SegmentNodeState.java:139) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentNodeState.getStableId(SegmentNodeState.java:122) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentNodeState.fastEquals(SegmentNodeState.java:633) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentNodeStore$Commit.execute(SegmentNodeStore.java:604) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.SegmentNodeStore.merge(SegmentNodeStore.java:265) [oak-segment-tar-0.0.13-SNAPSHOT.jar:na]
at org.apache.jackrabbit.oak.segment.HeavyWriteIT.heavyWrite(HeavyWriteIT.java:85) [test-classes/:na]
{noformat}

This is a problem with the test, not a regression:

{noformat}
Segment not found: 8399230c-9338-47e3-acf5-b92d326cf171. SegmentId age=7473ms,gc-count=32,gc-status=success,store-generation=29,reclaim-predicate=(generation<=27),segment-generation27
{noformat}

This means the missing segment was successfully gc'ed at GC #32. Its generation was 27 while the store just got bumped to generation 29. This causes a cleanup of all generations <= 27. 

The test itself calls {{FileStore.gc()}} in quick succession while at the same time writing to the store. This is likely to at some point cause a write to be based on an already collected segment. I suggest to fix this by increasing the number of retained generations to a sufficiently high value (for this test). 

On a side node, this issue (and being able to to a root cause analysis) validates the additional logging that we added with OAK-2405! 




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)