[jira] [Commented] (SOLR-12923) The new AutoScaling tests are way to flaky and need special attention.

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Commented] (SOLR-12923) The new AutoScaling tests are way to flaky and need special attention.

JIRA jira@apache.org

    [ https://issues.apache.org/jira/browse/SOLR-12923?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16794010#comment-16794010 ]

ASF subversion and git services commented on SOLR-12923:
--------------------------------------------------------

Commit f1e37f44175e10db7abfc88590e695374af59ff3 in lucene-solr's branch refs/heads/branch_8_0 from Chris M. Hostetter
[ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=f1e37f4 ]

SOLR-12923: Fix some issues w/concurrency and exception swallowing in SimClusterStateProvider/SimCloudManager

There are 3 tightly related bug fixes in these changes:

1) ConcurrentModificationExceptions were being thrown by some SimClusterStateProvider methods when
   creating collections/replicas due to the use of ArrayLists nodeReplicaMap. These ArrayLists were changed
   to use synchronizedList wrappers.
2) The Exceptions from #1 were being swallowed/hidden by code using SimCloudManager.submit() w/o checking
   the result of the resulting Future object. (As a result, tests waiting for a particular ClusterShape
   would timeout regardless of how long they waited.)   To protect against "silent" failures like this,
   this SimCloudManager.submit() has been updated to wrap all input Callables such that any uncaught errors
   will be logged and "counted."  SimSolrCloudTestCase will ensure a suite level failure if any such failures
   are counted.
3) The changes in #2 exposed additional concurrency problems with the Callables involved in leader election:
   These would frequently throw IllegalStateExceptions due to assumptions about the state/existence of
   replicas when the Callables were created vs when they were later run -- notably a Callable may have been
   created that held a reference to a Slice, but by the time that Callable was run the collection (or a
   node, etc...) refered to by that Slice may have been deleted.  While fixing this, the leader election
   logic was also cleaned up such that adding a replica only triggers leader election for that shard, not
   every shard in the collection.

While auditing this code, cleanup was also done to ensure all usage of SimClusterStateProvider.lock was
also cleaned up to remove all risky points where an exception may have been possible after aquiring the
lock but before the try/finally that ensured it would be unlocked.

(cherry picked from commit 76babf876a49f82959cc36a1d7ef922a9c2dddff)


> The new AutoScaling tests are way to flaky and need special attention.
> ----------------------------------------------------------------------
>
>                 Key: SOLR-12923
>                 URL: https://issues.apache.org/jira/browse/SOLR-12923
>             Project: Solr
>          Issue Type: Sub-task
>      Security Level: Public(Default Security Level. Issues are Public)
>          Components: Tests
>            Reporter: Mark Miller
>            Priority: Major
>
> I've already done some work here (not posted yet). We need to address this, these tests are too new to fail so often and easily.
> I want to add beasting to precommit (LUCENE-8545) to help prevent tests that fail so easily from being committed.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]