You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@ignite.apache.org by "Andrew Mashenkov (JIRA)" <ji...@apache.org> on 2018/09/27 13:15:00 UTC

[jira] [Created] (IGNITE-9722) MVCC: Test CacheMvccTransactionTest.testMvccCoordinatorChangeSimple fails.

Andrew Mashenkov created IGNITE-9722:
----------------------------------------

             Summary: MVCC: Test CacheMvccTransactionTest.testMvccCoordinatorChangeSimple fails.
                 Key: IGNITE-9722
                 URL: https://issues.apache.org/jira/browse/IGNITE-9722
             Project: Ignite
          Issue Type: Bug
          Components: cache, mvcc
            Reporter: Andrew Mashenkov
             Fix For: 2.7


The test make transactional putAll() and then check data with Scan query and getAll().

ScanQuery can miss some results on unstable topology which is expected as it doesn't lock topology.
But getAll() can unexpectedly return stale results.

To reproduce the issue, one can change read mode SCAN->GET (passed to checkAndGetAll() call) in checkPutGet() method. 

 

It looks like rebalance can overwrite local backup entries or remap is broken and we read local partition in moving state. Disabling fastLocGet in GridPartitionedGetFuture.map() resolves the issue.

 



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