Alexander Belyak created IGNITE-6451:
----------------------------------------
Summary: AssertionError: null in GridCacheIoManager.onSend on stop
Key: IGNITE-6451
URL: https://issues.apache.org/jira/browse/IGNITE-6451
Project: Ignite
Issue Type: Bug
Components: general
Affects Versions: 1.8
Reporter: Alexander Belyak
Priority: Minor
If we stop node while sending message (after GridCacheIoManager.onSend test if
grid is stopping) - we get AssertionError, for example, from:
{noformat}
java.lang.AssertionError: null
at
org.apache.ignite.internal.processors.cache.GridCacheMessage.marshalCollection(GridCacheMessage.java:481)
~[ignite-core-1.10.3.ea15-SNAPSHOT.jar:2.0.0-SNAPSHOT]
at
org.apache.ignite.internal.processors.cache.query.GridCacheQueryResponse.prepareMarshal(GridCacheQueryResponse.java:134)
~[ignite-core-1.10.3.ea15-SNAPSHOT.jar:2.0.0-SNAPSHOT]
at
org.apache.ignite.internal.processors.cache.GridCacheIoManager.onSend(GridCacheIoManager.java:917)
[ignite-core-1.10.3.ea15-SNAPSHOT.jar:2.0.0-SNAPSHOT]
{noformat}
I think we need more reliable approach to stop grid, ideally - we must stop all
activity as first step of stopping grid and go to next step only after it. Or
we can just add many tests in code like after each
cctx = ctx.getCacheContext(cacheId)
do
if (cctx == null && ...kernalContext().isStopping())
return false; //<= handle parallel stop here to correctly cancel operation
I think its important because no one can trust db with assertions in logs!
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
