[
https://issues.apache.org/jira/browse/IGNITE-4111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16719953#comment-16719953
]
Amelchev Nikita commented on IGNITE-4111:
-----------------------------------------
[~ascherbakov], Thank you for reporting this issue!
I still can't get the NPE with the reproducer and on TC it doesn't fail too
([tests
log|https://ci.ignite.apache.org/viewLog.html?buildId=2522954&tab=buildResultsDiv&buildTypeId=IgniteTests24Java8_Cache9#testNameId-3818521143170875066]).
But I have investigated the issue and problem is that anyone client tries to
connect to this node by communication SPI before disco SPI initialized
implementation. There is a question about how other nodes try to connect to a
node that is not in topology yet.
In any case, this is a bug and I have prepared the fix and direct reproducer.
This method called for compatibility resolving. In the fix, it will return
empty collections of nodes until impl initialized.
[~agoncharuk], could you take a look [at
PR|https://github.com/apache/ignite/pull/5650], please?
> Communication fails to send message if target node did not finish join process
> ------------------------------------------------------------------------------
>
> Key: IGNITE-4111
> URL: https://issues.apache.org/jira/browse/IGNITE-4111
> Project: Ignite
> Issue Type: Bug
> Components: general
> Reporter: Semen Boikov
> Assignee: Amelchev Nikita
> Priority: Minor
> Fix For: 2.8
>
> Attachments: test onFirstMessage hang.log
>
>
> Currently this scenario is possible:
> - joining node sent join request and waits for
> TcpDiscoveryNodeAddFinishedMessage inside ServerImpl.joinTopology
> - others nodes already see this node and can send messages to it (for example
> try to run compute job on this node)
> - joining node can not receive message: TcpCommunicationSpi will hang inside
> 'onFirstMessage' on 'getSpiContext' call, so sending node will get error
> trying to establish connection
> Possible fix: if in onFirstMessage() spi context is not available, then
> TcpCommunicationSpi should send special response which indicates that this
> node is not ready yet, and sender should retry after some time.
> Also need check internal code for places where message can be unnecessarily
> sent to node: one such place is
> GridCachePartitionExchangeManager.refreshPartitions - message is sent to all
> known nodes, but here we can filter by node order / finished exchage version.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
