[
https://issues.apache.org/jira/browse/IGNITE-8501?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16475977#comment-16475977
]
Stanislav Lukyanov commented on IGNITE-8501:
--------------------------------------------
The easiest way to reproduce this seems to be deploying a service with a
long-running `init()`. In that case, a client may see that the deployment has
already completed (e.g. `deployClusterSingleton` method returned), but
`GridServiceProxy::invokeMethod` will still observe a
`GridServiceNotFoundException`.
Need to add a test case like that to the
`GridServiceProxyClientReconnectSelfTest`.
> Retry on GridServiceNotFoundException in GridServiceProxy needs to be fixed
> ---------------------------------------------------------------------------
>
> Key: IGNITE-8501
> URL: https://issues.apache.org/jira/browse/IGNITE-8501
> Project: Ignite
> Issue Type: Bug
> Reporter: Stanislav Lukyanov
> Assignee: Stanislav Lukyanov
> Priority: Major
> Fix For: 2.6
>
>
> `GridServiceProxy::invokeMethod` attempts to invoke a method of an Ignite
> service and performs retries in case the invocation procedure throws
> `GridServiceNotFoundException` or `ClusterTopologyCheckedException` (these
> exceptions may be thrown in case the service assignments were already
> calculated, but the service instance was not yet created and initialized on
> the affinity server).
> After the fix IGNITE-7904 the exception type thrown by the remote invocation
> code changed to `IgniteCheckedException` (with a cause being
> `GridServiceNotFoundException` or `ClusterTopologyCheckedException`). Because
> of that, the retry doesn't work now.
> Need to fix the code to correctly handle new exception chain.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
