[
https://issues.apache.org/jira/browse/IGNITE-3862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15669937#comment-15669937
]
Dmitry Karachentsev commented on IGNITE-3862:
---------------------------------------------
Thanks Semen for review.
* Fixed javadoc.
* I'm not sure that GridServiceProxy should throw ComputeTaskTimeoutException,
in most cases it throws just IgniteException. Maybe I need to create specific
exception for that case (f.e. GridServiceAcquireTimeoutException)?
* Opened a ticket for compute serialization error
https://issues.apache.org/jira/browse/IGNITE-4231
* Use LT in GridTaskWorker and GridJobWorker to reduce amount of same errors
when GridServiceProxe retries to get service (it ignores
GridServiceNotFoundException and ClusterTopologyCheckedException).
> GridServiceProxy invocation never times out
> -------------------------------------------
>
> Key: IGNITE-3862
> URL: https://issues.apache.org/jira/browse/IGNITE-3862
> Project: Ignite
> Issue Type: Bug
> Components: managed services
> Affects Versions: 1.7
> Reporter: Valentin Kulichenko
> Assignee: Dmitry Karachentsev
> Priority: Critical
> Fix For: 1.8
>
>
> {{GridServiceProxy}} uses compute for remote invocation. In some cases an
> exception on server side can cause the closure execution never finish. For
> example, this happens when the exception is thrown during the serialization
> of the result.
> Need to add additional {{IgniteServices.serviceProxy(..)}} method that will
> additionally allow to specify custom timeout.
> This timeout should limit the number of retries (there is an infinite loop
> now) and also be passed to {{callAsyncNoFailover}} to avoid hangs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
