[
https://issues.apache.org/jira/browse/FLINK-10247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16606958#comment-16606958
]
Chesnay Schepler commented on FLINK-10247:
------------------------------------------
A possibly less invasive option would be to introduce a dedicated thread-pool
into the MetricQueryService that is used for processing messages. So we keep
the MQS in the same actor system as the dispatcher, but pass all messages to
the thread-pool for processing.
> Run MetricQueryService in separate thread pool
> ----------------------------------------------
>
> Key: FLINK-10247
> URL: https://issues.apache.org/jira/browse/FLINK-10247
> Project: Flink
> Issue Type: Sub-task
> Components: Metrics
> Affects Versions: 1.5.3, 1.6.0, 1.7.0
> Reporter: Till Rohrmann
> Priority: Critical
> Fix For: 1.6.1, 1.7.0, 1.5.4
>
>
> In order to make the {{MetricQueryService}} run independently of the main
> Flink components, it should get its own dedicated thread pool assigned.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
