[
https://issues.apache.org/jira/browse/TEZ-4048?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16779880#comment-16779880
]
Gopal V commented on TEZ-4048:
------------------------------
The real problem is that a huge number of events being logged aren't
significant to diagnostics - workload management in LLAP dequeuing tasks is
showing up as KILLED tasks which is why the queue is overflowing right (& the
output writer is writing to S3 right now, which is another slow point).
> Make proto history logger queue size configurable
> -------------------------------------------------
>
> Key: TEZ-4048
> URL: https://issues.apache.org/jira/browse/TEZ-4048
> Project: Apache Tez
> Issue Type: Improvement
> Affects Versions: 0.9.next
> Reporter: Prasanth Jayachandran
> Assignee: Prasanth Jayachandran
> Priority: Minor
> Attachments: TEZ-4048.1.patch
>
>
> Currently, the queue size is hard-coded to 10K which may be small for some
> bigger cluster. Make it configurable and bump up the default.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
