[
https://issues.apache.org/jira/browse/TEZ-14?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14213147#comment-14213147
]
Bikas Saha commented on TEZ-14:
-------------------------------
bq. for ta.isSpeculationEnabled(), switch to a class member var?
I did that at first but then had to find a conf and pass it in every where. So
changed to member.
bq. private static final long MAX_WAITTING_TIME_FOR_HEARTBEAT = 9 * 1000; -
this might be problematic
The value is large enough that it should mostly not get hit. This is used to
simulate a heart beat if one is not seen for this timeout.
bq. The real test will be to run this on some larger clusters and verify it in
practice.
Getting this out in the code will let folks do this easily. And then let us
iterate on those findings.
bq. Follow-up jira which should likely be done before a 0.6 release - add
counters
Yes. Will add counters. Also the change in TEZ-1773 will help mine this
information across time for better data on usefulness.
[~sseth] Please let me know by EOD if further comments are on the way.
> Support for speculation of slow tasks
> -------------------------------------
>
> Key: TEZ-14
> URL: https://issues.apache.org/jira/browse/TEZ-14
> Project: Apache Tez
> Issue Type: Improvement
> Reporter: Bikas Saha
> Assignee: Bikas Saha
> Attachments: TEZ-14.1.patch, TEZ-14.2.patch, TEZ-14.3.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)