[
https://issues.apache.org/jira/browse/HADOOP-2141?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12542852
]
Milind Bhandarkar commented on HADOOP-2141:
-------------------------------------------
In my past life, using 2x the average to determine outliers has worked well (I
don't know the theory behind it ;-) Regarding other questions,
We have seen stuck tasks mostly for reduces, (maybe because reduces write to
DFS), but I would prefer a uniform treatment for tasks, regardless of
map/reduce. Some streaming users do write to DFS in map tasks as side-effects
(taking care of each attempt writing to a separate file/directory). This will
help them as well.
Is there a strong reason for disabling it for maps ?
> speculative execution start up condition based on completion time
> -----------------------------------------------------------------
>
> Key: HADOOP-2141
> URL: https://issues.apache.org/jira/browse/HADOOP-2141
> Project: Hadoop
> Issue Type: Improvement
> Components: mapred
> Affects Versions: 0.15.0
> Reporter: Koji Noguchi
> Assignee: Arun C Murthy
> Fix For: 0.16.0
>
>
> We had one job with speculative execution hang.
> 4 reduce tasks were stuck with 95% completion because of a bad disk.
> Devaraj pointed out
> bq . One of the conditions that must be met for launching a speculative
> instance of a task is that it must be at least 20% behind the average
> progress, and this is not true here.
> It would be nice if speculative execution also starts up when tasks stop
> making progress.
> Devaraj suggested
> bq. Maybe, we should introduce a condition for average completion time for
> tasks in the speculative execution check.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
