[
https://issues.apache.org/jira/browse/AURORA-1050?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14288270#comment-14288270
]
Bill Farner commented on AURORA-1050:
-------------------------------------
A bisect between 116ee2daa85aa4d0b7528b4457454d0a0debefbd and
1c53b94810e5b778ebf9c52da29f06b6a1cf7c29 indeed called out the commit i
suspected:
{noformat}
6cf0dea80d4fae5450e015da9a5be9a943ac393c is the first bad commit
commit 6cf0dea80d4fae5450e015da9a5be9a943ac393c
Author: Zameer Manji <[email protected]>
Date: Fri Jan 16 17:23:34 2015 -0800
Bump the minimum allocation for thermos to 256MB of RAM.
This increase will prevent jobs with small resource requirements from
OOMing if
there is no executor overhead.
Testing Done:
./gradlew clean test
Bugs closed: AURORA-1021
Reviewed at https://reviews.apache.org/r/30000/
:040000 040000 e20e1cee4b38ffb67f281d07f73240e393475e05
59685c8db99b4b1691ac0972d36e4c27131d0f43 M src
bisect run success
{noformat}
> test_end_to_end.sh is failing on master
> ---------------------------------------
>
> Key: AURORA-1050
> URL: https://issues.apache.org/jira/browse/AURORA-1050
> Project: Aurora
> Issue Type: Bug
> Components: Testing
> Reporter: Bill Farner
> Priority: Blocker
>
> A task is going lost with the following message:
> {noformat}
> LOST : Task uses more resources cpus(*):0.75; disk(*):64; mem(*):257;
> ports(*):[31281-31281] than available cpus(*):3.97; disk(*):19997;
> mem(*):256; ports(*):[31000-32000]
> {noformat}
> I suspect this was caused by the fix for AURORA-1021, will git-bisect and
> confirm.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)