mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mesos Reviewbot Windows <revi...@mesos.apache.org>
Subject Re: Review Request 67104: Fixed a race condition in the allocator metrics.
Date Sat, 12 May 2018 02:21:33 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/67104/#review202984
-----------------------------------------------------------



PASS: Mesos patch 67104 was successfully built and tested.

Reviews applied: `['67104']`

All the build artifacts available at: http://dcos-win.westus.cloudapp.azure.com/mesos-build/review/67104

- Mesos Reviewbot Windows


On May 12, 2018, 1:02 a.m., Greg Mann wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/67104/
> -----------------------------------------------------------
> 
> (Updated May 12, 2018, 1:02 a.m.)
> 
> 
> Review request for mesos, Benjamin Mahler and Vinod Kone.
> 
> 
> Bugs: MESOS-8904
>     https://issues.apache.org/jira/browse/MESOS-8904
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This patch addresses a race condition in which the removal of
> a role from the allocator's quota sorter races with execution
> of a callback tied to a `PullGauge`. The gauge's callback
> assumed that the role would be present in the sorter, but it's
> possible for the role to be removed before the callback is
> executed.
> 
> 
> Diffs
> -----
> 
>   src/master/allocator/mesos/hierarchical.cpp 1000968be6a2935a4cac571414d7f06d7df7acf0

> 
> 
> Diff: https://reviews.apache.org/r/67104/diff/1/
> 
> 
> Testing
> -------
> 
> The test `MasterQuotaTest.RemoveSingleQuota` was modified to include a call to '/metrics/snapshot',
both with and without a framework registered in the role for which quota is set. It's not
easy to ensure that such a test provokes the race condition 100% of the time, but the test
would always expose it within a few iterations.
> 
> Before this patch, the test would crash reliably within several iterations. After this
patch, the test can be run many times with no failures.
> 
> 
> Thanks,
> 
> Greg Mann
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message