[QA] Browser test are now throttled

[Chris McMahon]

On Sun, Sep 7, 2014 at 7:00 PM, S Page <spage at wikimedia.org> wrote:

> On Sat, Aug 30, 2014 at 5:30 AM, Antoine Musso <hashar+wmf at free.fr> wrote:
>
>> Le 30/08/2014 09:57, Željko Filipin a écrit :
>>
>> > Did it help? Do we have a graph at dashboard[1] that would confirm it?
>>
>
> Echo and Flow tests were all green Saturday Sept 6 for the first time \o/
>

Great, I've been seeing more green since the throttling also.


>
> But then 4 out of 6 Flow test runs today (
>   Sep 7, 2014 8:33:11 PM
>   Sep 7, 2014 6:51:45 PM
>   Sep 7, 2014 6:13:05 PM
>   Sep 7, 2014 3:54:07 AM
> UTC times I think) had one or two $#@!
>
> getaddrinfo: Name or service not known (SocketError)
>
> failures :-(
>
>
And I often watch the builds run.  AFAICT, even though the browser test
builds are throttled, the SocketErrors spring up when when there are a
large number of non-browser-test Jenkins build executors running.

Something I want to do this quarter is to figure out where these
performance problems exist, and figure out a better arrangement for Jenkins
master and slave hosts to support all the build executors we need to
support.

SauceLabs gave us 10 simultaneous VMs to use, I would like to use them all.
-C




> _______________________________________________
> QA mailing list
> QA at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/qa
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.wikimedia.org/pipermail/qa/attachments/20140908/11e84d54/attachment.html>