Last modified: 2014-10-24 18:00:29 UTC

Wikimedia Bugzilla is closed!

Wikimedia migrated from Bugzilla to Phabricator. Bug reports are handled in Wikimedia Phabricator.
This static website is read-only and for historical purposes. It is not possible to log in and except for displaying bug reports and their history, links might be broken. See T72049, the corresponding Phabricator task for complete and up-to-date bug report information.
Bug 70049 - Increase number of Jenkins slaves to spread load and prevent browser test failures on beta
Increase number of Jenkins slaves to spread load and prevent browser test fai...
Status: NEW
Product: Wikimedia Labs
Classification: Unclassified
Infrastructure (Other open bugs)
unspecified
All All
: Unprioritized normal
: ---
Assigned To: Nobody - You can work on this!
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2014-08-26 16:45 UTC by Chris McMahon
Modified: 2014-10-24 18:00 UTC (History)
11 users (show)

See Also:
Web browser: ---
Mobile Platform: ---
Assignee Huggle Beta Tester: ---


Attachments

Description Chris McMahon 2014-08-26 16:45:40 UTC
per Antoine: 

Have labs ops raise our quota of cpu/mem/disk
	
then create a bunch of 1 CPU slaves dedicated to browser tests
if sauce is 10 VM, we can go with 10 more CPU in labs project and 10 dedicaed jenkins slaves. They are fairly easy to setup
Comment 1 Antoine "hashar" Musso (WMF) 2014-08-26 16:47:52 UTC
Moving to 'infrastructure' component.

Have to poke Andrew / Marc-André to figure out whether the quota for the 'integration' project can be raised to let us create all those additional VM.   I have no clue what kind of stress it is going to cause on labs hardware pool.

We might have to allocate some budget to throw more hardware to the labs infrastructure supporting all the instances.
Comment 2 Andrew Bogott 2014-08-26 16:52:03 UTC
What are your current quotas, and what do you need them raised to?  It should be much of a problem in any case, overprovisioning of CPUs works fine and doesn't much matter unless all the VMs are maxing out their processors at the same time.
Comment 3 Andrew Bogott 2014-08-26 16:53:10 UTC
Um... shouldn't.  shouldn't be much of a problem.
Comment 4 Antoine "hashar" Musso (WMF) 2014-08-26 21:36:22 UTC
The current quotas are:

Cores: 25/30
RAM: 51200/102400
Floating IPs: 0/0
Instances: 9/12
Security Groups: 0/10

Ref: https://wikitech.wikimedia.org/w/index.php?title=Special:NovaProject&action=displayquotas&projectname=integration


For the context, we are now running one browser test max per instance. Currently that means only three jobs run in parallel out of 40 or 50 jobs to run, some very long.

During our weekly meeting, Chris proposed to have moaaar instances created so we can run more jobs in parallel.

We have 5 CPU left, if we can get the # of instances bumped to 9 + 5 = 14, that will let us create 5 m1.small (1 CPU, 2GB RAM, 20GB disk).

We can raise the CPU and # of instances by 5 later on if we need to double the number of browser tests executors.
Comment 5 Antoine "hashar" Musso (WMF) 2014-08-26 21:37:50 UTC
Or we can just bump core to 40 and instances to 20. Giving us some more room.

Andrew: if you need budget to throw more hardware, I am sure we can get some since beta / CI are heavily relying on the labs infrastructure now.
Comment 6 Greg Grossmeier 2014-10-22 00:12:30 UTC
More hardware coming: https://rt.wikimedia.org/Ticket/Display.html?id=8244
Comment 7 Krinkle 2014-10-24 18:00:29 UTC
I created two extra slaves last month. Bringing us to a total of eight. 4 Ubuntu Precise, 4 Ubuntu Trusty.

Note You need to log in before you can comment on or make changes to this bug.


Navigation
Links