Page MenuHomePhabricator

Bump labs quota for 'integration' project
Closed, ResolvedPublic

Description

We have recently switched MediaWiki from PHP 5.3 to PHP 5.5. As such the workload as been shifted from Precise instances to Trusty instances.

To accommodate for the new workload, we have added six labs instances ci.medium (2 CPU, 2GB RAM). Turns out there is not enough memory to run two jobs in parallel (we had a bunch of oom errors). Thus we changed the Jenkins slaves to only run one job.

To get more jobs run in parallel, we would need to add 8 more c1.medium slaves instances to the integration project. We have reached the quota, and this task is about raising it.

For 8 more executor slots we would need 8 ci.medium or:

Quotaci.medium8 of them
CPU216
RAM2GB16GB
Disk40GB320GB

We would need:

  • CPU quota raised to 97 (85 + 12)
  • instance quota raised to 37 (29 + 8)

The quota summary is:

StuffCurrentNewDelta
Cores81/8597/97+ 16
RAM151552/204800~ 167+ 16 GB
Instances29/2937+8

Not sure whether labs infra can handle it. If you get enough resources, maybe 100 CPU and 40 instances quota?

Event Timeline

Where would we change this so it is raised.

@Paladox: this task is for the WMF Labs admins to take care of. Quotas are handled in the OpenStackManager on wikitech (iow: not in Gerrit) afaiui.

Andrew claimed this task.
Andrew added a subscriber: Andrew.

done!

Wonderful @Andrew \O/ Let us know if integration/contintcloud cause too much stress on labs infra.