fc25 mock is extremely slow

Description

Now that we build fc25 rpms for Lago I noticed that the tests themselves(not mock setup) are much slower than el7/fc24. On average they took 500 seconds on el7/fc24/fc23, while on fc25 they take ~700 seconds. I don't think this is specific to Lago(or at least I see no reason ATM), so worth checking when other projects start building fc25 if it is a fc25 issue.

Activity

Show:

Nadav Goldin February 1, 2017 at 8:02 PM

After a while, unable to replicate. So closing for now.

Nadav Goldin December 26, 2016 at 9:05 AM

Of course this needs some more testing. I didn't say it was related to mock, but actually to fc25 itself.

Barak Korren December 26, 2016 at 9:00 AM

For a patch that does nothing:

When using mock cache:
http://jenkins.ovirt.org/job/jenkins_master_check-patch-fc25-x86_64/14/ Took 21 sec on vm0001.workers-phx.ovirt.org
http://jenkins.ohttp://jenkins.ovirt.org/job/jenkins_master_check-patch-el7-x86_64/346/virt.org/job/jenkins_master_check-patch-el7-x86_64/346/ Took 21 sec on vm0090.workers-phx.ovirt.org
http://jenkins.ovirt.org/job/jenkins_master_check-patch-fc24-x86_64/171/ Took 29 sec on el7-vm25.phx.ovirt.org
http://jenkins.ovirt.org/job/jenkins_master_check-patch-el6-x86_64/171/ Took 21 sec on el7-vm36.phx.ovirt.org

When not using cache:
http://jenkins.ovirt.org/job/jenkins_master_check-patch-fc25-x86_64/13/ Took 1 min 46 sec on vm0001.workers-phx.ovirt.org
http://jenkins.ovirt.org/job/jenkins_master_check-patch-el7-x86_64/346/ Took 1 min 30 sec on vm0090.workers-phx.ovirt.org

So nothing seems inherently and significantly slower so far.

Barak Korren December 26, 2016 at 8:35 AM

This probably is specific to Lago, or, more precisely, to using libvirt/qemu/kvm.

More data is needed, need to compare simple loop or a simple i/o test running inside fc25 v.s. an other OS.

Fixed

Details

Assignee

Reporter

Priority

Created December 26, 2016 at 8:28 AM
Updated February 28, 2017 at 2:33 PM
Resolved February 1, 2017 at 8:02 PM