15:01:18 <lennyb> #startmeeting third-party
15:01:19 <openstack> Meeting started Mon Dec  4 15:01:18 2017 UTC and is due to finish in 60 minutes.  The chair is lennyb. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:22 <openstack> The meeting name has been set to 'third_party'
15:01:28 <lennyb> Hello
15:02:10 <mmedvede> o/
15:02:27 <lennyb> mmedvede, how are you nowdays?
15:03:11 <lennyb> mmedvede, anything you would like to discuss/share?
15:03:29 <mmedvede> all is well, thanks
15:04:03 <mmedvede> I wonder if any third-party CI operators started rolling out zuulv3
15:04:25 <lennyb> mmedvede, dont know. we are still in zuulv2.
15:04:50 <lennyb> do you know how much work it takes to move to v3?
15:04:57 <mmedvede> I am working on getting puppet-openstackci to better support v3
15:05:08 <mmedvede> #link https://storyboard.openstack.org/#!/story/2001367
15:05:39 <mmedvede> lennyb: I do not know how much work exactly, but I think quite a bit
15:08:03 <mmedvede> my plan is to get a test environment deployed using puppet-openstackci first
15:08:19 <lennyb> mmedvede, thanks. Do you know if I understand correctly there is no Jenkins in Zuulv3. So who runs the jobs ? Zuul? We should rewrite all JJB yamls?
15:08:51 <mmedvede> lennyb: yes, no jenkins. There are zuul executors instead that run ansible playbooks
15:09:20 <mmedvede> my understanding is superficial, as I did not really touch it with my my own hands yet
15:10:33 <lennyb> mmedvede, I see. We have in plans checking out zuul3, but I dont see it in the near future. We have 3 production environment based on zuul2.
15:11:52 <lennyb> mmedvede, anything else on this issue?
15:12:03 <mmedvede> ouch, 3 is a lot
15:12:45 <lennyb> mmedvede, o yeah :), So it takes time to maintain them
15:13:41 <mmedvede> lennyb: I have nothing else. Should mention in case any operators use puppet-openstackci and want to get to zuulv3 faster, I could use some help, probably with testing and reviewing
15:14:44 <lennyb> mmedvede, I will check if I can help you with this.
15:15:19 <lennyb> mmedvede, I dont remember if I asked you this already, do you run stable/ocata and stable/pike in your setup?
15:20:37 <mmedvede> lennyb: in setup of what?
15:20:53 <mmedvede> we run tests on stable/pike and ocata patches, yes
15:20:59 <mmedvede> our clouds are mitaka
15:22:22 <lennyb> mmedvede, I had some strange error that also tushar saw. So I added a small workaround #link https://review.openstack.org/#/c/507391/ . I wander if only us faced this issue
15:26:51 <mmedvede> lennyb: looks like we have the same error for stable/pike at least
15:27:02 <mmedvede> https://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/cinder/25/525025/1/check/devstack-plugin-ceph-dsvm-xenial/7eab392/libvirt/libvirtd.txt.gz
15:29:05 <lennyb> mmedvede, try to use my patch
15:29:23 <mmedvede> I wonder if it has something to do with using kvm acceleration (there are another errors below in the log pointing to that)
15:29:35 <mmedvede> lennyb: are you using accel=kvm or tcg?
15:29:57 <mmedvede> iirc OpenStack CI was using more of tcg
15:32:57 <lennyb> mmedvede, where should I check for accel?
15:34:39 <mmedvede> lennyb: qemu command should have that
15:35:45 <mmedvede> switching to tcg is not a good solution, but if you also use accel=kvm it could explain why upstream CI has no problems
15:36:29 <mmedvede> you can check qemu args used in libvirt/qemu/instance-* logs
15:36:47 <lennyb> mmedvede, ok, I will check it. thanks.
15:37:12 <mmedvede> I'll test your patch
15:37:34 <lennyb> mmedvede, I have accel=kvm #link http://52.169.200.208/40/521640/6/check-nova/Nova-MACVTAP-ML2-Sriov/4c15428/logs/qemu/instance-0000000a.log.gz
15:38:43 <mmedvede> good, this upstream job does use tcg http://logs.openstack.org/39/521639/6/check/legacy-tempest-dsvm-neutron-full/de7577d/logs/libvirt/qemu/instance-0000000a.txt.gz
15:40:45 <lennyb> mmedvede, I also see another libvirt error, that is not consistent "qemuDomainLookupByUUID:1569 : Domain not found:"
15:41:13 <lennyb> #link http://52.169.200.208/59/525159/2/check-mlnx/Neutron-ML2-Sriov/63491b6/logs/libvirtd.log.gz
15:42:44 <mmedvede> lennyb: I see that libvirt error in both failing and passing tests
15:43:26 <mmedvede> a lot of libvirtd errors could also be normal
15:43:52 <mmedvede> actually, this is not an error, it is just a debug info message
15:44:07 <lennyb> mmedvede, do you know how can I set accel=tcg? I want to check something
15:44:56 <lennyb> mmedvede: #link http://52.169.200.208/59/525159/2/check-mlnx/Neutron-ML2-Sriov/63491b6/logs/devstack@n-cpu.service.log.gz
15:44:57 <lennyb> libvirtError: Node device not found: no node device with matching name 'net_enp6s0f4_fa_16_3e_5e_3d_3a'
15:46:27 <mmedvede> to set accel=tcg on OpenStack level you can set the setting in nova.conf:
15:48:54 <mmedvede> [libvirt]
15:48:55 <mmedvede> virt_type=qemu
15:49:45 <mmedvede> you can trace backwards from there what you need to change in devstack-gate or devstack settings if you use either
15:49:48 <lennyb> mmedvede, thanks.
15:52:35 <lennyb> mmedvede, we are reaching end of meeting. Thanks for the tips, I will do some checking.
15:52:56 <mmedvede> thanks
15:55:23 <lennyb> mmedvede, see you next week. thanks for showing up
15:56:56 <lennyb> #endmeeting