Tuesday, 2017-10-03

*** chas has joined #openstack-powervm00:00
*** edmondsw has joined #openstack-powervm01:01
*** edmondsw has quit IRC01:17
*** chhavi has joined #openstack-powervm03:15
*** esberglu has joined #openstack-powervm03:17
*** esberglu has quit IRC03:17
*** thorst has joined #openstack-powervm03:33
*** thorst has quit IRC03:50
*** esberglu has joined #openstack-powervm04:10
*** esberglu has quit IRC04:14
*** chas has quit IRC06:13
*** chas has joined #openstack-powervm06:13
*** esberglu has joined #openstack-powervm06:55
*** esberglu has quit IRC06:55
*** esberglu has joined #openstack-powervm07:48
*** esberglu has quit IRC07:53
*** esberglu has joined #openstack-powervm10:33
*** esberglu has quit IRC10:33
*** esberglu has joined #openstack-powervm11:27
*** esberglu has quit IRC11:31
*** edmondsw has joined #openstack-powervm11:50
*** thorst has joined #openstack-powervm11:58
*** thorst_ has joined #openstack-powervm12:00
*** thorst has quit IRC12:02
*** thorst_ has quit IRC12:10
*** thorst has joined #openstack-powervm12:20
*** thorst has quit IRC12:24
*** thorst has joined #openstack-powervm12:44
*** esberglu has joined #openstack-powervm12:51
esberglu#startmeeting powervm_drviver_meeting13:01
openstackMeeting started Tue Oct  3 13:01:58 2017 UTC and is due to finish in 60 minutes.  The chair is esberglu. Information about MeetBot at http://wiki.debian.org/MeetBot.13:01
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.13:02
*** openstack changes topic to " (Meeting topic: powervm_drviver_meeting)"13:02
openstackThe meeting name has been set to 'powervm_drviver_meeting'13:02
esbergluDidn't realize the recurring meetings had expired, sent out new ones just now13:03
esbergluefried: edmondsw: thorst: You guys around?13:05
esberglu#topic In-Tree Driver13:05
*** openstack changes topic to "In-Tree Driver (Meeting topic: powervm_drviver_meeting)"13:05
esbergluNova is going through the specs this weeks so hopefully they get to ours13:06
edmondswesberglu nova is supposed to be going through specs today, so you may get a ping13:06
edmondswhopefully that's just an easy approve, though, since they didn't raise any issues at the PTG13:06
edmondswwe'll see :)13:06
edmondswlet me know if you get pinged13:07
esbergluedmondsw: Yeah I'm guessing any problems will just be semantics. Will do13:07
esbergluConfig Drive is waiting on reviews13:07
esbergluI'm still working through some of the OVS test cases and cleaning it up a little13:07
esbergluThat's all for me IT13:07
edmondswyou rebased OVS so that it's not dependent on that currently-abandoned change, right?13:08
edmondswI think I owe you a review on config drive. I'll hold off on OVS until you're done cleaning it up13:08
esbergluedmondsw: Yeah OVS is based on config drive now. OVS is not ready for reviews, I'll let you know when it is13:09
edmondswI don't have anything else for IT13:10
esberglu#topic Out-of-Tree Driver13:10
*** openstack changes topic to "Out-of-Tree Driver (Meeting topic: powervm_drviver_meeting)"13:10
edmondswstill need reviews for ceph and iscsi OOT13:11
edmondswand there are some thing that need to merge if the OOT ci is working again... is it?13:11
esbergluedmondsw: Something went in since yesterday afternoon that broke stacking again13:12
edmondswof course13:12
esbergluedmondsw: Pretty typical of this stretch in the release cycle13:12
esbergluLots of changes going in13:12
edmondswesberglu yeah... once you get CI passing, ping me and I'll add rechecks on some OOT stuff13:13
edmondswif efried doesn't beat me to it :)13:13
esbergluedmondsw: Sure thing13:13
edmondswprobably all for OOT right now13:14
edmondswoh, docs...13:14
edmondswwe have a lot of docs cleanup that we need to do OOT... kind of a mess13:14
edmondswI realized that we weren't even publishing docs other than master since newton, and we got that fixed yesterday (tx esberglu, adreznec)13:15
edmondswbut our docs content is also pretty stale13:15
edmondswI've started cleaning some of it up under https://review.openstack.org/#/c/508173/13:15
edmondswthere will be other things that need to updates as well13:16
esbergluLets add a section to the etherpad with a list of outdated docs and all pitch in there as time allows13:16
*** kylek3h has joined #openstack-powervm13:16
edmondsw++ yeah, I'll add something if I didn't already13:16
esbergluI'm still targeting the IT doc updates, just been sidetracked by CI lately13:17
edmondswone of the things I noticed is that we need to be more careful as reviewers to make sure that new conf options are documented when added13:17
edmondswor we need to decide that all conf options aren't documented...13:18
edmondswand when options change, we need to check docs to make sure the docs don't need changing as well13:18
edmondswthat will all tie into the WIP review I linked13:18
edmondswenough on that for today13:19
edmondswI've got nothing else for OOT13:19
esberglu#topic Device Passthrough13:19
*** openstack changes topic to "Device Passthrough (Meeting topic: powervm_drviver_meeting)"13:19
esbergluHaven't heard from efried yet, probably because I didn't send out meeting invites in time13:20
edmondswI saw some chatter on the ML yesterday about vGPUs13:20
edmondswefried, you should read that thread if you haven't already13:20
edmondswwere referenced a few times13:21
edmondswbut I think they ended up where I hoped they would13:21
edmondswI don't have any other updates on this, so if efried isn't around...13:21
edmondswesberglu next13:23
esberglu#topic PowerVM CI13:23
*** openstack changes topic to "PowerVM CI (Meeting topic: powervm_drviver_meeting)"13:23
esbergluLike I said OOT stacking is broken again13:23
esbergluThat's #1 on my list today13:23
esbergluPlanning on knocking out a bunch of small backlog items throughout the week as able if I get it running well again13:24
esbergluNothing else to report for now13:24
edmondswI approved one of your proposals and asked a question in the other13:25
esbergluedmondsw: Yep I merged the 1st and responded to the 2nd13:25
edmondswk, will look after13:25
esberglu#topic Open Discussion13:26
*** openstack changes topic to "Open Discussion (Meeting topic: powervm_drviver_meeting)"13:26
esbergluNothing else from me13:26
edmondswI just looked at our TODOs etherpad and realized why I hadn't put OOT docs updates on it...13:26
edmondswit's for IT work13:26
edmondswI will create an OOT todos etherpad. May move some things from the IT one as appropriate13:27
edmondswand link to each other13:28
esbergluedmondsw: Yeah there is definitely some OOT stuff on the IT one13:28
edmondswsome things need to happen both OOT and IT... will need to think about where we want to cover those13:29
edmondswprobably more detail in one place or the other, but mentioned on both13:29
esbergluWe could make a general todos as well. But that seems like a lot of pages13:30
edmondswI won't do anything too drastic until efried is around to discuss13:30
esbergluedmondsw: Sounds good. That all for today?13:30
edmondswand I want to look it over and form a strong opinion of my own :)13:30
edmondswyeah, think we're done13:31
esbergluCool. Time to battle with devstack13:31
*** openstack changes topic to "This channel is for PowerVM-related development and discussion. For general OpenStack support, please use #openstack."13:31
openstackMeeting ended Tue Oct  3 13:31:34 2017 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)13:31
openstackMinutes:        http://eavesdrop.openstack.org/meetings/powervm_drviver_meeting/2017/powervm_drviver_meeting.2017-10-03-13.01.html13:31
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/powervm_drviver_meeting/2017/powervm_drviver_meeting.2017-10-03-13.01.txt13:31
openstackLog:            http://eavesdrop.openstack.org/meetings/powervm_drviver_meeting/2017/powervm_drviver_meeting.2017-10-03-13.01.log.html13:31
esbergluedmondsw: +2 from sdague on the spec13:46
edmondswesberglu great13:47
*** kylek3h has quit IRC13:59
efriedOh, hey esberglu edmondsw.  I thought it was weird that my Tuesday morning was clear.13:59
efriedSorry I missed it.14:00
*** kylek3h has joined #openstack-powervm14:00
esbergluefried: No worries its on me for not sending invites14:00
efriedI have been following the vGPU thread on the dev ML.14:00
efriedI don't have a problem with multiple etherpads as long as they're cross-linked, but agree some stuff is intertwined (like IT ports to OOT).  Use your judgment; I'm fine with whatever.14:01
efriedAnything else that needs my attention?14:02
esbergluefried: OOT CI stacking is broken again. I think I see the problem and have a fix, testing it out now. Not sure if thats blocking you anywhere14:03
efriednope.  Nothing in the world is merging anyway.14:03
efriedToday is spec review day in nova, so I'll be helping out there all day.14:04
efriedesberglu Your spec isn't showing up in the dashboard:  https://goo.gl/QidAVs14:09
esbergluefried: Maybe because sdague already hit it with a +2?14:10
efriedesberglu It should be showing up in the "needs final +2" section.14:10
esbergluefried: Wait no there's a section for that14:10
efriedWait, is it showing up for you?14:13
esbergluefried: No14:13
*** miltonm has joined #openstack-powervm14:13
esbergluIt has everything that the others do. Link to the bp in the commit, the topic is set correctly14:14
*** thorst has quit IRC14:15
efriedesberglu The bp link isn't set to point to the spec.  But I don't see how that could affect the dashboard.14:18
esbergluefried: You mean the bp link on this page?14:20
efriedesberglu Here: https://blueprints.launchpad.net/nova/+spec/powervm-nova-it-compute-driver  where it says "set the URL for this specification"14:22
edmondswefried I think I see why it's not showing up for us...14:22
edmondswyou reviewed it, so it wouldn't show for you14:22
efriedI thought that too; but esberglu didn't review it.14:22
edmondswthere's also NOT+owner%3Aself so it wouldn't show for esberglu14:22
edmondswI bet it shows for everyone but the 3 of us and sdague14:23
efriedbut yeah, if I take that out, it shows up.  So fine.14:24
*** thorst has joined #openstack-powervm14:31
*** esberglu has quit IRC14:49
*** esberglu has joined #openstack-powervm14:49
edmondswesberglu do you ever compare neo-os-ci/ci-ansible/roles/devstack-compute/templates/local.conf.j2 and powervm-ci/devstack/master/outoftree/local.conf ?14:53
edmondswwonder if any of the changes we've made in the latter would be useful in the former14:53
edmondswobviously they can't be identical14:53
*** esberglu has quit IRC14:53
*** thorst has quit IRC15:10
*** thorst has joined #openstack-powervm15:11
*** thorst has quit IRC15:11
*** esberglu has joined #openstack-powervm15:20
esbergluefried: edmondsw: https://review.openstack.org/#/c/509208/15:32
esberglu^ OOT stacking fix15:33
edmondswesberglu +115:35
edmondswI assume we will pull that patch to get OOT working while we wait for them to merge15:36
esbergluedmondsw: Putting up a patch in powervm-ci15:37
esbergluedmondsw: 598115:43
edmondswesberglu +215:54
*** tjakobs has joined #openstack-powervm15:56
esbergluefried: edmondsw: Quick, get your rechecks in while you can :)15:58
edmondswesberglu :)16:07
edmondswefried do you know anything about the local cinder volume support we have in nova-powervm OOT? is that for the cinder LVM driver?16:34
*** chas has quit IRC16:34
efriededmondsw Sorry, no idea.  I just know whatever I can read from the code.  Let me know if you want to go over it.16:35
*** chas has joined #openstack-powervm16:35
*** chas has quit IRC16:36
*** chas has joined #openstack-powervm16:37
tjakobsedmondsw I can't speak for the volume/local.py, but I was using the cinder LVM driver for my initial iSCSI tests16:38
edmondswtjakobs ok, tx... I don't see how that would work other than through volume/local.py, so I think that answers my question... at least that LVM is one type of local cinder volume support16:39
edmondswtjakobs wait, what would that have to do with iSCSI?16:41
tjakobsit creates a logical volume on the c-vol node, and then exports through iSCSI for the compute node to connect to16:46
edmondswtjakobs oh... so I misread your previous comment. That wouldn't go through volume/local.py, that would go through volume/iscsi.py16:48
tjakobsYeah, I'm not sure how to use local.py, I was more commenting on how the LVM driver works (from my experience)16:49
edmondswto use local.py you'd have to run c-vol on the compute node16:50
tjakobsbut what volume_driver would you use? My c-vol and n-cpu are running on the same system, but I've never run into the local.py file16:51
edmondswtjakobs if I understand correctly, you would setup c-vol to use the LVM driver and then you could just attach to the VM through local.py instead of needing to setup iSCSI16:57
*** esberglu has quit IRC16:58
edmondswtjakobs are you saying the LVM driver requires iSCSI?16:58
tjakobsedmondsw from what i've done with it, yes16:58
*** esberglu has joined #openstack-powervm16:58
edmondswhmm... so I don't know what cinder driver you'd use, then16:59
edmondswtjakobs did you maybe configure it in some way that made it export as iscsi?16:59
tjakobsI did add a `iscsi_helper = lioadm` in cinder.conf, but i'm pretty sure it defaults to tgtadm17:01
edmondswtjakobs sounds like you don't have to use icsci with the lvm driver: http://cloudgeekz.com/71/how-to-setup-openstack-to-use-local-disks-for-instances.html17:02
*** chas has quit IRC17:04
edmondswtjakobs I also found this: https://ask.openstack.org/en/question/92001/cinder-lvm-volume-local-to-instance/17:04
tjakobsedmondsw look at https://github.com/openstack/cinder/blob/master/cinder/volume/drivers/lvm.py#L103-L10817:05
tjakobshmmm i'll look at those links17:05
edmondswtjakobs maybe the target_driver only comes into play if it's not local?17:06
edmondswI mean, if the volume isn't on the same system as the VM17:06
*** chhavi has quit IRC17:07
tjakobsI think you'd have to specify something if thats the case, b/c I have cinder makings volumes on my system, then the same system connects to those volumes over iSCSI17:07
*** efried has quit IRC17:24
edmondswtjakobs yeah, looks like LVM is tied to iscsi... but there are two drivers with driver_volume_type=local: BlockDeviceDriver and (in some cases) DrbdManageDrbdDriver17:38
*** tjakobs has quit IRC17:52
*** tjakobs has joined #openstack-powervm18:00
*** esberglu has quit IRC18:20
*** esberglu has joined #openstack-powervm18:21
*** esberglu has quit IRC18:21
*** esberglu has joined #openstack-powervm18:22
*** chas has joined #openstack-powervm19:04
*** chas has quit IRC19:09
edmondswfyi, there's a ML thread saying they're going to go back to zuul v2 again for a bit19:11
*** efried has joined #openstack-powervm19:43
efriedesberglu Not sure if you already celebrated, but your spec merged.20:08
efriedand bp approved20:09
esbergluefried: Yep I saw that. Woohoo!20:09
efriedDo you want the bp assigned to you?20:10
efriedI did that.20:10
esbergluefried: cool20:18
edmondswgood stuff20:51
*** chas has joined #openstack-powervm21:06
*** chas has quit IRC21:10
openstackgerritMerged openstack/networking-powervm master: Updated from global requirements  https://review.openstack.org/50731621:17
openstackgerritMerged openstack/nova-powervm master: change kernel backstore to fileio  https://review.openstack.org/50754821:18
tjakobsedmondsw: is there a reason some operations have a cli and some don't (regarding you comment on 498864 in support-matrix.ini)21:30
edmondswtjakobs I think cli is just indicating what corresponding CLI performs that action21:31
edmondswis it not set for some of the other things in nova-powervm even when it is set in nova?21:31
tjakobscorrect, there are several21:32
tjakobsattach/detach for a quick example21:32
edmondswefried I assume this is just a miss, not intentional... you agree?21:33
efriededmondsw tjakobs what are we talking about?21:33
efriedOkay, I see21:34
efriedIt's possible the 'cli' field was added later, and we never filled it in?21:34
efriedBut no, there's no intentional omission there.21:34
edmondswk... so I'll throw a TODO on the etherpad about filling out the others21:35
efriedWhen we're integrated with the main support matrix, the CLI goes with the whole block anyway.21:35
efriedCouldn't hurt to have it in ours, I suppose21:35
efriedThough I guess maybe someone should validate that it works for us.21:35
edmondswefried why wouldn't the CLI commands work for us?21:36
efriedWho knows.  I would think that would represent a bug in any situation where it happens.  So we can declare support, and let users open bugs.21:37
edmondswyep. The only way the CLI wouldn't work is if the API doesn't work, and that's kinda the point of the support matrix is to indicate what works21:41
openstackgerritTaylor Jakobson proposed openstack/nova-powervm master: Add resize attached volume support  https://review.openstack.org/49886421:43
*** esberglu has quit IRC21:53
openstackgerritTaylor Jakobson proposed openstack/nova-powervm master: Add resize attached volume support  https://review.openstack.org/49886421:57
*** edmondsw has quit IRC21:57
*** miltonm has quit IRC21:58
*** miltonm has joined #openstack-powervm22:00
*** tjakobs has quit IRC22:27
*** chas has joined #openstack-powervm23:07
*** chas has quit IRC23:11

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!