Tuesday, 2018-06-19

*** apearson has joined #openstack-powervm01:42
*** apearson has quit IRC01:59
*** AlexeyAbashkin has joined #openstack-powervm05:01
*** AlexeyAbashkin has quit IRC05:34
*** AlexeyAbashkin has joined #openstack-powervm05:39
*** tjakobs has joined #openstack-powervm05:44
*** tjakobs has quit IRC05:55
*** AlexeyAbashkin has quit IRC06:07
*** AlexeyAbashkin has joined #openstack-powervm08:25
*** edmondsw has joined #openstack-powervm11:50
efriededmondsw: placement handling extra specs.  Yes, it does stuff with them, indirectly.  This is in context of https://review.openstack.org/#/c/571111/3/specs/rocky/approved/add-emulated-virtual-tpm.rst -- looking...11:54
efriedwtf is tpm?11:54
efriedthis one, I guess?  https://en.wikipedia.org/wiki/Trusted_Platform_Module11:54
* efried reads spec11:55
edmondswI'm sure you'll have some comments in there11:56
edmondswone of my comments (still in draft) is that they need to talk to you placement guys11:56
edmondswbut I wanted to understand better myself how/whether extraspecs are considered in placement11:57
edmondswsince they are proposing an extraspec to indicate whether a VM should get a virtual TPM or not11:57
edmondswand ideally that would factor into placement so you could only schedule to a host that supports that11:57
edmondswefried I just posted my comments12:01
efriededmondsw: The answer is that placement only considers the extra specs it considers.  And indirectly, of course - the scheduler parses the extra specs to construct the querystring we use to ask placement for allocation candidates.12:01
edmondswok, I thought it would be something like that12:02
edmondswi.e. some extra specs are special12:02
efriededmondsw: Something like hw:vtpm could be translated through to placement-isms, but might make more sense for it to be treated as an actual resource.12:02
efriedHowever, stephenfin's comment in PS1 was a leading one: if it's not a "consumable" resource, anything we do to inventory it would be a hack.12:02
edmondswefried I could envision some drivers limiting how many vTPMs you could have on a host, and others not limiting12:03
efriededmondsw: What does "unlimited" look like?  total=$maxint ?12:03
edmondswefried the current solution that these wind river guys have uses a backing file on the hypervisor12:04
edmondswbut the logical next step is to actually use the physical TPM, and that of course is a limited resource12:04
efrieda "physical TPM" - is that like a PCI card?  A chip on the board?12:04
edmondswas long as it's just a backing file, though, you're only limited by the max number of files of the hypervisor and space on its disk12:05
edmondswchip on the board12:05
edmondswusually, anyway12:05
*** apearson has joined #openstack-powervm12:05
edmondswand in hindsight, I probably should have pinged you in the placement channel12:06
edmondswso everyone else could see this12:06
*** apearson has quit IRC12:22
*** apearson has joined #openstack-powervm12:59
*** esberglu has joined #openstack-powervm13:50
edmondsw#startmeeting PowerVM Driver Meeting14:00
openstackMeeting started Tue Jun 19 14:00:48 2018 UTC and is due to finish in 60 minutes.  The chair is edmondsw. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
*** openstack changes topic to " (Meeting topic: PowerVM Driver Meeting)"14:00
openstackThe meeting name has been set to 'powervm_driver_meeting'14:00
edmondsw#link https://etherpad.openstack.org/p/powervm_driver_meeting_agenda14:01
edmondswping esberglu gman-tx mdrabe mujahidali chhagarw14:01
edmondsw#topic In-Tree Driver14:01
*** openstack changes topic to "In-Tree Driver (Meeting topic: PowerVM Driver Meeting)"14:01
edmondsw#link https://etherpad.openstack.org/p/powervm-in-tree-todos14:01
edmondswI don't think we have any in-tree updates this week that I can think of... anyone else?14:02
efriednot I14:02
*** mujahidali has joined #openstack-powervm14:02
edmondsw#topic Out-of-Tree Driver14:02
*** openstack changes topic to "Out-of-Tree Driver (Meeting topic: PowerVM Driver Meeting)"14:02
edmondsw#link https://etherpad.openstack.org/p/powervm-oot-todos14:03
edmondswmdrabe I saw you were starting to look at MSP stuff14:03
edmondsware you working on nova-powervm changes for that?14:03
mdrabeYea gonna whip that up in nova-powervm soon14:03
edmondswefried sounds like you're off the hook :)14:03
mdrabeI'm assuming we'll want test for that in devstack14:04
mdrabeI'll probably need help with that14:04
edmondswmdrabe anything we put into nova-powervm should be tested with devstack14:04
edmondswmdrabe I'm going to be writing up a wiki to help folks14:04
edmondswmdrabe if you're working on it before that email goes out, ping me14:05
mdrabeYep thnx14:05
efriedIn reviewing py3-first changes I noticed our tox.inis - at least for networking-powervm, probably the other -powervms as well - is totally borqué.  I started looking into it a little bit, but mostly don't really know what I'm doing.  We should probably move to stestr before we try too hard to fix existing unused envs in there.14:05
edmondswefried I know essentially nothing about stestr... seem like it will be hard to transition?14:06
efriedProbably not.  There's an email thread somewhere or other from mtreinish, which is where I would start.  And if help needed, he's the one I would ask.14:06
edmondswi.e., you're not taking that further at the moment14:07
efriedpriorities :)14:07
edmondswI added it to the TODO list14:07
edmondswesberglu, I think you said you're done porting changes from IT reviews back to OOT, correct?14:08
edmondswI removed that workitem from the TODO etherpad14:08
esbergluedmondsw: No I'm done with IT follow ups14:08
esbergluI haven't ported yet, I have a couple local changes that aren't quite done14:09
edmondswoh, misunderstood... I'll put it back then14:09
esbergluFor snapshot and base disk adapter, haven't looked at localdisk or vSCSI backports yet other than stuff I noted while developing14:09
edmondswso are you planning to finish that out, or are you done?14:10
edmondswif you have a list of things that need to be done there, you could throw it in the etherpad under that item14:10
esbergluedmondsw: I'm planning on finishing that. It's all pretty minor changes so I've just been poking when I have a chance14:10
edmondsw+1 and thank you14:10
edmondswchhagarw doesn't seem to be online today, but real quick on iscsi...14:11
edmondswhttps://review.openstack.org/#/c/567575/ merged14:12
edmondswand she is now working on https://review.openstack.org/#/c/576034/14:12
edmondswI'm not sure the approach she started with is the best, and I've commented accordingly14:12
edmondswwould love others thoughts there14:13
edmondswefried mdrabe etc.14:13
efriedoh, I was waiting until you were +2 to even look.14:14
edmondswhmm... I can't get to http://review.openstack.org/ now...14:15
efriedjust came back14:15
edmondswok yep14:15
edmondswI'm trying to setup an env to test https://review.openstack.org/#/c/567964/14:15
edmondswtjakobs has been working on a couple rbd changes that I think we can merge pretty easily once NovaLink and pypowervm support is in14:17
edmondswand the change to use loop backstore type is in a similar boat14:17
edmondswI've also been working on docs a bit when I can spare a minute (so not often)14:18
edmondswI think that's all for OOT14:18
edmondsw#topic Device Passthrough14:18
*** openstack changes topic to "Device Passthrough (Meeting topic: PowerVM Driver Meeting)"14:18
edmondswefried ^14:18
efriedSpec for reshape-provider-tree merged.  nrp-in-alloc-cands series hasn't moved much in the last week.  Consumer gens getting close, but not there yet.14:19
edmondsw"consumer gens"?14:19
efriedconsumer generations14:19
edmondswoh, right14:20
efriedFinally been reviewing cyborg specs, had some back-and-forth with Sundar about os-acc, getting closer on that.14:20
edmondswwhat's os-acc?14:20
edmondswthis like os-vif but for accelerators?14:20
efriedlike os-vif but for accelerators14:20
edmondswtx for commenting on the vTPM spec14:21
edmondsw#topic PowerVM CI14:22
*** openstack changes topic to "PowerVM CI (Meeting topic: PowerVM Driver Meeting)"14:22
efriedNot sure if I mentioned last week, but the libvirt update_provider_tree and shared DISK_GB stuff merged, which is a big milestone.14:22
edmondsw#link https://etherpad.openstack.org/p/powervm_ci_todos14:22
edmondswefried sorry, thought you were done14:22
efriednow I am14:22
edmondswmujahidali where do we stand with the CI?14:23
edmondswesberglu said yesterday that it was broken and he was working with you to redeploy14:23
mujahidalicurrnetly I am redeplying the compute nodes14:23
edmondswesberglu I'm not even seeing powervm in http://ci-watch.tintri.com/project?project=nova anymore14:24
edmondswdoes it get removed if we don't report on anything for a while?14:24
esbergluedmondsw: Yeah it only goes back 24 hours, so if we haven't voted in 24 it disappears14:25
edmondswmujahidali think we'll be back up today?14:25
esberglumujahidali was having some issues running the Ansible scripts. Are you still stuck on that same issue?14:25
mujahidaliedmondsw:most probably14:25
mujahidaliesberglu: problem was resolved14:26
esbergluWhat was the issue?14:26
mujahidaliI restarted neo-714:27
mujahidaliit was pingable but not able to ssh14:27
mujahidaliand most of the vms are in error state14:27
esbergluI thought you were having another issue with the playbook failing?14:28
esbergluWas that error just because neo7 was down?14:28
esbergluYou sent me a log where the Update resolv.conf file task failed14:28
mujahidaliyeah, that was due to virtual env, asnible was asking for "python netaddr" and it was already installed14:29
mujahidaliso I googled it and did a pip install for the same14:29
esberglumujahidali: Okay cool let me know if you hit any other issues while deploying14:29
edmondswgoogle for the win!14:30
mujahidaliesberglu: thanks for the quick help :)14:30
esbergluI've given mujahidali some notes for where we are with multinode CI and the next steps to take there14:30
esbergluI'm assuming we want to give that priority over the stable branch vSCSI CI he was doing?14:30
edmondswesberglu I thought we were really close to done on the vscsi stuff14:31
esbergluedmondsw: Yeah I think we had all branches except ocata working14:31
edmondswworking for queens and pike now, just ocata left?14:31
mujahidaliesberglu: I am also looking into multinode, will come back to you after trying the steps mentioned in your mail.14:32
edmondswif we're going to put the ocata work on hold to focus on multinode, which makes sense to me, let's remove it from the changeset and update the commit to be specific to pike/queens14:32
esbergluedmondsw: Honestly might not be worth the trouble of porting to ocata for now14:32
edmondswso we can go ahead and merge that for those release and work on ocata separately14:32
esbergluYeah what you said14:32
edmondswesberglu right14:32
edmondswyou got that, mujahidali?14:33
mujahidalimake sense14:33
edmondswshould be a quick/easy change, and then you can focus on multinode (after you get the CI back up, of course)14:33
edmondswanything else for CI/14:34
esbergluThe only other CI task is getting those additional systems added to the CI pool14:34
mujahidaliyeah, will look into that14:35
edmondswmujahidali FYI, I hit an issue trying to do a greenfield NovaLink install over the weekend14:35
edmondswyou will hit the same issue until we update the bootp server14:35
edmondswI will try to do that today14:35
edmondswping me if I forget to update you when that's fixed14:36
edmondswone other thing I remembered14:36
edmondswwe had a slack conversation last week while you were out mujahidali14:36
edmondswabout an issue that came up where something updated a file and broke the CI14:36
edmondswgo back and read that thread when you get a chance14:37
edmondswnext time we redeploy the CI for non-emergency (i.e. not what you're doing now) we should talk about updating some things there to prevent that issue in future14:37
mujahidalisure, will look into that, I think it's the opnestack-ci channel14:38
edmondsw#topic Open Discussion14:39
*** openstack changes topic to "Open Discussion (Meeting topic: PowerVM Driver Meeting)"14:39
edmondswanything else before we close?14:39
edmondswalright, tx everyone14:41
*** openstack changes topic to "This channel is for PowerVM-related development and discussion. For general OpenStack support, please use #openstack."14:41
openstackMeeting ended Tue Jun 19 14:41:36 2018 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:41
openstackMinutes:        http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-06-19-14.00.html14:41
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-06-19-14.00.txt14:41
openstackLog:            http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-06-19-14.00.log.html14:41
*** AlexeyAbashkin has quit IRC16:52
*** mujahidali has quit IRC17:43
*** mujahidali has joined #openstack-powervm17:43
*** Mujahid_ has joined #openstack-powervm17:48
*** mujahidali has quit IRC17:51
*** Mujahid__ has joined #openstack-powervm17:52
*** Mujahid_ has quit IRC17:56
*** Mujahid_ has joined #openstack-powervm18:12
*** Mujahid__ has quit IRC18:15
*** Mujahid_ has quit IRC19:07
*** AlexeyAbashkin has joined #openstack-powervm19:26
*** AlexeyAbashkin has quit IRC19:37
openstackgerritEric Berglund proposed openstack/nova-powervm master: WIP: Snapshot In-tree Backports  https://review.openstack.org/57663019:45
openstackgerritEric Berglund proposed openstack/nova-powervm master: WIP: Snapshot In-tree Backports  https://review.openstack.org/57663020:03
*** edmondsw has quit IRC20:47
*** apearson has quit IRC21:32
*** esberglu has quit IRC21:36
*** esberglu has joined #openstack-powervm21:37
*** esberglu has quit IRC21:41

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