Tuesday, 2018-10-30

*** tonyb has quit IRC06:50
*** mujahidali has joined #openstack-powervm13:49
edmondsw#startmeeting PowerVM Driver Meeting14:01
openstackMeeting started Tue Oct 30 14:01:03 2018 UTC and is due to finish in 60 minutes.  The chair is edmondsw. Information about MeetBot at http://wiki.debian.org/MeetBot.14:01
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:01
*** openstack changes topic to " (Meeting topic: PowerVM Driver Meeting)"14:01
openstackThe meeting name has been set to 'powervm_driver_meeting'14:01
edmondsw#link agenda: https://etherpad.openstack.org/p/powervm_driver_meeting_agenda14:01
edmondsw#topic In-Tree Driver14:01
*** openstack changes topic to "In-Tree Driver (Meeting topic: PowerVM Driver Meeting)"14:01
edmondswI'm waiting for a 2nd +2/+A on https://review.openstack.org/#/c/610174/14:01
edmondswI think that's the only thing we've currently got going in-tree14:02
efriedIt was +A but then got bumped for the uuidsentinel change.14:02
efriedI might try to convince Jay to fast-approve it since the rebase was trivial.14:02
efriedBut this patch deserves a bit of a closer look since...14:03
efried...where it is now officially the responsibility of the virt driver to set allocation ratios and reserved values.14:03
efriedBecause https://review.openstack.org/#/c/613991/ is ripping out the bits where the resource tracker sets those outside of the virt driver, for the update_provider_tree path.14:03
edmondswi.e. we have to merge 468560 before those others can merge?14:04
efriednope. Turns out I "predicted" the future here, and the powervm patch is already copacetic.14:04
efriedOnce that second one merges, we'll get to do an optional optimization to call the base ComputeDriver method to calculate the disk reserve14:05
efriedthis: https://review.openstack.org/#/c/613126/4/nova/virt/driver.py@86014:05
efriedright now that's inlined: https://review.openstack.org/#/c/468560/4/nova/virt/powervm/driver.py@20314:06
efriedIn other in-tree news, don't forget to vote for the T release name.14:06
edmondswnot sure I'm following, but I'll take a look after the meeting and it'll probably all come together14:06
edmondswI don't think I got an email about the T name vote yet...14:06
efriedit's a public poll. So you can vote once per IP address. (but obviously just vote once)14:07
edmondswah, that explains it14:07
efriedAre you on the Train train?14:07
edmondswyeah... though I also liked Thomas (the tank engine)14:08
edmondswand would have been cool to have suggested the name that got picked14:08
efriedOh, was that your suggestion?14:08
edmondswbut I'll vote for Train14:08
efriedThomas is on there14:08
efriedand it's ordered, and you can have ties14:08
edmondsw#topic Out-of-Tree Driver14:10
*** openstack changes topic to "Out-of-Tree Driver (Meeting topic: PowerVM Driver Meeting)"14:10
edmondswsimilar to IT, we have https://review.openstack.org/#/c/613342/ needing a 2nd +2/+A14:10
edmondswmdrabe ^14:10
edmondswmdrabe I assume that's an easy test with PowerVC (for the bug you requested earlier) and then you can +214:11
edmondsw(looking at the code as well, of course)14:12
mdrabethnx will look14:12
mdrabeedmondsw: Did we wanna backport that?14:12
edmondswI don't think I'll push backporting... I view it more as hardening than as a vulnerability14:12
efried(Update: Success convincing Jay to fast-approve https://review.openstack.org/#/c/468560/)14:13
mdrabeSounds good14:13
edmondsw#topic Device Passthrough14:13
*** openstack changes topic to "Device Passthrough (Meeting topic: PowerVM Driver Meeting)"14:13
efriedOh, there was one more OOT thing...14:13
edmondsw#topic OOT part 214:14
*** openstack changes topic to "OOT part 2 (Meeting topic: PowerVM Driver Meeting)"14:14
efried(edmondsw: there's also #undo)14:14
edmondswah, nice14:14
efriedGot an email to our pypowervm github account, assume y'all got it too.14:14
efriedAbout requests <=2.19 having a security vulnerability.14:15
edmondswyeah, I got that14:15
efriedI checked the nova project, since we usually try to align with that.14:15
edmondswwe should not be affected because pypowervm is only used with localhost14:15
edmondswremoving the MITM concern14:15
efriedAnd it's at 2.14.214:15
efriedokay, I didn't look into the vulnerability itself.14:16
efriedJust whether it was feasible for us to bump.14:16
efriedWe're behind openstack anyway there.14:16
edmondswglad you brought it up14:16
efriedrequirements project has upper constraint at 2.2014:16
edmondswI don't know that we need to keep up with other OpenStack projects... kind of the point of having l-c in each project now instead of global like it used to be14:17
edmondswu-c will always be the latest available that we haven't found an issue with14:17
edmondswbut l-c should be the oldest available that we don't have an issue with14:17
efriedyeah, sure14:17
efriedwell, we don't have l-c in pypowervm14:17
efriedand we at least need to make sure we don't have conflicts.14:18
edmondswl-c for pypowervm is essential the lower bound listed in requirements.txt14:18
efriedbut in this case it would let us clean up that req a bit. Right now it's a three-parter.14:18
edmondswefried so what exactly are you suggesting?14:19
efriedWhat I don't see is any patches bumping l-c in nova at large. Wonder if it just hasn't percolated down there yet.14:19
efriedsorry, s/nova/openstack/14:19
edmondswefried I'm guessing that it's still percolating14:19
efriedI'm not really suggesting anything yet. I kinda want to find out if anyone in openstack is concerned.14:19
efriedif so, we should bump our req to 2.20.0 just to align.14:20
efriedif not... I guess if you think the vulnerability is n/a for us, we could leave it.14:20
edmondswthere I disagree... we don't need to align, any more than openstack projects need to align with each other14:20
edmondswthis is why they moved l-c into individual projects, so that they don't have to align14:20
efriedSorry, I didn't mean align for the sake of alignment.14:21
edmondswyou meant specifically for this bug?14:21
efriedBut if there's a known vulnerability and it's concerning enough that openstack bumps minimum to 2.20.0, then it's probably a good idea if we do too, even if it's not strictly necessary (at the moment).14:21
efriedBut don't we have people consuming pypowervm in a remote fashion?14:21
edmondswI didn't think it supported that, without hacking14:22
edmondswmaybe I need to take a closer look at what local2remote does14:22
edmondswI thought vanilla pypowervm could only be used locally14:22
efriedeven so14:23
efriedfrom the other side, any harm in bumping the minimum?14:23
edmondswok, the more I think of it, and just confirmed... local2remote isn't changing pypowervm code to make remote possible14:24
edmondswpypowervm does support remote14:24
edmondswso... yeah... we might want to bump14:24
edmondswwe could leave it the way it is, and say that you're welcome to update requests but we just won't make you do it14:24
edmondswbut probably better to bump14:25
edmondswlet me make sure there is a python-requests 2.20.0 on both RHEL and Ubuntu 16.04 and 18.04 first14:26
edmondsw#topic Device Passthrough14:27
*** openstack changes topic to "Device Passthrough (Meeting topic: PowerVM Driver Meeting)"14:27
edmondswefried ^14:27
efriedokay, the new file format spec got a couple reviews since last week. I was kinda waiting for more, but at this point I guess I'll incorporate those changes and push another rev soon.14:28
efriedotherwise, no progress.14:29
efriedalso will need to respin the actual dev passthrough spec to account for having factored out --^14:30
efriedbut haven't started that yet either.14:30
efriedEOM unless questions14:30
edmondswno questions. Haven't gotten to the spec yet14:30
edmondsw#topic PowerVM CI14:31
*** openstack changes topic to "PowerVM CI (Meeting topic: PowerVM Driver Meeting)"14:31
edmondswmujahidali ^14:31
mujahidaliAfter redeploying the neos CI looks quite good that earlier but there is some problem in accesing the http://ci-watch.tintri.com/14:32
edmondswas long as the CI is working :)14:34
edmondswany update on other things?14:35
mujahidaliWe are able to stack IT for py-3devstack(tempest was failing)14:35
edmondswyou figured out the issues? I haven't seen any commit proposals for fixes14:35
mujahidalithe issue was in OOT14:36
edmondswI thought you were also looking at a pypowervm issue14:36
mujahidaliand I didn't get chance to test pypowervm changes.14:36
edmondswmujahidali are you going to propose a commit for an OOT fix?14:37
mujahidaliI need to test it first.14:37
mujahidaliand I will be pulled for some time as a backup for Bharath until he is back(for PVC-build).14:38
edmondswyep, understood14:39
edmondswanything else?14:39
mujahidalithat's it.14:39
edmondsw#topic Open Discussion14:40
*** openstack changes topic to "Open Discussion (Meeting topic: PowerVM Driver Meeting)"14:40
edmondswI'm only seeing python-requests 2.9.1 on Ubuntu 16.04.... :(14:41
efriedthat wouldn't even satisfy our minimum, would it?14:41
edmondswI'll keep looking into that14:42
mujahidaliany idea why it's http://ci-watch.tintri.com/ not working ?14:43
edmondswmujahidali no clue14:43
edmondswefried looks like Ubuntu didn't bump 2.9.1 to 2.20.0... they just patched 2.9.1 with the fix for this issue14:44
edmondswcome to think of it, that kind of thing is common, and explains why you're not seeing other projects bumping versions14:44
efriedmm, interesting.14:44
efriedI'll see if anyone in -infra knows about ci-watch14:45
edmondswalright, thanks everyone14:47
*** openstack changes topic to "This channel is for PowerVM-related development and discussion. For general OpenStack support, please use #openstack."14:47
openstackMeeting ended Tue Oct 30 14:47:06 2018 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)14:47
openstackMinutes:        http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-10-30-14.01.html14:47
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-10-30-14.01.txt14:47
openstackLog:            http://eavesdrop.openstack.org/meetings/powervm_driver_meeting/2018/powervm_driver_meeting.2018-10-30-14.01.log.html14:47
*** mujahidali has quit IRC14:54
edmondswefried I looked over the stuff you brought up above when we were talking about IT... all LGTM15:28
efriedroger that15:54
efriededmondsw: Mr. The Tuba Guy approved the tempfile patch.16:10
edmondswgreat, tx16:10
efriededmondsw: and mdrabe +2'd the oot one. Do we +W our own around here, or should we ask him to do it?16:11
edmondswI'm fine doing it myself, but it does look nicer if you do it :)16:12
*** chhagarw has joined #openstack-powervm16:39
efriededmondsw: Done.16:51
efriedAnd in other news, mmedvede sent an email to the maintainers of tintri asking if they're continuing to support it.16:51
efriedI know the nova team uses it, so there should be more motivation than just us.16:51
*** chhagarw has quit IRC17:29
efriededmondsw: The oot patch is broke. References to those consts. You got a fup patch coming quick?21:29
efriededmondsw: We might could be a little clever about this. We can make the CreateAndConnectCfgDrive job output the cfg drive name, then pull it in from the PowerOn job. Instead of duplicating the name gen logic, just pass the conf option to the job (or reference the conf option directly from within the job, but that's kind of dirty).21:33
efriedThe tricky part is figuring out how to make that input optional, so we can ignore if CreateAndConnectCfgDrive didn't run. My taskflow is a bit rusty, not sure offhand how to do that.21:34
efriedbut for now you could just put the consts back.21:35
openstackgerritEric Fried proposed openstack/nova-powervm master: Use tempfile for powervm config drive  https://review.openstack.org/61334221:36
openstackgerritEric Fried proposed openstack/nova-powervm master: Use tempfile for powervm config drive  https://review.openstack.org/61334221:37
efriededmondsw, mdrabe: Rebased to yank it out of the gate ^21:38
efriedFortunately the IT patch didn't have that problem. (I guess we must have a TODO to port that delete-cfg-drive functionality in-tree?)21:40
mdrabeYea probably21:45
*** tonyb has joined #openstack-powervm23:03
*** chhagarw has joined #openstack-powervm23:25
*** chhagarw has quit IRC23:30

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