Tuesday, 2019-02-26

cdentstill not totally 100%, but haven't felt this "hey, it's fun to fix code" in a long time00:00
*** s10 has joined #openstack-placement00:01
cdentefried, mriedem: i'm having a discussion with management tomorrow in which I need to convince them that me working on getting shared disk working in nova is good for everyone. I've already pointed out how it's going to make vmware datastores better via https://github.com/cdent/placement-disk-models . Either of you have anything else to add?00:03
cdentOther than "well, somebody has to do it"?00:03
efriedcdent: I no longer have the push from PowerVM to make SSP pretty, but maybe edmondsw does.00:05
cdentthat might actually be a reasonable argument "well eric was getting pushed on that, but not anymore"00:05
efriedafter misinterpreting that ^ a couple of times, I finally get what you mean.00:12
efriedBut I would still be interested in helping out on it.00:12
efriedI have lots of background and some pieces I want to see done a certain way (so they don't exclude "boutique" drivers like PowerVM).00:12
cdentSure, I'm thinking of it solely as a way of expressing need00:12
efriedokay, I'm getting in trouble for having my hands on a keyboard instead of a wood chisel. See y'all tomorrow.00:13
efriedo/00:13
cdentso many companies like to pretend like some magical open source bunny will do the work00:13
cdentsee ya00:13
*** s10 has quit IRC00:18
mriedemi've got nothing00:27
edleafecdent: Put that on your business card: "Magical Open Source Bunny"00:27
cdentI have business cards?00:27
melwittpeople have needed shared storage working in nova for years. but it not working is so latent (as long as nova has existed) and other things keep getting prioritized higher than it00:29
melwittI don't think that helps you really, but that's the situation00:30
cdentmelwitt: yeah. I'm not clear on how best to phrase that "the has been so important for so long, but not quite important enough to burst the damn"00:30
cdents/the has/this has/00:30
melwittI think it's because it's the accounting that's wrong and it can be worked around, for the most part00:31
melwittit's like, this sucks but we want other stuff more, that can't be worked around00:31
cdentthe advent of placement has dorked some of the ways that vmware's driver was doing workarounds00:31
cdentso I'm hoping that will be at least somewhat compelling, but who knows00:32
melwittyeah. I mean, it's particularly interesting because for example, my company, we ship with shared storage, ceph. so having shared disk accounting work properly is important,00:33
melwittbut each cycle there's always something else. like, we need numa live migration more. or we need sriov live migration more. and it just keeps happening00:33
cdentI remember eoghan getting cranky with me about 6 months into placement's life: what have you guys go to show for youselves? Is shared disk working?00:33
cdent(not properly cranky, just the usual "sigh, openstack" cranky)00:34
melwittyeah, it's tough. fundamentally changing how we do resource handling is a ton of work and at first not going to see those kind of benefits in the early stages00:37
cdenteach day, in every way...00:38
cdentgabbi-based integration tests are passing now, so I'm gonna bed00:39
cdentgoodnight all00:39
*** cdent has quit IRC00:39
openstackgerritMerged openstack/placement master: Update the doc in _get_provider_ids_matching()  https://review.openstack.org/63861100:45
openstackgerritMerged openstack/placement master: Don't use OVO in Trait and TraitList objects  https://review.openstack.org/63680501:04
openstackgerritMerged openstack/placement master: Don't use OVO for Inventory and InventoryList  https://review.openstack.org/63680601:04
*** tetsuro has joined #openstack-placement01:35
*** tetsuro has quit IRC01:38
*** tetsuro has joined #openstack-placement01:39
*** mriedem is now known as mriedem_afk01:39
*** tetsuro has quit IRC01:53
*** tetsuro has joined #openstack-placement01:57
*** tetsuro has quit IRC02:00
*** tetsuro has joined #openstack-placement02:03
*** tetsuro has quit IRC02:05
openstackgerritMerged openstack/placement master: Prepare in_tree allocation candidates  https://review.openstack.org/63572202:05
openstackgerritMerged openstack/placement master: in_tree[N] alloc_cands with microversion 1.31  https://review.openstack.org/63572302:06
*** tetsuro has joined #openstack-placement02:10
*** mriedem_afk has quit IRC02:24
*** mriedem has joined #openstack-placement02:29
*** mriedem has quit IRC03:12
*** tetsuro has quit IRC03:34
openstackgerritMerged openstack/placement master: Don't use OVO with ResourceProvider and ResourceProviderList  https://review.openstack.org/63680703:52
*** takashin has left #openstack-placement05:31
*** tetsuro has joined #openstack-placement06:14
melwitthas anyone seen an error like this before from placement? 500 Circular reference detected coming from the /usages API http://logs.openstack.org/73/638073/7/check/grenade-py3/5a86c65/logs/screen-placement-api.txt.gz#_Feb_26_02_38_18_42644706:22
melwitt(my patch adds calls to the /usages API ^)06:23
*** tetsuro has quit IRC06:31
*** tetsuro has joined #openstack-placement07:06
*** helenafm has joined #openstack-placement08:20
*** tssurya has joined #openstack-placement08:30
*** ttsiouts has joined #openstack-placement08:49
*** tetsuro has quit IRC08:57
*** ttsiouts has quit IRC09:07
*** ttsiouts has joined #openstack-placement09:08
*** ttsiouts has quit IRC09:13
*** ttsiouts has joined #openstack-placement09:14
*** dklyle has quit IRC09:25
*** e0ne has joined #openstack-placement09:25
*** dklyle has joined #openstack-placement09:27
*** tssurya has quit IRC09:30
*** tssurya has joined #openstack-placement09:31
*** cdent has joined #openstack-placement09:43
*** IvensZambrano has joined #openstack-placement09:53
*** ttsiouts has quit IRC11:07
*** ttsiouts has joined #openstack-placement11:08
*** ttsiouts has quit IRC11:12
*** IvensZambrano has quit IRC11:40
*** IvensZambrano has joined #openstack-placement11:41
*** IvensZambrano has quit IRC12:00
*** IvensZambrano has joined #openstack-placement12:01
*** ttsiouts has joined #openstack-placement12:08
*** snevi has joined #openstack-placement12:26
*** IvensZambrano has quit IRC12:30
*** cdent has quit IRC12:30
*** zzzeek has quit IRC12:30
*** openstackgerrit has quit IRC12:30
*** rubasov has quit IRC12:30
*** zzzeek has joined #openstack-placement12:31
edmondswcdent sorry, I'm going to pretty much echo what others have said... shared disk is important, it's been important a long time, but it's never been quite important enough to prioritize over other things and make it happen.12:45
edmondswand now I'm not really working on OpenStack anymore :( just keeping an eye on things and trying to help my old team as I can, so I'm not going to be a driving force12:47
*** mriedem has joined #openstack-placement12:50
*** ttsiouts has quit IRC12:51
*** alex_xu has quit IRC12:51
*** ttsiouts has joined #openstack-placement12:51
*** alex_xu has joined #openstack-placement12:52
*** ttsiouts has quit IRC12:56
*** ttsiouts has joined #openstack-placement13:03
*** cdent has joined #openstack-placement13:30
*** diga has joined #openstack-placement13:35
cdentgibi: if you're around there's a minor test fix which could use your kick: https://review.openstack.org/#/c/639168/13:37
*** rubasov has joined #openstack-placement13:43
gibicdent: I've kicked13:57
cdentthank you!13:57
*** ttsiouts has quit IRC15:13
*** ttsiouts has joined #openstack-placement15:13
*** e0ne has quit IRC15:16
*** ttsiouts has quit IRC15:18
*** ttsiouts has joined #openstack-placement15:34
cdentefried: in a bizarre twist of fate, I'm momentarily at a loss for something to do (because brain can only work on certain things at certain phases of the moon). Is it cool with you if I rebase/fix/tidy your dry-ing of the post-ovo-ectomy stuff?15:38
cdentseems like a second set of eyes might be a good thing15:38
efriedcdent: I don't mind, would only be concerned for lack of reviewers if you wind up having to recuse yourself :)15:41
* cdent shrugs15:41
efriedcdent: If you are looking for something to do, there's three placement blueprints that aren't yet started.15:41
cdentwe probably need more of gibi and tetsuro in this stack at some point15:41
efriedmm15:41
cdenta) not looking for that kind of work (right now), b) ideally feature related work would be done by non-regulars (huge (and ironic/knowing/teasing-efried) emphasis on the ideally)15:42
*** e0ne has joined #openstack-placement15:53
* gibi need more brains15:57
*** tssurya has quit IRC16:09
*** belmoreira has quit IRC16:20
mriedemcdent: if you're looking for stuff to do, osc-placement is falling quite behind in placement api microversion support16:30
cdentmriedem: see a16:30
mriedemand we've also long talked about adding a meta command  to that to allow setting inventory allocation ratios on all providers in an aggregate16:31
cdentwhen I'm not in the current phase of the moon I'm going to investigate making it easier for us to use storyboard to track that kind of thing16:33
cdentsince they sound a lot  like stories16:33
cdentnon-bug issues, I guess16:34
*** diga has quit IRC16:34
*** openstackgerrit has joined #openstack-placement16:35
openstackgerritChris Dent proposed openstack/placement master: Factor listiness into an ObjectList base class  https://review.openstack.org/63732516:35
openstackgerritChris Dent proposed openstack/placement master: Move _set_objects into ObjectList  https://review.openstack.org/63732816:35
openstackgerritChris Dent proposed openstack/placement master: Move *List.__repr__ into ObjectList  https://review.openstack.org/63733216:35
openstackgerritChris Dent proposed openstack/placement master: Clean up ObjectList._set_objects signature  https://review.openstack.org/63733516:35
melwittcan anyone help me with this error I'm getting calling the placement /usages API? I'm calling it in a new patch to count quota usage from placement http://logs.openstack.org/73/638073/8/check/tempest-full-py3/161317d/controller/logs/screen-placement-api.txt.gz#_Feb_26_07_25_17_36362416:53
cdentmelwitt: that should be fixed with the merge of.... /me looks16:58
*** helenafm has quit IRC16:58
cdenthttps://review.openstack.org/#/c/639216/16:59
cdenthttps://bugs.launchpad.net/nova/+bug/181763316:59
openstackLaunchpad bug 1817633 in OpenStack Compute (nova) "Listing placement usages causes a circular reference error" [High,Triaged] - Assigned to Chris Dent (cdent)16:59
cdentit's an artifact of the ovo-ectomy16:59
*** ttsiouts has quit IRC16:59
melwittsweet, thanks. I had passing tempest last on feb 21, so I wondered if it had to be something recent17:00
*** ttsiouts has joined #openstack-placement17:00
cdentmelwitt: as an aside it managed to merge because the problem only shows up when using mysql, so it was caught by this stuff: https://review.openstack.org/#/c/601614/17:00
cdent(which is now based on the fix to the bug)17:00
melwittah ok17:02
cdent(it turns out tempest never calls /usages )17:03
*** ttsiouts has quit IRC17:05
openstackgerritChris Dent proposed openstack/placement master: WIP: Use native list for lists of Usage  https://review.openstack.org/63939117:13
cdentjaypipes, efried : next step in the ectomy, initially to make jay happy, but it seems pretty good if we're inclined17:13
cdentjaypipes, gibi, mriedem : could one of you look at https://review.openstack.org/#/c/639216/ as it is causing real problems for people (inc. melwitt )17:15
openstackgerritMerged openstack/placement master: Remove NOTEs about _RE_INV_IN_USE  https://review.openstack.org/63623617:16
melwittI updated my patch to Depends-On it, to see what happens17:16
gibicdent: done17:22
cdentthanks gibi17:22
*** e0ne has quit IRC17:24
openstackgerritMerged openstack/placement master: Fix a bad granular gabbi test  https://review.openstack.org/63916817:36
*** snevi has quit IRC18:42
openstackgerritMerged openstack/placement master: Cast Usage.usage to int  https://review.openstack.org/63921618:55
*** alex_xu has quit IRC19:20
*** mriedem has quit IRC19:21
*** mriedem has joined #openstack-placement19:22
*** e0ne has joined #openstack-placement19:26
*** e0ne has quit IRC19:46
*** e0ne has joined #openstack-placement19:46
*** e0ne has quit IRC20:31
cdentefried, jaypipes, edleafe, melwitt : as you seem likely to be awake or around, question about storyboard20:50
edleafeshoot20:51
cdentI talked with the #storyboard folk about starting to use it without migrating, since we're talking about letting old stuff get done or expire on launchpad20:51
cdentthat turns out to be okay, so I'm planning to do the necessary to have placement use storyboard in gerrit20:51
cdentthis is also safe because it's connection to launchpad is already broken20:51
cdent(by the new governance)20:52
edleafeIsn't that just a one-line change?20:52
cdentthe question is: what to do about os-resource-class and os-traits?20:52
cdentdo we want them to be in a group with placement20:52
cdentor their own thing20:52
cdent"group" is basically how you say "these several repos use the same stories"20:52
cdentoh, and osc-placement too20:52
edleafeAs long as they worked the same, it wouldn't bother me to have them separate20:53
cdentto what end? I think it might be easier together20:53
edleafeIOW, nothing like placement on SB, os* on launchpad20:53
cdentas there's a (slim) chance it might help us pay attention to orphans like osc-placement a bit better20:53
edleafeOh, I thought you were looking to separate them20:53
edleafeI'd rather they be together20:54
cdentefried, jaypipes, melwitt, mriedem : anything to add?20:54
efriedI prefer os-rc and os-traits and osc-placement be "under" the placement umbrella entirely.20:55
mriedemif you move os-* under SB then you need to migrate the existing launchpad stuff20:56
edleafemriedem: that's what made me think that cdent wanted to keep them separate20:56
mriedemlooks like os-resource-classes isn't in launchpad,20:56
mriedemand os-traits doesn't have bugs or blueprints20:57
mriedemi'm pretty sure osc-placement has bugs in LP though20:57
cdentmriedem: 'xactly. plus right now any connection between gerrit and launchpad for any of that stuff is broken20:57
mriedemhttps://launchpad.net/placement-osc-plugin20:57
cdentmy thinking was that for the duration of the cycle we hybrid20:57
mriedem^ isn't broken20:57
mriedemhttps://bugs.launchpad.net/placement-osc-plugin20:58
cdenti don't mean that it's not there20:58
cdentI mean that reporting on progress from gerrit changes20:58
mriedemwhy would that be broken for osc-placement?20:59
cdentI may be wrong, but my understanding was that when we started using a different set of acl's things broke, but I may be conflating other problems that were happening somewhere else21:00
mriedemi'll test it quick for that repo21:00
mriedemsince the bug link in the docs is wrong21:00
cdentin any case: what is the outcome that you want? what I was proposing that we only migrate stuff that we can demonstrate matters, not wholesale21:01
openstackgerritMatt Riedemann proposed openstack/osc-placement master: Fix bugs link in contributing doc  https://review.openstack.org/63943621:02
mriedembased on those 3 projects being in placement under governance https://governance.openstack.org/tc/reference/projects/placement.html21:03
mriedemi'd just move them all to storyboard if placement is going to be in storyboard21:03
mriedemas noted, osc-placement is the only one that had stuff in launchpad21:03
mriedemos-traits and os-r-c doesn't21:03
mriedemlooks like the bug tracking reporting may be busted https://bugs.launchpad.net/placement-osc-plugin/+bug/181779021:04
openstackLaunchpad bug 1817790 in placement-osc-plugin "bug link in contributor docs is wrong" [Medium,New]21:04
mriedemlp hasn't updated yet21:04
cdentyeah, that's the stuff I was talking about21:04
cdentI haven't seen it work in anything placement related in quite some time21:04
cdentbrb21:05
*** cdent has quit IRC21:05
*** cdent has joined #openstack-placement21:05
cdentsorry, my client got all horked up21:05
mriedemcdent: i'm not sure if melwitt is around today so maybe just post whatever your proposal is to that thread i started about feature tracking in the ML21:13
mriedemand we'll get lazy consensus there21:13
cdentmriedem: I think what may not be clear here is that we can go ahead and create the stuff in storyboard and it doesn't impact the current state of the bugs in launchpad. I'm going to go ahead and create the project-config change and will link it here and make sure the commit message is sufficiently verbose21:14
cdentmriedem: the reason I want to get things started is because we seem to be gathering things that need to be remembered and no good place to put them. storyboard seems like that place21:15
cdentand since it won't break stuff (since stuff is already broken), may as well21:16
mriedemok, still probably good to update that thread in the ML with that information21:16
cdentaye21:17
melwittI'm not sure exactly what the question is, but if some placement things are going to use storyboard, it seems like all of them should, to make things less confusing21:21
cdentmelwitt: yeah the question sort of evolved. that answers one of them21:21
cdentthe other, more sore of latent one was: how, when and what should be migrated21:22
cdentbecause we have some things already in progress in nova's launchpad, we've discussed letting those things ride21:22
melwittoh, I see21:24
melwittI guess if there's a storyboard tool for migrating in-progress things (that will copy all existing comments over) then you could. if the existing bug comments don't get copied, I'd think it better to leave them in launchpad. though the danger there is, what if it takes months to close out the bugs, then you have them in separate places and people will be confused21:26
melwitt(some bugs in launchpad, other bugs in storyboard)21:27
*** e0ne has joined #openstack-placement21:28
cdentyes, there is such a migration tool and I think that will get involved at some point but we don't have to do it at the same time as creating the storyboard stuff21:29
cdentre confusion: it's already pretty confused :)21:29
cdentI think during the transition, no matter we do, we are going to have pay extra attention and do extra handling21:30
melwittwhat do you mean by "create the storyboard stuff"? start using storyboard for new bugs?21:31
cdentmelwitt: no just have a placement project exist21:32
cdent(in storyboard)21:32
melwittI guess in general, I'd aim to have everything move and start using storyboard at the same time, and leave lp behind at the same time, if possible21:32
melwittto ease confusion, but as you said, I'm sure whichever way you go with, people will manage21:32
cdentWhat I'm trying to do is make sure that storyboard is already there so that we don't have to wait for that part once we have fully decided how we are going to do the rest of it21:32
melwittis there a lot involved in making it be there? (I've never added storyboard to a thing before)21:33
cdentyou make a project-config change some magic happens in the background21:35
cdentso no, not much21:35
cdentit's the migration that is a bit more involved21:35
cdentbut it would be nice to be able to spend some time learning about how to make storyboard go, and we recently being having a lot of things that sounds like stories that need to be written down somewhere21:36
cdentand while an etherpad would be fine, a task tracker would be even better21:36
melwittok. I don't have a strong opinion about it, just that it would seem nice not to have things split between the two tools for any time, but I'm sure it will be ok whichever way you do it21:36
mriedemi'd be willing to bet the SB migration tool will not handle migrating only a subset of nova bugs to a project in SB21:40
mriedemi.e. only those that are still open with a 'placement' tag21:41
cdentmriedem: there's a tag filter dealy according to fungi21:41
mriedemok we'll see :)21:41
mriedemis there a dry-run option?21:41
cdentprobably21:41
mriedembecause if that tag filter is buggy your migration is going to run for a few months21:41
cdentwell, if were doing _my_ migration we'd not be doing any migration21:41
cdentso our migration might run for months, but I hope not. apparently the main way to do a dry run is to migrate to a dummy project21:42
mriedemthe other thing is,21:42
mriedemof these 35 open nova 'placement' bugs https://bugs.launchpad.net/nova/?field.searchtext=&orderby=-importance&field.status%3Alist=NEW&field.status%3Alist=CONFIRMED&field.status%3Alist=TRIAGED&field.status%3Alist=INPROGRESS&field.status%3Alist=INCOMPLETE_WITH_RESPONSE&field.status%3Alist=INCOMPLETE_WITHOUT_RESPONSE&assignee_option=any&field.assignee=&field.bug_reporter=&field.bug_commenter=&field.subscriber=&field.structu21:42
mriedemsubscriber=&field.tag=placement&field.tags_combinator=ANY&field.has_cve.used=&field.omit_dupes.used=&field.omit_dupes=on&field.affects_me.used=&field.has_patch.used=&field.has_branches.used=&field.has_branches=on&field.has_no_branches.used=&field.has_no_branches=on&field.has_blueprints.used=&field.has_blueprints=on&field.has_no_blueprints.used=&field.has_no_blueprints=on&search=Search21:42
mriedemyikes21:43
cdentI'd really much rather we select only the important stuff, and rewrite it as humans into the new project, but I understand that sort of thing makes people nervous or unhapy21:43
mriedemthey aren't all placement bugs21:43
mriedemthey are nova bugs dealing with placement21:43
cdentright21:43
cdentwhich is again, why I think a mechanical migration is not great21:43
cdentwe should just suck it hup and do some human work21:43
mriedemthe open bug backlog is probably small enough to just port things over manually and close the bugs that get "migrated" from lp21:43
cdentsince it is only 35...21:43
cdentyes21:43
mriedemdoes uh google index storyboard stuff yet?21:44
cdentlot of those 35 are either not germane or stale21:44
* cdent doesn't konw21:44
mriedemsearching "story openstackclient bug" yields only lp21:44
mriedemand osc has been in sb for awhile21:44
mriedemanywho21:45
mriedemnot really your problem21:45
cdentI would guess that storyboards heavy use of js is a factor21:45
cdentyeah, and it's not like there's much of a choice21:45
mriedemyour problem is it's 9:45pm and you're thinking about this21:45
cdentI spent must of the earlier parts of the day laying in the sun outside21:45
mriedemisn't there some boiled meat to be had?!21:45
cdentand if I didn't do this now I would get to speak with your charming gob21:46
cdentno, what we have is some accidentally over-garlicked hummus21:46
cdenttruly astonishing21:47
*** ttsiouts has joined #openstack-placement21:56
cdentgood night21:58
*** cdent has quit IRC21:58
*** e0ne has quit IRC22:03
*** ttsiouts has quit IRC22:16
*** ttsiouts has joined #openstack-placement22:16
*** ttsiouts has quit IRC22:21
*** ttsiouts has joined #openstack-placement22:28
*** ttsiouts has quit IRC23:30
*** ttsiouts has joined #openstack-placement23:31
*** ttsiouts has quit IRC23:35

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