Monday, 2015-04-20

*** david-lyle has quit IRC02:55
*** flaper87 has quit IRC06:45
*** flaper87 has joined #openstack-relmgr-office06:45
ttxSergeyLukjanov: ping me when around08:24
ttxflaper87: same08:24
flaper87ttx: here08:24
ttxflaper87: ohai! I was looking into zaqarclient tests08:25
ttxIn particular
ttxBut looks like tests arte broken on all branches08:25
ttxIf I am to trust,n,z08:25
ttxAnything we could/should do there ?08:26
flaper87ttx: yeah, I'm working on fixing that. It's actually related to devstack and not the client. Will try to tackle that asap08:26
ttxcool, because this may block fixing requirements if we don't08:26
flaper87I know what's going on, which is 1 step forward :P08:26
flaper87Now I need to fix it :D08:26
ttx(we may be able to work around it, but simplerr to just get it fixed08:26
* flaper87 stops procrastinating and fixes the gate08:26
ttxflaper87: cool thanks08:27
flaper87ttx: np, thank you for raising it08:27
*** zz_johnthetubagu is now known as johnthetubaguy08:29
SergeyLukjanovttx, /me somewhere here09:14
ttxSergeyLukjanov: I think we got it covered thanks09:20
ttxSergeyLukjanov: maybe +2
SergeyLukjanovttx, 175303 merged09:49
johnthetubaguyttx: we should talk nova RC2 again soon, hows things looking now?09:52
ttxjohnthetubaguy: looking better09:53
ttxjohnthetubaguy: I'd like a quick talk with dhellmann first, but I think stable/kilo is looking relatively well now, good enough to open RC2 windows and merge backports09:54
ttxwe still need to work on master and libraries, but RC2s on the main projects should be ok09:54
ttxjohnthetubaguy: could we talk about that in a couple of hours, once I disucssed status with Doug ?09:57
johnthetubaguyttx: yeah, thats fine09:57
ttxjohnthetubaguy: congrats on your election btw09:57
ttxjohnthetubaguy: hope you'll be able to free up enough time ;)09:58
johnthetubaguyttx: yeah, finger crossed, my manager seemed supportive before the election at least!10:11
sdaguemorning ttx10:26
sdaguewhen you get a chance, can we chat about the oslo messaging bug / requirements fix?10:27
ttxsdague: yes! Maybe after my lunch / after your breakfast :)10:29
sdaguesounds good, just let me know10:29
ttxsdague: available now11:54
sdagueok, cool11:54
sdagueso... oslo.messaging11:54
ttxMy understanding is that you need a new version cut and a requirements update11:55
sdagueactually, in the other order11:55
* ttx chekc tsh req update11:55
ttxThe req update would only affect oslo.messaging, right11:55
ttxor would we also bumpt the min oslo.messaging in requirements ?11:56
ttx(as a step 3)11:56
sdaguewell, that's a good question11:58
sdaguehere is the situation11:58
sdague1.8.1 with py amqp 1.3, is basically going to be a disaster for people11:58
sdaguebecause 1.8.1 enables heartbeats by default11:59
sdagueand amqp 1.3 easily goes into an infinite loop fail state with that code11:59
ttxsdague: I guess that raises the question of what warrants a min bump. Looks like the line is a bit grey12:00
ttxespecially around release time when bumping mins has a cost to distros12:00
ttxIf we take lifeless terminology, the min cap is part of the "known bad" definition12:01
sdagueso, my feeling is a 1.8.2 with heartbeat off by default, and requiring versions of amqp that should work, seems good12:01
ttxIf it is 1.8.1 only we could !=1.8.1 ?12:01
sdagueso, that would mean reverting to 1.8.0, effectively12:01
sdaguewhich drops code that people want, this heartbeat code12:02
ttxwell no, we would still do 1.8.212:02
ttxso that would mean "1.8.0 or 1.8.2"12:02
ttxnot sure if that would work, in practice12:02
ttxchecking how long ago 1.8.0 was done12:03
sdagueyeh, that being said, 1.8.1 will work ok if you pull up the requirements as well12:03
ttxAlso sounds weird that you would enable heartbeats in a .Z, but meh12:03
sdagueyeh, it's a new feature, and it's enabled by default12:03
ttxsounds a bit too much for a .Z in my book :)12:04
ttxhmm, if 1.8.1 works if you pull up reqs, I guess we don't have to bump the min12:04
ttxjust unsure when we'll be able to push that, waiting for dhellmann to have a pow-wow on that12:05
sdagueI still think we need the 1.8.2 that forces the working level of requirements, because there are a number of folks that discovered this issue12:05
ttxthat = releasing a 1.8.212:05
ttxsdague: ++12:05
sdagueI'd be ok not making 1.8.2 the required min in kilo, we should probably just recommend it in release notes though12:05
ttxsounds good12:06
ttxLet's wait for dhellmann to get up before we unblock anything12:06
ttxBeen working over the weekend to advance the work on unblocking reqs, and I think we are looking better now12:07
ttxmestery: ping me when around12:36
ttxredrobot: ping me when around re: Barbican RC112:38
mesteryttx: o/12:41
ttxmestery: hi! I need your advice on one backport that seems to be needed to unblock stable/kilo testing12:42
mesteryttx: shoot!12:42
ttxLooks like we need to unbork the branch12:42
mesteryttx: Looking ...12:43
ttxthat is neutron-fwaas12:43
ttxthe others all could merge stuff in12:44
ttx"stuff" being
ttxIf you agree that shoul dbe part of the release, +212:45
mesteryheh, yes12:45
mesteryI don't see a problem with being a part of the release12:46
ttxjust found it weird enough as a backport to ask for your blessing first12:46
mesterySo +212:46
ttxrather than nikapproving it12:46
ttxmestery: cool thx12:46
mesteryttx: Looking through the list, the important ones all have backports out12:46
mesteryAnything else is gravyu12:46
mesteryAlthough I would like to cherry-pick this one, it finally merged over the weekend:
openstackLaunchpad bug 1438040 in neutron "fdb entries can't be removed when a VM is migrated" [High,Fix committed] - Assigned to Kyle Mestery (mestery)12:47
mesteryDoing it now12:47
ttxok, we'll consider the list for RC2 at the meeting tomorrow12:47
mesteryNM, that one is already there12:47
ttxall branches should be fine by then12:47
ttxdhellmann: around ? Would like to discuss status after the weekend work12:51
* ttx brbs12:53
*** superdan is now known as dansmith13:31
*** russellb has quit IRC13:40
*** russellb has joined #openstack-relmgr-office13:45
sdaguettx: quick thing, when you are rb14:22
sdaguecould you make the columns for - (the cross project submissions) word wrap?14:22
ttxsdague: I regularly do. Damn Google Docs just adds unwrapped lines14:24
ttx(ignores existing formatting)14:24
ttxlet me clean it up again14:25
ttxsdague: done14:25
ttxsdague: OK, let's assume dhellmann is on travel-back day. I'd like to sanity-check that we can probably open RC2 windows at this point14:35
ttxOver the weekend I managed to push a .gitreview change in every integrated release project stable/kilo branch14:36
sdagueok, let me get another cup of coffee, back in a couple of minutes, but I'll read scrollback14:36
ttxThe only broken stable/kilo branch at this point is zaqarclient, but that's because its tests are universally broken, nothing stable/kilo specific14:37
ttx(flaper87 is working on fixing that)14:37
ttxSo I think at this point we can safely open RC2 windows for integrated projects14:37
ttxWe may even be able to issue the .Z kilo library releases mentioned at point 1314:38
ttxI don't think those would affect the unborking on the requirements/master/library situation14:39
ttx(which we can discuss next)14:40
sdagueok, that's all probably true14:40
ttxhmm, wondering about one thing14:40
ttximagine a lib with 1.0.0 in stable kilo and no 1.1.0/liberty version yet14:41
ttxif we issue 1.0.1 in kilo and master req is uncapped, that will be used for master testing14:41
ttxI.. don't think that would be bad, right ?14:42
ttx(not good, but not bad either ?)14:42
sdagueit could break the world14:42
sdaguebecause of the requirements mismatch problem14:42
ttxoh, aye14:42
ttxinheirting the capped deps from stable/kilo14:42
ttxso we actually need to complete point 9.5 before14:43
sdagueso what's the deal with barbican -
ttxbarbican hasn't done its RC1 yet14:43
sdaguealso... I think this unwind is making me think we should never have done this14:44
ttxbeen trying to get hold of redrobot14:44
sdaguearen't they way past when the RC1 is expecteD?14:44
ttxwell incubated projects were in theory not affecting the rest of the process14:44
ttxso they can in theory do what they want, they are not even under release management14:44
ttxthey can even miss the release, at least in theory14:45
ttxwon't hold it on an incubated project14:45
sdagueok, except, they can break everything because of this requirements issue14:45
ttxright. Fun eh14:46
sdaguerequirements/project.txt seems like it has to be the list of the things that get active release management14:46
ttxwell, only because some integarted project depends on them14:46
sdagueand those are far from being true14:46
ttxright, we need to staff up14:46
ttxI also wonder if some sanity-checking of tags on libraries by release management wouldn't be a good thing too14:47
ttxsince they seem to happen at the weirdest time14:47
ttxjust because PTls can tag14:47
ttxbut tha's another discussion14:48
ttxsdague: would we start 9.5 before 9.4 is completed ?14:48
*** russellb has quit IRC14:48
ttxThat depends on whether any lib depends on barbicanclient kiteclient and zaqarclient14:49
sdagueso, honestly, barabican client might be untested enough we can get around it14:49
sdagueyeh, I don't think they do14:49
sdagueso yes, I think 0.9.5 can move forward14:49
sdagueI'd start with trying to get all of oslo released14:49
ttxAlso neeed to doublecheck notmyname's view on
ttxI think he is "right"14:50
sdaguenotmyname is mostly right, I don't know about the sphinx line though14:50
ttxgiven that swift didn't sync anything, they don't need uncap14:50
ttxright, but they shouldn't break the world14:51
ttxOK, I'll list all libraries that need a liberty release14:51
ttxso we start 9.514:51
ttxI expect dhellmann to do Oslos when he shows up14:52
ttxsdague: thanks for your help validating/fixing my assumptions :)14:52
sdagueno worries14:53
ttxI'll say that we almost had a boring release14:53
*** russellb has joined #openstack-relmgr-office14:53
ttxjohnthetubaguy: I guess we can lok into a RC2 Nova window now14:54
ttxif you have some time14:54
johnthetubaguyttx: hey, I have a meeting in 5mins I need to go to, but we can talk for 5 mins?14:54
ttxjohnthetubaguy: let's wait for end of your meeting14:55
ttxI have one in 5 min too14:55
johnthetubaguyttx: ah, OK, np14:56
johnthetubaguyttx: FWIW I think we need this one fixing urgently:
openstackLaunchpad bug 1445335 in OpenStack Compute (nova) "create/delete flavor permissions should be controlled by policy.json" [Critical,Confirmed] - Assigned to Divya K Konoor (dikonoor)14:57
ttxjohnthetubaguy: ack... Note that it's a good exercise to look into all those late "critical" bugs and wonder how we could catch them in testing next time14:58
johnthetubaguyttx: totally agreed, our policy testing is very weak at this point, it would seem :(14:59
johnthetubaguyttx: I think sdague's push for more functional in tree tests is a great step, we just need to add the tests...14:59
ttxsdague: actually if I understand correctly, all libs need a new liberty release to include the recent uncapping. So no point in listing which ones need it15:05
sdaguettx: yes15:06
sdaguewell, the point in listing is to figure out which are outstanding15:06
ttxRight, can do the full list15:07
redrobotttx o/  catching up on IRC15:14
ttxredrobot: so.. we kind of need an RC1 now15:15
ttxas it is blocking the rest of the process (didn't expect that, but now it does)15:15
ttxredrobot: at least we need a stable/kilo branch15:15
redrobotttx the last showstopper CR landed this morning, so I could release an RC1 now, but we may need an RC2 pending a bug in discussion right now.15:16
ttxredrobot: sounds good15:16
ttxcould you approve the open-liberty patch ?15:16
ttxWe may even be ablke to backport the bug you mention between the cutting of the stable/kilo branch and the tagging of RC215:16
redrobotttx ok, I'll get the Liberty CR merged shortly.15:17
ttxAlso, any reason to hold on ?15:18
ttxThat sounds unrelated to me...15:18
ttxredrobot: ^15:18
redrobotttx so we were wanting to release a barbicanclient 3.1.0 to include Kilo features, and I'm afraid uncapping the reqs now would break things when we release 3.1015:21
redrobotttx 3.1.0*  ...  but I need to check what the RC1 cap was for barbicanclient.15:21
redrobotttx if other projects are capped at barbicanclient <3.1.0, then I think we could merge that uncap now, and backport any bugfixes into a 3.0.x branch.15:22
ttxredrobot: yes, it's a bit late to do a major update to the "kilo" lib15:23
ttxYou can release a 3.1.0 as a "liberty" version of the lib15:24
ttxthat way that won't break testing15:24
redrobotttx ok, I doubt any other projects would be using the features not implemented in the client yet15:24
ttxso yes, please +2/APRV
ttxwe need the uncap in to unbreak the world15:24
redrobotttx done15:25
ttxredrobot: thx:15:27
ttxmorganfainberg: around?15:28
morganfainbergttx: yes I am.15:29
ttxIf you have kiteclient +2/APRV, could you push ?15:29
morganfainbergJust getting moving for the day.15:29
ttxmorganfainberg: then we can find time to consider opening RC215:29
morganfainbergDon't have access for kite client.15:29
ttxdman, who does15:29
morganfainbergBarbican team I think15:29
notmynamettx: when you get a moment....15:30
ttxredrobot: I may also need you to push
morganfainbergThough it might be a dead project.15:30
morganfainbergDon't think it was really developed.15:30
ttxhmm, so we could ignore it15:31
* ttx checks if they ever merged the offending requirements15:32
redrobotttx workflowed the CR, but yeah, Jamie Lennox is the main contributor to that project and he hasn't done any work on it in some time.15:32
ttxredrobot: ok15:32
ttxmorganfainberg: ping me when you're awake enough to discuss RC2 opening15:33
morganfainbergttx: awake enough now.15:34
ttxI need a sec though :)15:34
morganfainbergSure. I'm writing emails and updating bugs / closing bugs that would break people.15:35
ttxmorganfainberg: ok, around now15:43
morganfainbergttx: :)15:43
ttxso.. keystone rc215:43
morganfainbergYay. Rc2 :P15:44
ttxWe kind of already opened it to merge
ttxso let's make it official15:44
morganfainbergSounds good.15:44
ttxThen we can check what's on,n,z15:44
ttxand see if anything else would be interesting to add15:45
morganfainbergThere are about 5-6 bugs I would like to land to rc2. The one you linked was the #1 priority15:45
ttxbah, already opened15:45
* ttx cleans it up15:46
morganfainbergOne of the bugs is just man page updates / sample config. Since we had the rc2 window.15:46
morganfainbergSo minimal things to hit rc2 but all are useful / good fixes.15:47
ttxThe 3 listed there are proposed for merging on stable/kilo15:48
morganfainbergThat looks about right.15:48
openstackLaunchpad bug 1443598 in Keystone kilo "backend_argument containing a password leaked in logs" [High,In progress] - Assigned to Doug Hellmann (doug-hellmann)15:48
morganfainbergThe only other thing should be man page / sample config update(s)15:48
openstackLaunchpad bug 1441300 in Keystone "keystone-manage man page updates" [Medium,Fix committed] - Assigned to Lance Bragstad (lbragstad)15:49
morganfainbergYeah. We need to generate specifically for the branch though.15:49
morganfainbergCan't just backport.15:49
ttxthey have backports proposed already15:49
morganfainbergSince Liberty dev has already happened.15:49
morganfainbergI'll have the man page/sample config up today.15:49
ttxmorganfainberg: do they need to be regenerated ? Already proposed at
morganfainbergThe other one I'd be ok with also landing. Also no backport.15:50
ttxIf they need an update, you should -2 this one15:50
morganfainbergI'll 2x check that one make sure they are generated correctly.15:51
morganfainbergAnd push a change up if they aren't.15:51
morganfainbergThis morning.15:51
ttxOK, does that look complete enough at this point ?
morganfainbergThe other bug re: backend arguments should be in rc2.15:52
*** russellb has quit IRC15:53
ttxquick checking other fixcommitted bugs15:54
openstackLaunchpad bug 1421968 in Keystone "List Endpoint Groups Associated with project not routed" [Medium,Fix committed] - Assigned to Lin Hua Cheng (lin-hua-cheng)15:55
openstackLaunchpad bug 1430951 in Keystone "Revocation causes duplicate (and overly broad?) events in revocation table" [High,Fix committed] - Assigned to Alexander Makarov (amakarov)15:55
ttx1421968 was explicitely moved to L15:55
morganfainbergyeah that one isn't going to land in K15:56
morganfainbergit's a massive mess to unwind15:56
morganfainbergand has been broken for a looooong time15:56
morganfainbergor wait15:56
morganfainbergthere were three of those bugs that were similar15:56
morganfainberglet me see which one that was15:56
morganfainbergthis one can land in K15:57
morganfainbergjust landed late15:57
morganfainbergso missed rc115:57
morganfainbergthe other one was group-based revocations15:57
morganfainbergand that was a mess15:57
*** russellb has joined #openstack-relmgr-office15:58
morganfainbergbut i want to say this was an odd batch15:58
morganfainberglets not target it to rc215:58
ttxnotmyname: we can talk now15:59
notmynamettx: thanks15:59
notmynamettx: we're working on RC2 patches (or getting them lined up). should have a good list by the end of my day16:00
ttxnotmyname: I guess we should push that requirement bump16:00
notmynamettx: but I'm curious on the status of bumping the dependency in global requirements16:00
notmynamedo you need anything from me on it?16:00
ttxI'd have preferred to get Doug's opinion on it, since he was on top of the master issues16:00
notmynamewhen will he be around?16:00
ttxbut if sdague blesses it, I can't find a good reason why not to push it now myself, given it only impacts swift16:01
ttxsdague: what do you think of ? I think it's safe at this point16:02
ttxespecially with swift manual dep update16:02
sdaguettx: yeh, that seems fine16:03
sdagueit's a known RC bug right?16:03
notmynameas soon as it lands, we'll land a patch to update the requirements and fix the issue in the code16:03
sdagueshould we get the bug ref in the commit message?16:04
ttxoops, approved them16:04
notmynameI can post-facto create on if the process requires it16:04
ttxnotmyname: let's open the rc2 milestone for swift16:04
sdaguewell, normally stuff like that requires a bug in the commit message for tracking purposes16:05
sdaguebut, ttx already approved, so be it16:05
notmynamethanks ttx16:05
ttxI'll require the bug on the swift fix :P16:05
notmynamettx: what does "open the rc2 milestone" mean for what we do today?16:05
ttxYou can create the bug corresponding to the fix and target it to that ^16:06
notmynamecan we also land so we can get patches to master before backporting for rc2?16:06
notmynameah, nm. refesh shows it's there already :-)16:06
ttxnotmyname: When the bug is created, would be good to reference it in
ttxthen I can approve that change if (1) it corresponds to a bug being targeted to RC2 and (2) it's been fixed in master already16:08
ttxthe goal of the process is to avoid losing fixes between branches16:08
ttxSo you'll need an equivalent of for the master branch16:08
ttxand get that merged16:09
notmynameFWIW, looks like I can +2/+A on stable/kilo for swift16:09
ttxand looks like I can't16:09
ttxprobably some mismatch when we pushed the special pre-release ACL16:10
notmynamethe tables have turned ;-)16:10
ttxnotmyname: so I'll rely on you :)16:10
ttxnotmyname: so.. create LP bug, target it to RC2, reference it in, create a master equivalent of and merge it16:11
ttxthen we can approve
ttxIs there anything else to merge for that fix ?16:11
ttxIs there any other fix you'd like to see before we respin RC2 ?16:11
notmynamethere are a few others. I'll have a good picture by the end of my day today16:12
ttxhmm, err16:12
notmynameneed to validate those listed and get them landed16:12
notmynameif necessary16:12
ttx is actually very confusing16:12
ttxIt's for master but with topic "stable/kilo"16:13
ttxso I assumed it was a stable/kilo thing16:13
notmynameah, that looks wrong16:13
ttxIt's fine, just needs the stable/kilo backport :)16:14
notmynameI'll take care fo that today16:14
ttxthat may explain why I couldn't +2 it :)16:15
ttxtables turn again16:15
ttxYou seem to have on the backport list too16:15
ttxWould be good to have a RC2-targeted bug for taht one as well16:16
ttxnotmyname: then we can pick up the result at the sync tomorrow16:17
ttxsounds good. Talk to you tomorrow16:17
johnthetubaguyttx: hi, I am more free now, when you are16:18
ttxjohnthetubaguy: I think we lost the window. Let's talk at the 1:1 sync tomorrow16:18
* ttx needs to cut Barbican RC1 now16:18
johnthetubaguyttx: OK, np16:19
ttxredrobot: -- keep me informed if you need an RC216:34
redrobotttx will do, thanks.16:34
sdaguettx: silent just brought up - again in #openstack-oslo16:41
ttxsdague: I agree on the plan, not sure the timing is right though16:43
sdaguettx: ok16:43
ttxi.e. not sure what this requirements/master merge will (or rather won't in the current state) trigger.16:43
ttxthe sync jobs and some test jobs being desactivated and all16:44
ttxso let's wait until tomorrow and dhellmann's return16:44
*** russellb has quit IRC16:55
*** russellb has joined #openstack-relmgr-office16:58
*** russellb has quit IRC17:36
*** russellb has joined #openstack-relmgr-office17:40
*** russellb has quit IRC17:47
*** david-lyle has joined #openstack-relmgr-office17:50
*** russellb has joined #openstack-relmgr-office17:52
*** david-lyle_ has joined #openstack-relmgr-office18:01
*** david-lyle has quit IRC18:01
*** david-lyle_ is now known as david-lyle18:06
*** david-lyle has quit IRC18:20
*** david-lyle has joined #openstack-relmgr-office18:20
*** david-lyle has quit IRC18:41
*** johnthetubaguy is now known as zz_johnthetubagu19:19
*** asalkeld has joined #openstack-relmgr-office21:15
*** david-lyle has joined #openstack-relmgr-office22:04
*** david-ly_ has joined #openstack-relmgr-office23:44
*** david-lyle has quit IRC23:45
*** david-ly_ is now known as david-lyle23:46

Generated by 2.14.0 by Marius Gedminas - find it at!