Tuesday, 2011-05-10

*** dragondm has quit IRC00:01
*** mattray has quit IRC00:01
*** mattray has joined #openstack-meeting00:03
*** mattray has quit IRC00:30
*** westmaas1 has joined #openstack-meeting01:01
*** westmaas1 has quit IRC01:02
*** adjohn has joined #openstack-meeting01:03
*** anticw has quit IRC01:19
*** anticw has joined #openstack-meeting01:19
*** dendrobates is now known as dendro-afk02:01
*** med_out is now known as med04:36
*** med is now known as medberru04:36
*** medberru is now known as medberry04:36
*** dendro-afk is now known as dendrobates06:36
*** dendrobates is now known as dendro-afk09:06
*** anticw has quit IRC09:55
*** anticw has joined #openstack-meeting09:57
*** adjohn has quit IRC10:09
*** GasbaKid has joined #openstack-meeting12:03
*** dendro-afk is now known as dendrobates12:16
*** rackerhacker has quit IRC12:26
*** rackerhacker has joined #openstack-meeting12:33
*** GasbaKid has quit IRC12:47
*** dprince has joined #openstack-meeting12:55
*** dendrobates is now known as dendro-afk13:06
*** dendro-afk is now known as dendrobates13:28
*** edconzel has joined #openstack-meeting13:36
*** edconzel has quit IRC13:40
*** edconzel has joined #openstack-meeting13:46
*** blamar_ has joined #openstack-meeting14:23
*** blamar_ is now known as blamar14:23
*** jkoelker has joined #openstack-meeting14:31
*** dendrobates is now known as dendro-afk14:42
*** rnirmal has joined #openstack-meeting14:47
*** dragondm has joined #openstack-meeting14:59
*** medberry is now known as med_out15:56
*** mattray has joined #openstack-meeting16:22
*** troytoman-away is now known as troytoman16:41
*** dprince has quit IRC16:46
*** dprince has joined #openstack-meeting16:57
*** dragondm has quit IRC18:17
*** dprince_ has joined #openstack-meeting18:30
*** dragondm has joined #openstack-meeting18:49
*** markvoelker has joined #openstack-meeting18:56
*** mattray has quit IRC19:08
*** dprince_ has quit IRC19:18
*** dprince has quit IRC19:18
*** eday has joined #openstack-meeting20:10
*** creiht has joined #openstack-meeting20:12
*** bcwaldon_ has joined #openstack-meeting20:29
*** bcwaldon has joined #openstack-meeting20:35
*** carlp has joined #openstack-meeting20:41
*** jk0 has joined #openstack-meeting20:46
*** johan___ has joined #openstack-meeting20:46
*** dabo has joined #openstack-meeting20:46
*** comstud has joined #openstack-meeting20:47
*** comstud has left #openstack-meeting20:47
*** comstud has joined #openstack-meeting20:48
*** salv-orlando has joined #openstack-meeting20:48
*** User has joined #openstack-meeting20:52
*** User has joined #openstack-meeting20:52
*** Tushar has joined #openstack-meeting20:52
*** pquerna has joined #openstack-meeting20:53
*** mattray has joined #openstack-meeting20:53
*** jwilmes has joined #openstack-meeting20:53
*** spectorclan_ has joined #openstack-meeting20:53
*** danwent has joined #openstack-meeting20:57
*** exlt has joined #openstack-meeting20:59
*** mattray has quit IRC20:59
*** markwash_ has joined #openstack-meeting20:59
*** mattt has joined #openstack-meeting21:00
*** jdurgin has joined #openstack-meeting21:00
*** dsockwell has joined #openstack-meeting21:00
* creiht bows21:00
* exlt lurks..21:00
dsockwellyeah i'm just going to listen in as well21:00
ttxGreetings from Budapest !21:00
vishy^^ my arm is broken21:00
uvirtbotvishy: Error: "^" is not a valid command.21:00
*** johnpur has joined #openstack-meeting21:01
ttxI'm not completely sober so please let me know if I don't make sense :)21:01
* jaypipes curties.21:01
jaypipesor curtsies. joke fail.21:01
* dsockwell slaughters a goat21:01
dabo \o21:01
ttxok, let's get started !21:01
openstackMeeting started Tue May 10 21:01:56 2011 UTC.  The chair is ttx. Information about MeetBot at http://wiki.debian.org/MeetBot.21:01
openstackUseful Commands: #action #agreed #help #info #idea #link #topic.21:01
ttxWelcome everyone to our weekly OpenStack team meeting...21:02
ttx#link http://wiki.openstack.org/Meetings21:02
*** jeffkramer has joined #openstack-meeting21:02
ttx#topic Actions from previous meeting21:03
*** openstack changes topic to "Actions from previous meeting"21:03
ttx* KnightHacker, zns to get the nobottle branch merged asap21:03
ttxany news on that front ?21:03
jaypipesttx: I'm subscribed to the GitHub repo and there's been no progress on nobottle.21:04
ttxhmm. I guess we should carry on then21:04
jaypipesttx: we had a 2 hour sprint planning with the Keystone team last wednesday and they are still assigning roles, responsibility etc21:04
ttx#action  KnightHacker, zns to get the nobottle branch merged asap21:04
ttx* antonym to raise a thread about IRC separation on the ML21:04
ttxI've seen that raised, did we reach a conclusion on that thread yet ?21:05
edayyeah, there was a decision email at the end, I think21:05
*** blamar_ has joined #openstack-meeting21:05
pvothe thread was to keep them the same for now?21:05
edayadding -dev21:05
ttxbetween vacation and UDS I'll admit not having followed every thread lately21:05
soren*cough* slacker *cough*21:06
* soren is no better, though. :)21:06
ttxsoren: so you finally finished that beer, I see21:06
kpepplei thought we agreed to add -dev also21:06
sorenttx: I have no idea what you're talking about.21:06
edaykpepple: yup21:06
ttxok, anyone disagreeing on that ?21:06
ttxI don't really care either way.21:07
edayshould probably join it :)21:07
sorenThe idea of making the split a per-topic one was dropped?21:07
jaypipesI believe so.21:07
*** sandywalsh has joined #openstack-meeting21:07
edaysoren: correct21:07
ttx#action antonym to make agreed channel split happen21:07
ttx#topic General release status21:08
edaychannel is there, everyone join and lets close the action item :)21:08
*** openstack changes topic to "General release status"21:08
ttxeday: he needs to post an email to officialize it on the ML21:08
antonymttx: will do, it's pretty much done21:08
ttxUnder the new release management rules, we are in open development stage, so the coordinated release requirements at this point are quite basic21:09
ttxOnce the Diablo plans are mostly finalized, I'll switch to tracking mode21:09
ttxThe idea being to track what features land and give the PTLs an idea of how well their plan is actually executed21:09
*** _0x44 has joined #openstack-meeting21:09
ttxStarting next week I'll also have a deeper look into how I could adapt the release status page to the new world order21:10
ttxQuestions before we switch to per-project status ?21:10
ttxok then21:11
ttx#topic Nova status21:11
*** openstack changes topic to "Nova status"21:11
ttxThe Diablo plan is at: https://blueprints.launchpad.net/nova/diablo21:11
vishyI've been going through the blueprints and prioritizing and targetting21:11
ttxvishy: I'd say it's still being worked on ?21:11
vishyI think I've got all of the major ones targetted21:11
ttxI admit that I haven't had a chance to look into it that much...21:12
vishyi need to find some people to assign a couple of the essential ones to21:12
ttxyes, I'd prefer if all the >Low actually have assignees21:12
vishybut it is going well (slower than i expected, there were/are a whole bunch of obsolete/hanging bps21:12
ttxvishy: targeting to diablo should give you a good list at https://blueprints.launchpad.net/nova/diablo21:13
ttxwithout needing to deprecate all the old stuff21:13
ttxthough cleaning up cannot hurt :)21:13
ttxThe first milestone, diablo-1, is scheduled for June 221:14
ttxThat means cutting the milestone release branch in 3 weeks.21:14
ttxLooking at https://launchpad.net/nova/+milestone/diablo-1 we have 19 blueprints targeted...21:14
ttx...which sounds a bit optimistic to me, but we'll see :)21:15
daboThe distributed scheduler merge prop should be landing in about 2 weeks. It will be a fairly large and disruptive change, so will that interfere with diablo-1?21:15
vishyttx: yes I may be moving a few of those21:15
*** grapex has joined #openstack-meeting21:15
ttxdabo: I don't think so, it's good if it actually lands in an early milestone...21:15
vishyI'd like to get distributed scheduler and multinic in asap21:15
johnpurvishy: agree, do it asap21:16
jaypipesvishy: FYI, Glance's diablo-2 milestone is where I'm targeting keystone integration, if you want to align on that...21:16
dabocool - I'll keep everyone informed as to its progress21:16
vishyjaypipes: cool, I think we're trying to get super ghetto version in in diablo-1, but our complete integration won't be until diablo-421:16
vishywe're going to go through a few iterations...21:17
jaypipesvishy: yup, cool.21:17
ttxvishy: anything else you wanted to mention ?21:17
westmaaswhat should we do if we think things will move from milestone 1 to 2?21:17
westmaasannounce in these meetings, update the bp? both?21:18
jaypipeswestmaas: both.21:18
jaypipeswestmaas: and if it involves other teams, post something to the ML.21:18
johnpurnotmyname: is swift aligning around keystone integration at the end of june?21:18
notmynameif you would like us to :-)21:19
vishyttx: nope21:19
ttxjohnpur: maybe we can talk about that when we switch topic to swift21:19
jaypipeswestmaas: at least, that's what I'll be doing ;)21:19
johnpurttx: ok21:19
ttxOther questions for the Nova PTL ?21:19
westmaasjaypipes: just looking for someone to copy :)21:19
jaypipesttx: I do, actually21:19
jaypipesttx, vishy: so, although I didn't get through a whole lot of reviews in my review day yesterday, I did notice a HUGE number of pending merge requests... perhaps we can send an email to the ML targeting a number of the high priority ones to review?21:20
vishyjaypipes: good idea21:20
jaypipesvishy: something like ttx's freeze emails.21:21
jaypipesvishy: in particular, I note that the NTT volume branch blueprints are in needs code review status...21:21
jaypipesvishy: ec2 volume stuff...21:22
_0x44Are all of those merge requests still expecting to be merged? When I did my review day I saw quite a few that seemed abandoned pre-summit waiting for summit decisions but were never updated.21:22
ttxyes and due to TZ differences the comment.fix iterations take a bit longer with japanese branch merge proposals21:22
vishyok I'll take a look at the merge list and fire out an email with important ones?21:22
vishyand request that people update their branches or put them to wip if they aren't ready21:23
ttx_0x44: the abandoned ones should no longer be in "needs review" status21:23
jaypipes_0x44: it's a shared responsibility of the reviewer and PTL to follow up with folks about stagnating merge proposals...and get stuff into Work in Progress if it is being updated.21:23
pvovishy: that would help. Quite a few branches are a couple of weeks old21:23
jaypipesvishy: yep, that sounds good. lemme know if you want assistance at all.21:24
ttxideally the branch review listing page would match the priorities21:24
ttxhaven't updated it though21:24
comstudi made a merge prop today for a bugfix that's kinda important to get in for xenserver21:24
comstudnot sure the bug was prioritized, however21:25
ttxwill have a fresh look at it next week so that it's useful again21:25
jaypipesI'd like to see https://bugs.launchpad.net/nova/+bug/771512 prioritized to Critical...21:25
uvirtbotLaunchpad bug 771512 in nova "Timeout from API with 50 Simultaneous Builds" [High,Triaged]21:25
antonymjaypipes: +121:25
jaypipesanyone know the folks at Linden and might be able to raise a fire?21:26
ttxjaypipes: any reason why you can't do that yourself ?21:26
comstudjay: that's going in progress after this meeting21:26
ttx(raising to critical)21:26
jaypipesthe reason I'd like to set it to Critical is because once mtaylor gets going with CI, that bug will cause major havoc.21:26
pvojaypipes: its causing havoc for our testing now.21:26
jaypipespvo: well, yes, I know :)21:26
comstudit's next on my list21:26
antonymyep :)21:26
mtaylorthat's fine - cause all the havoc you want :)21:26
* comstud steals it right now21:26
jaypipesok, super :)21:27
ttxmtaylor: o/21:27
ttxok, can we switch to glance status now ?21:27
jaypipesI'm ready.21:27
ttx#topic Glance status21:28
*** openstack changes topic to "Glance status"21:28
ttxThe Glance Diablo plans are at: https://blueprints.launchpad.net/glance/diablo21:28
jaypipesHere's the milestone we are focused on: https://launchpad.net/glance/+milestone/diablo-121:28
jaypipesFor the next 3 weeks, we are focusing on API improvements that will make glenc happy :)21:28
ttxsounds god to me21:28
ttxgood even21:28
glenc:D happy glenc is happy21:28
ttxjaypipes: other announcements, comments ?21:29
jaypipesI've already talked with soren about coordination for packaging, but vishy, we need to discuss affects on nova integration21:29
jaypipesvishy: mostly around the upgrade path for existing nova/glance installations. We can take it offline later on.21:29
ttxAny question for Jay ?21:30
jaypipesvishy: and if you have a chance, pls take a look at https://code.launchpad.net/~jaypipes/glance/api-version/+merge/6013021:30
jaypipesttx: good to move on I think.21:31
ttx#topic Swift status21:31
*** openstack changes topic to "Swift status"21:31
ttxnotmyname: did you have time to sort out your versioning scheme and milestone plan ?21:31
johnpurjaypipes: you are scheduled out only to diablo-2?21:31
*** mattray has joined #openstack-meeting21:31
notmynamenot yet, but there won't be much change from our current 1.X.Y21:31
ttxok, was wondering:21:32
ttxWe currently have a 1.4 "series", with no blueprints attached.21:32
ttxWe could have a "diablo" series instead, with milestones that happen to have whatever version number you want...21:32
*** midodan has joined #openstack-meeting21:32
jaypipesjohnpur: stuff to be discussed with usharesoft going into d3 and beyond... more to come.21:32
johnpurttx: we talked about syncing version nomenclature?21:32
ttxthough that doesn't make a lot of sense branch-wise, I guess21:33
*** devcamca- has left #openstack-meeting21:33
notmynamethe blueprints https://blueprints.launchpad.net/swift are what we have21:33
notmynamebut no, they aren't targeted to a diablo release21:33
ttxjohnpur: not yet... so far Swift wanted their own versioning21:33
*** devcamca- has joined #openstack-meeting21:33
johnpurjaypipes: cool. was wondering about the image conversion stuff that isn't showing up.21:33
ttxnotmyname: right, I'm trying to get a "diablo" picture21:33
ttxand "1.4" might just not be synonymous21:34
ttxsince you might decide other values for "X" above... IIUC21:34
notmynamewe've been doing some internal planning here, but there isn't anything for "diablo" yet21:34
notmynameand agreed about 1.4 not necessarily == 1.421:34
ttxnotmyname: ok, I guess we'll wait a bit more then :)21:34
johnpurnotmyname, ttx: can we talk about getting a common version scheme? maybe outside of this meeting... it is a pain to do queries against non-normalized version strings.21:35
ttxjohnpur: we need to have a way to query what falls into the "diablo" timeframe and will become part of openstack 2011.321:35
johnpurnotmyname: what is your plan to support keystone?21:35
johnpurttx: yes!21:35
ttxjohnpur: that might not imply forcing Swift to adopt the same version numbers for milestones21:36
notmynamejohnpur: yes, all that's needed to do is some wsgi middleware21:36
ttxjohnpur: agreed that we can discuss that offline21:36
jaypipesnotmyname: I think johnpur was asking about a general timeframe :)21:36
johnpurjaypipes: :)21:37
jaypipesnotmyname: nova is trying for end of this month, glance is trying for end of july to integrate with keystone, as an FYI.21:37
notmynamewe don't know what needs to be done yet. we can be done when we see what keystone requires21:37
jaypipesnotmyname: sorry, I meant end of june, not july...21:37
*** katkee has joined #openstack-meeting21:38
*** pino_ has joined #openstack-meeting21:38
jaypipesnotmyname: ok, however I think some things will come down to what *you* require from keystone, which is why communication with the keystone folks will be important. just a though, not trying to harrass you :)21:39
notmynamegholt and ziad have been talking about it21:39
ttxnotmyname: any other announcements or comments ?21:39
jaypipesok, coolio.21:39
notmynamettx: nope21:39
ttxQuestions for the Swift team ?21:39
ttxok then...21:40
ttx#topic Open discussion21:40
*** openstack changes topic to "Open discussion"21:40
ttxIn case you missed the press, news from UDS here is that Ubuntu Server is switching their default solution for Ubuntu Cloud Infrastructure from Eucalyptus to OpenStack.21:40
ttxNot really a surpise, but still, good news21:41
jaypipeshmm, I guess we should fix that 50 concurrent clients bug then ;P21:41
pvoha. nah.21:41
sorenjaypipes: I'm quite close on that, by the way.21:42
jaypipessoren: pls communicate with comstud :)21:42
sorenjaypipes: Oh, he's working on it, too?21:42
jaypipessoren: yep21:42
sorenI mean *I'm* commenting on the bug and stuff.  :)21:42
sorencomstud: Well if you've got it, fine. I'll just stop playing around.21:44
antonymsoren: yeah, saw the of the updates on the bug report21:44
vishyfirst person to patch eventlet wins21:44
comstudyeah, i just read all of the notes21:44
comstudI didn't realize it had been triaged so much21:44
comstudI don't care who does it as long it gets addressed RSN21:44
comstudi can take a look at patching eventlet21:45
sorenI have a pathc that fixes it for us, but not generally, so I'm not really happy with it.21:45
* vishy wants to see that patch21:45
comstudme too21:45
antonymme too21:45
johnpurredbo is an eventlet expert too21:45
antonymi can give it a whirl in our environment21:45
jaypipesvishy, notmyname: we still on to discuss commonalities in a few minutes on this channel?21:45
sorenIt fixes it in the case whee we never really block for the full socket connect timeout at time, but that's not good enough for upstreaming.21:45
ttxsomething else someone wants to mention before we close the meeting ?21:45
vishyjaypipes: yes21:45
ttxthe PTLs need the room :)21:45
antonymi mainly just want to get past 50 builds so i can get some other bugs to come crawling out :D21:46
jaypipesttx: others are welcome to linger as far as I'm concerned.21:46
ttxoh right, just need to formally pass the bucket21:46
ttxso let's just do it... now21:47
*** openstack changes topic to "Openstack Meetings: http://wiki.openstack.org/Meetings | Minutes: http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/"21:47
openstackMeeting ended Tue May 10 21:47:38 2011 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)21:47
openstackMinutes:        http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-05-10-21.01.html21:47
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-05-10-21.01.txt21:47
openstackLog:            http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-05-10-21.01.log.html21:47
ttxThanks everyone !21:47
*** salv-orlando has quit IRC21:47
*** dabo has left #openstack-meeting21:48
*** jk0 has left #openstack-meeting21:48
*** pquerna has left #openstack-meeting21:48
vishynotmyname: here?21:48
comstudsoren: i'm not at all attached to the bug, if you want to continue rolling with it. otherwise, if it frees you up, i can take it.  i'd love to see your patch, though21:48
*** spectorclan_ has quit IRC21:49
jaypipesok, so we're supposed to discuss areas of code that should move to a common library.21:49
vishyso I think there are two broad topics21:49
jaypipesvishy, notmyname: how would you like to handle this? should we simply propose pieces/parts of code that are good candidates?21:49
vishy1) Can we create a common "service starter kit"?  A bunch of groups are requesting it.21:50
jaypipesvishy: like a skeleton project?21:50
vishy2) How can we make the projects as similar as possible for users and administrators21:50
vishyI think those things are driving the integration discussion.21:51
jaypipesvishy: I would definitely support making a skeleton project. that actually would be a good way to find: a) common utility code to go into the openstack-common library and b) best practice things in test cases, etc21:51
notmynameI think that (2) is key. and implies enough of (1) that it doesn't need to happen21:51
notmynamealthough I'm willing to be convinced, I don't see the benefit in a skeleton project21:52
jaypipesnotmyname: well, I think a skeleton project is really useful for new projects to start from...21:52
vishynotmyname: i don't quite understand that...21:52
vishynotmyname: all the projects do logging/wsgi/eventlet/rest apis/etc...21:53
notmynameguidelines to do (2) is good21:53
notmynamelike we discussed earlier21:53
notmynamecommon ways to start daemons, etc21:53
vishyit might be nice to have a place to start so a new project doesn't have to re-invent the wheel21:53
sorencomstud: As I said, it's a dirty hack. It doesn't *really * solve the problem, and I didn't keep it, so I'd have to reproduce it. I'd rather spend my time actually fixing the problem rather than making my way back to a half-arsed solution to it.21:53
jaypipesnotmyname: right, but a skeleton project has all those things "in place". The code becomes the best practice "document"21:54
jaypipessoren: heya, sorry, could you move that thread to openstack-dev? :)21:54
sorenjaypipes: :p21:54
jaypipessoren: thx man21:54
vishyall of the other projects are going to copy one of the projects anyway21:55
*** johan___ has left #openstack-meeting21:55
jaypipesvishy: right.21:55
vishyso it might be nice if they had a "best" starting point21:55
notmynamethings that join openstack should be established projects anyway21:55
jaypipesvishy: it would have been very useful for keystone to have a skeleton project.21:56
jaypipesnotmyname: how so?21:56
notmynameideas don't join openstack. code does21:56
jaypipesnotmyname: for instance, the volume, lbaas and network projects...21:56
jaypipesthose are all ideas, not code.21:56
vishykeystone, reddwarf, and quantum have all requested such a skeleton21:56
edayjaypipes: he's talking about ppb approval, not new things destined for OS21:56
notmynamevishy: guidelines for how this works are good21:57
edaynotmyname: I can't remember, were you around for the PPB meeting last week?21:58
jaypipesnotmyname: I think we are saying that creating a skeleton project *is* the guideline...21:58
notmynameeday: no, but I read over the logs21:58
markwash_(not a PTL) how changeable would a skeleton project be in case we found a better way to do one of the "common" things in the future21:59
vishywell we can at least make some progress on the guidelines21:59
*** ryu_ has joined #openstack-meeting21:59
edaynotmyname: ok. a lot of this is due to the "integation" part of it, it seems the PPB mostly wants highly integrated projects for approval, so it's not going to be a random existing project joining21:59
vishythen someone can create a skeleton based on the guidelines21:59
jaypipesmarkwash_: it would be an evolving set of best practices common to all openstack projects.21:59
danwentquick note to anyone joining and looking for the networking discussion, meeting has been moved to tomorrow at the same time due to a conflict on the channel21:59
jaypipeseday, notmyname: we're not talking about PPB stuff here...22:00
*** markvoelker has quit IRC22:00
edaywell, hopefully 90% of the 'skeleton' project would be in common libs being used by other projects too22:00
vishyeday: agreed22:00
vishyso we have one proposal that was made so far: daemonization22:00
edayjaypipes: wanted to be sure there was understanding of motives/goals first22:00
*** edconzel has quit IRC22:00
*** bcwaldon has quit IRC22:01
vishynext on my list would be configuration (especially for logging)22:01
jaypipeseday: totally agreed. but what a skeleton project does is prevent people from, say, copying Glance (or Swift, or Nova, etc) when creating their new endpoint project, and instead, the copy the skeleton project.22:01
edayjaypipes: I understand, and not saying it shouldn't exist at all :)22:01
jaypipesvishy: Gflags will be the big issue in that regard I'm afraid.22:01
*** Tushar has quit IRC22:02
vishyjaypipes: i understand that, but I think we can at least define a common logging.conf file and have each project parse it appropriately22:02
jaypipesvishy: I'm definitely cool with that.22:02
creihtisn't that already solved in python?22:02
vishycreiht: not really22:03
jaypipescreiht: yes, but the different projects have, for instance, different defaults for how they handle logging (syslog vs file, format of log records, etc)22:03
edayvishy: how so? logging.config seems sufficient22:03
*** grapex has quit IRC22:04
vishyeday: seems a bit low-level to me22:04
jaypipescreiht: currently, each project sets up its logging with different routines. In glance, it's in glance.common.config. In swift, it's in swift.common.utils, in nova, it's in nova.logging. Would be good to have standardized routines in a common library for initializing/setting up logging.22:05
creihtjaypipes: I'm not against that, I'm just saying why not use what is already built into python for that?22:05
creihtthe swift team has already talked about doing that internally22:05
jaypipescreiht: *sigh*. I'm not suggesting otherwise :)(22:05
vishycreiht: I suppose if we supply a really good default file it might work22:05
jaypipescreiht: I'm saying instead of code doing:22:06
jaypipesfrom swift.common.utils import setup_logging22:06
jaypipesyou'd do from openstack.common import logging22:06
jaypipesor something like that...22:06
creihtjaypipes: why even do that if you can do22:06
creihtfrom logging.config import...22:06
notmynameso the concern is the import line? I thought the concern was the backend of how loggis is used22:06
vishyjaypipes: i don't know if we're really going to get that far that way22:07
vishyjaypipes: i expect every project will have custom logging handlers etc to inject extra information22:07
jaypipescreiht, notmyname: omg. it's not about the import line. it's about the set of code lines that establishes logging functionality.22:07
jaypipesfor instance, in Glance, we have a --log-config option, that enables logging to be overridden with a logging.config file...22:08
creihtall I said was, I'm not against having a more common logging setup, all I suggested was to use what is already available in python22:08
vishyi'm more concerned about the command line options for telling it to syslog vs file log, verbosity, etc...22:08
jaypipeswould be great if ALL projects had a --log-config option, and that it was handled the same for all projects...22:08
jaypipesvishy: that's what I'm talking about as well!22:08
*** letterj has joined #openstack-meeting22:08
creihtor a standard python logging config that could be configured in a config file? :)22:09
vishycreiht: sure, as long as it easy to make these switches22:09
jaypipescreiht: sorry, could you elaborate on what you mean by that?22:09
creihtall I'm trying to say is python already seems to have a set of best practices for this, so why not just use that22:10
edayjaypipes: don't bother with command line args, as you convinced me to do in burrow. burrow.conf can just include a common logging.conf22:10
edaymuch easier, btw :)22:10
jaypipeseday: that's how glance works, too...22:10
jaypipescreiht: nobody is arguing to use anything other than standard python stuff... I'm not sure why you keep bringing that up?22:11
creihtjaypipes: because you keep asking me to? :)22:12
* creiht goes back into hiding22:12
jaypipesthis is so very frustrating.22:12
*** letterj has left #openstack-meeting22:13
*** jdurgin has left #openstack-meeting22:13
vishyso jaypipes, can you do what you are doing with your log config file using a logging.config file?22:14
jaypipesvishy: yes.22:14
vishyI assume we can in nova, but it might require some refactoring22:14
edayso, ConfigParser files is primary config mechanism, optionally can support command line options to inject values into ConfigParser, and logging.config for logging formats? seems like we're in agreement :)22:14
creihtso maybe a better question is, if we rely on the python best practices for logging configuration, what questions are left to answer?22:14
jaypipesvishy: you can either do --log-config=PATH on the command line, or you can put a log_config=PATH in the config files.22:15
vishysounds fine22:15
jaypipescreiht: defaults.22:15
vishyso that takes care of daemonization and logging22:15
vishyare there any other big pain points for people using all of the projects together?22:16
jaypipescreiht: for instance, glance defaults to file, swift defaults to syslog.22:16
edayvishy: I think config too. everything but Nova uses ConfigParser, so we just need to convert nova away from gflags22:16
termieeverything but nova is much simpler than nova22:16
vishyeday: that sounds painful :)22:16
creihtI would argue is that nova is only more complicated because it uses gflags :)22:17
termiecreiht: lulz22:17
edayvishy: not saying it isn't painful, but needed if we want this level of openstack integration :)22:17
vishya lot of the stuff we are using flags for needs to move in general22:17
termiefor the record it is much easier to convert things to gflags than vice-versa22:17
termiegflags offers more features22:18
jaypipesbeg to differ on that one. :)22:18
comstudand termie will arm wrestle anyone who thinks overwise.22:18
creihtI think it has more to do with philosophy22:18
termiecreiht: /etc/object-server/1.conf?22:18
vishytermie: I don't think the main issue is with the feature set, it is that most people don't seem to like it...22:18
creihtswift is everything is in a config file, command line options only when absolutely neccesary22:18
creihttermie: that is just a workaround for the saio22:19
termiecreiht: in about 20 config files22:19
edaycreiht: doing the same with burrow22:19
creihtnova is everything in command line options22:19
jaypipeslet's please not do the gflags vs. optparse/ConfigParser right now...22:20
vishycreiht: that isn't exactly true, everything can be specified on the command line22:20
jaypipesI'm too tired to care :)22:20
*** grapex has joined #openstack-meeting22:20
vishyjaypipes: is there anything else?22:20
creihttermie: there is one config file per service (proxy, object, container, etc.) not unreasonable since those are most likely on different servers22:20
jaypipesvishy: yes, actually :)22:20
edayvishy: are you opposed to moving to configparser for nova?22:20
jaypipesvishy: there's a bunch of stuff that's duplicated in the various "utils.py" files that exist in all the projects.22:20
creihtI would also argue that ops people like config files more, and makes for better deployment22:21
jaypipesvishy: there's also best practices in regards to running tests, etc...22:21
jaypipesvishy: but we could leave those for another day.22:21
termiecreiht: flagfiles work the same way as a config file but they don't require weird workarounds to support multiple processes22:22
termiecreiht: but we were asked to table the debate a bit, so perhaps we should rest?22:22
vishyeday: I'm not opposed to it, I just think it is a big change that we can't make right now...22:22
creihtI could care less actually :)22:22
jaypipesvishy: I would agree with you.22:22
vishyjaypipes: testing++22:22
vishyjaypipes: titan team is putting together some documentation and examples for unittests22:23
jaypipescreiht: hmm, I thought you *couldn't* care less ;)22:23
edayvishy: will it ever be easier?22:23
creihtthat's the problem with a lot of the common stuff, is that it isn't so much about features but philosophy22:23
jaypipesvishy: k, that's good to hear.22:23
vishyeday: yes, after NaaS and BSaaS22:23
vishyjaypipes: i think with those examples and maybe some stuff from you on how you're doing functional testing22:24
vishywe are pretty good there22:24
jaypipesvishy: ok22:24
vishyutils stuff I don't see as a huge issue.22:24
jaypipesvishy: k.22:24
jaypipesthen it sounds like other than config/logging stuff, there's really no impetus to do a common library, since there's little agreement on philosophy.22:25
*** grapex has quit IRC22:25
vishyi mean if someone wants to go through them all and pull out common stuff, i wouldn't be opposed...I just don't think that part is a huge priority22:25
jaypipesvishy: I never said this stuff was a priority, just something we could work towards :)22:25
creihtand I still think that if something is important enough to pull out to be in common, that maybe we should evaluate if it would be better pull it out as a library instead22:26
creihtthere is cohesion around dameons, why not use that as a proof of concept first, and see where it goes from there?22:26
creihtweather daemon stuff is in its own lib, or in openstack.common, start with something and see how it goes?22:27
jaypipescreiht: there's no disagreement on that. that's happening.22:27
vishycreiht: that seems fine22:28
jaypipescreiht: re: pulling daemon stuff out into a separate non-openstack-common lib.22:28
vishyi just was hoping we could get something for these other projects to start with22:28
vishyso logging config, we can all add a blueprint for that?22:29
creihtIt seems pretty sad to me that if someone is seriously trying to start such a project, they need that much hand holding to start22:29
vishy(those that don't have it yet)22:29
*** blamar_ has quit IRC22:29
vishyand we tell other groups to just use glance as a base?22:32
vishyand pick and choose from other projects as needed?22:32
creihtIt seems like they should be able to figure that out for themselves?22:34
jaypipescreiht: you're not being particularly helpful here...22:34
jaypipesvishy: I still think a skeleton project would be useful, especially for some of these teams on projects that are used to Java, non-Python programming.22:35
vishyjaypipes: i agree22:35
vishyjaypipes: but who has time to make one?22:35
jaypipescreiht: nobody is *forcing* people to do this... we're just trying to lower the barrier.22:35
jaypipesvishy: I can do it.22:36
* termie is22:36
jaypipestermie: forcing people? :)22:36
vishyjaypipes: if you want to put something together that would be cool22:36
termietermie is can make time22:36
creihtright, because what we need to do to enable the development of scalable distributed systems is lower the barrier to entry22:36
vishytermie can do one too and they can fight to the death robowars style22:36
creihtsorry if I am a bit opinionated about that22:36
termievishy: works for me22:36
jaypipesI'm really not interested in fighting :)22:36
termiejaypipes: "collaborating" equal but separate22:37
vishyyour skeletons will fight to the death22:37
jaypipeshehe, ok no worries.22:37
vishymaybe you guys will accidentally agree on a bunch of stuff22:37
jaypipesvishy: termie agrees with people? ;P22:37
vishyand gflags vs config parsing can be a plugin :)22:37
termiei agree with eople who agree with me22:38
jaypipestermie: :)22:38
vishyreminds me of a friend who said: "I thought she had really good opinions.  She basically agreed with everything I said."22:38
termievishy: was i that friend?22:39
termievishy: 'cause i've said that22:39
notmynameso as not to drag this out too much, what needs to be done?22:39
vishytermie: haha no, that was my friend jon.  It was extra funny because he didn't really realize what he was saying22:39
vishynotmyname: sounds like 3 actions22:39
termievishy: ah22:39
vishy1: all projects support logging.config22:39
*** grapex has joined #openstack-meeting22:40
vishy2: jay to do daemon library breakout and we integrate with projects22:40
vishy3: termie and jay to make skele-projects22:40
vishydid i miss anything?22:40
jaypipesnotmyname: so you don't have to do anything. bonus.22:40
eday(future) kill-gflags :)22:40
vishy(we pushed the gflags discussion to later)22:40
termiep.s. broke out our gflags improvements and pinged the gflags team about them: http://github.com/termie/hflags22:40
creihtthat seems to be a recurring theme :)22:40
*** katkee has quit IRC22:42
vishyso done for today?22:42
notmynameI don't have anything else22:43
*** jeffkramer has quit IRC22:44
vishycool, thanks guys.  I'll draft an email to the list saying what was decided22:44
*** markwash_ has left #openstack-meeting22:45
* alekibango is here late... +1 for unity on logging.config22:48
*** eday has left #openstack-meeting22:49
*** troytoman is now known as troytoman-away22:54
*** rnirmal has quit IRC23:04
*** midodan has quit IRC23:12
*** jkoelker has quit IRC23:17
*** mattray has quit IRC23:24
*** mattray has joined #openstack-meeting23:26
carlpHey mattray23:26
mattraycarlp: howdy23:26
*** grapex has quit IRC23:27
*** danwent has quit IRC23:37
*** ryu_ has quit IRC23:39
*** pino_ has quit IRC23:39
*** danwent has joined #openstack-meeting23:41
*** _0x44 has left #openstack-meeting23:44
*** danwent has quit IRC23:44
*** johnpur has quit IRC23:48
*** danwent has joined #openstack-meeting23:53

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!