Tuesday, 2011-03-22

*** pvo has joined #openstack-meeting00:05
*** reldan has joined #openstack-meeting00:06
*** deshantm has quit IRC00:09
*** troytoman-away is now known as troytoman00:15
*** troytoman is now known as troytoman-away00:15
*** troytoman-away is now known as troytoman00:16
*** troytoman is now known as troytoman-away00:17
*** troytoman-away is now known as troytoman00:17
*** troytoman is now known as troytoman-away00:17
*** deshantm has joined #openstack-meeting00:24
*** ovidwu has joined #openstack-meeting00:34
*** reldan has quit IRC00:39
*** dragondm has quit IRC00:52
*** ovidwu has quit IRC00:52
*** ovidwu has joined #openstack-meeting00:54
*** deshantm has quit IRC00:58
*** deshantm has joined #openstack-meeting01:02
*** pvo has quit IRC01:17
*** dendrobates is now known as dendro-afk01:33
*** deshantm has quit IRC01:43
*** dendro-afk is now known as dendrobates02:08
*** deshantm has joined #openstack-meeting02:32
*** westmaas1 has quit IRC02:45
*** deshantm has quit IRC02:54
*** adjohn_ has joined #openstack-meeting04:05
*** adjohn_ has quit IRC04:07
*** adjohn_ has joined #openstack-meeting04:07
*** deshantm has joined #openstack-meeting04:11
*** adjohn_ has quit IRC04:12
*** adjohn_ has joined #openstack-meeting04:12
*** reldan has joined #openstack-meeting04:48
*** reldan has quit IRC06:20
*** romain_lenglet has joined #openstack-meeting07:10
*** romain_lenglet has left #openstack-meeting07:10
*** dendrobates is now known as dendro-afk07:29
*** reldan has joined #openstack-meeting08:29
*** alekibango has quit IRC09:35
*** reldan has quit IRC09:56
*** reldan has joined #openstack-meeting09:59
*** alekibango has joined #openstack-meeting10:32
*** romain_lenglet has joined #openstack-meeting10:52
*** romain_lenglet has left #openstack-meeting11:22
*** reldan has quit IRC11:23
*** dprince has joined #openstack-meeting12:05
*** reldan has joined #openstack-meeting12:22
*** pvo has joined #openstack-meeting12:53
*** chuck_ has joined #openstack-meeting12:56
*** zul has quit IRC12:57
*** chuck_ has quit IRC13:03
*** chuck_ has joined #openstack-meeting13:17
*** chuck_ is now known as zul13:19
*** zul has quit IRC13:19
*** zul has joined #openstack-meeting13:19
*** chuck_ has joined #openstack-meeting13:25
*** pvo has quit IRC13:26
*** zul has quit IRC13:27
*** troytoman-away is now known as troytoman13:44
*** dendro-afk is now known as dendrobates14:01
*** murkk has quit IRC14:14
*** pvo has joined #openstack-meeting14:29
*** murkk has joined #openstack-meeting14:35
*** dragondm has joined #openstack-meeting15:05
*** troytoman is now known as troytoman-away15:17
*** reldan has quit IRC15:18
*** chuck_ is now known as zul15:21
*** romain_lenglet has joined #openstack-meeting15:29
*** romain_lenglet has left #openstack-meeting15:37
*** dendrobates is now known as dendro-afk15:37
*** reldan has joined #openstack-meeting15:47
*** pvo_ has joined #openstack-meeting15:57
*** pvo has quit IRC15:57
*** dendro-afk is now known as dendrobates15:59
*** johnpur has joined #openstack-meeting16:03
*** troytoman-away is now known as troytoman16:09
*** pvo_ has quit IRC16:29
*** pvo has joined #openstack-meeting16:31
*** reldan has quit IRC16:38
*** reldan has joined #openstack-meeting16:54
*** jbryce has joined #openstack-meeting17:19
*** creiht has joined #openstack-meeting17:23
*** dabo has joined #openstack-meeting17:23
*** johnpur has quit IRC17:46
*** johnpur has joined #openstack-meeting17:54
*** jbryce_ has joined #openstack-meeting18:02
*** jbryce has quit IRC18:05
*** jbryce_ is now known as jbryce18:05
*** reldan has quit IRC18:31
*** romain_lenglet has joined #openstack-meeting18:41
*** troytoman is now known as troytoman-away18:46
*** old_devel has joined #openstack-meeting19:10
*** old_devel has left #openstack-meeting19:10
*** reldan has joined #openstack-meeting19:24
*** jaypipes has joined #openstack-meeting19:56
*** eday has joined #openstack-meeting20:00
*** jk0 has joined #openstack-meeting20:01
*** danwent has joined #openstack-meeting20:06
*** romain_lenglet has quit IRC20:09
*** romain_lenglet has joined #openstack-meeting20:09
*** dprince has quit IRC20:11
*** troytoman-away is now known as troytoman20:12
*** nixpert has joined #openstack-meeting20:15
*** romain_lenglet has quit IRC20:19
*** Ryan_Lane has joined #openstack-meeting20:28
*** old_devel has joined #openstack-meeting20:33
*** old_devel has left #openstack-meeting20:34
*** troytoman is now known as troytoman-away20:35
*** nixpert has quit IRC20:47
*** Vek has joined #openstack-meeting20:50
*** zul has quit IRC20:52
*** jlmjlm has joined #openstack-meeting20:53
*** masumotok has joined #openstack-meeting20:54
*** ewanmellor has joined #openstack-meeting20:57
*** chuck_ has joined #openstack-meeting20:57
sorenchmouel: It's spelled "/names" here :)20:58
*** justinsb has joined #openstack-meeting20:58
chmouelsoren: :)20:58
*** letterj has joined #openstack-meeting20:58
*** johan_ has joined #openstack-meeting20:59
*** jarrod has joined #openstack-meeting21:00
openstackMeeting started Tue Mar 22 21:00:08 2011 UTC.  The chair is soren. Information about MeetBot at http://wiki.debian.org/MeetBot.21:00
openstackUseful Commands: #action #agreed #help #info #idea #link #topic.21:00
sorenGreetings, everyone.21:00
sorenI'm filling in for ttx who's on the road.21:00
sorenPlease bear with me.21:00
sorenI'll try to type in a French accent.21:00
dendrobatessoren: I know ttx and you are a very poor ttx21:00
sorenToday's agenda:21:00
sorendendrobates: And vice versa, to be fair.21:01
soren#link http://wiki.openstack.org/Meetings21:01
*** sandywalsh has joined #openstack-meeting21:01
soren#topic Actions from previous meeting21:01
*** openstack changes topic to "Actions from previous meeting"21:01
soren* jaypipes to try to send review focus email before ttx sends the BMPFreeze focus email on Friday21:01
sorenttx sent one on Friday, listing the 18 Nova branches that should get review priority21:01
soren* soren to fix his review day reminder21:01
sorenI believe I did.21:02
sorendendrobates: Did you get an e-mail today, for instance?21:02
sorenAny missing action items from last week?21:02
soren#topic Current release stage: Development21:03
*** openstack changes topic to "Current release stage: Development"21:03
sorenWe are 2 days from FeatureFreeze, which will happen late on March 24.21:03
sorenFeature branches need to be reviewed, fixed and merged by then.21:03
soren#info Feature branches that are not merged in time should either be deferred to the next merge window (Diablo, opening April 14), or follow the FeatureFreeze exception process to try to sneak in Cactus.21:03
johnpursoren: when is the exact time on the 24th?21:03
sorenjohnpur: iz secret.21:04
johnpurfor those that cut it close...21:04
jaypipesjohnpur: 4:20pm EST.21:04
creihtEastern Secret?21:04
sorenjaypipes: Probably somewhere around there, yes.21:04
jaypipesthat was a joke...21:04
johnpurjaypipes: i thoughwe wuz all about the UTC here21:04
dendrobatessoren: I did.  I will be doing my reviews this evening21:04
sorendendrobates: Cool beans.21:04
ewanmellorhypervisor-vmware-vsphere-support is approved by Jay Pipes.  If Rick Harris or Ed Leafe or someone else from -core could review and approve, that would be awesome.21:04
westmaasthe BMP freeze was midnight Pacific21:04
westmaasand we used all that time :x21:05
sorenwestmaas: It happens when it happens. :)21:05
pvoewanmellor: are you looking for just code comments or integration test confirmation?21:06
sorenSee exception process at:21:06
soren#link http://wiki.openstack.org/FeatureFreeze21:06
sorenAnything else for this topic?21:06
westmaasalso see the deadline on the first line :)21:06
ewanmellorpvo: Citrix and Jay have reviewed, so I'm just looking for a green light to get it merged.21:06
*** adiantum has joined #openstack-meeting21:07
_cerberus_xs_resize could also use some love pretty plz21:07
sorenMoving on (apparently)21:07
soren#topic Cactus Release status21:07
*** openstack changes topic to "Cactus Release status"21:07
sorenPlease check:21:07
soren#link http://wiki.openstack.org/releasestatus/21:07
sorenReview effort should go in priority on the 18 planned feature branches mentioned in:21:07
*** nixpert has joined #openstack-meeting21:07
soren#link https://lists.launchpad.net/openstack/msg01377.html21:07
sorenSo far only 1 of those 18 have been merged.21:07
sorenIn particular those 3 (essential or prerequisites to other) need attention:21:08
sorenMan, cut-and-pasting stuff from ttx is easy. I could do this all day.21:08
ewanmellorsoren: Sorry, didn't realise "Development" was separate from "Release status" ;-)21:08
sandywalshhoping to take zones4 out of WIP tomorrow21:08
sorenewanmellor: Yeah, I should have just skipped through it faster :)21:08
*** User859 has joined #openstack-meeting21:09
tr3buchetsandywalsh: awesome!21:09
*** User859 has joined #openstack-meeting21:09
*** troytoman-away is now known as troytoman21:09
sorenOn the Nova stabilization effort:21:09
sorenLast week we had 35 bugs opened and 29 fixes committed21:09
sorenThis week we had 31 bugs opened and 29 fixes committed21:09
sorenThese numbers are current as of this morning. I think they're still accurate.21:10
sorenAnything else for this topic?21:10
*** jcookeman has joined #openstack-meeting21:11
sorenMoving on.21:11
soren#topic Open discussion21:11
*** openstack changes topic to "Open discussion"21:11
jcookemanHi folks. Im sorry Im late.21:11
letterjWhere do we stand on test environments?21:11
sorenExcellent question.21:12
pvopublic test environments?21:12
letterjAll types of them?21:12
sorenanotherjesse mentioned in an mp comment that he now had access to test hardware. How publically usable is that?21:12
* soren glances at other ex-Anso people for info21:12
*** nelson has joined #openstack-meeting21:12
pvoWe're almost ready to begin deploying our testing cluster but it isn't going to be publicly available.21:13
tr3buchetnot i expect21:13
pvojust need to get the packaging down21:13
johnpurpvo: what is the state of the RAX test huddle?21:13
pvohost machines provisioned.21:13
sorenpvo: Ok, so the test results won't even be visible outside Rackspace?21:13
pvoI talked with mtaylor about getting our debian squeeze packages finished.21:13
pvowe can make the results public, I think21:13
letterjpvo: Does that include a Swift cluster?21:13
pvono, this was just host machines for nova21:14
johnpurtest results should be public, no matter where the tests are conducted21:14
sorenpvo: What does that mean? They'll be hooked into Jenkins or will you publish anhtml page every two weeks?21:14
tr3bucheti wasn't aware anyone had plans for creation of a public test environment, did i miss something?21:14
johnpursoren: i am talking with mtaylor to get the testing and results automated21:14
pvosoren: I expect they'll be hooked into jenkins running the integration suite nightly.21:14
johnpurpvo: right on!21:15
sorenpvo: Ok, that's "public" in my book.21:15
pvoI plan to work with mtayor and whoever is doing the smoke tests21:15
* ttx waves from dinner joint wifi21:15
pvosoren: results public, but not the ability to hit the api21:15
*** old_devel has joined #openstack-meeting21:15
sorenpvo: Sure. Baby steps.21:16
pvosoren: : )21:16
letterjDon't we need to get swift included with that?21:16
jlmjlmSo, testing excl post-commit, to be possible pre-commit later on?21:16
pvoletterj: will circle with you after this21:16
letterjpvo: ok thanks21:16
johnpurletterj: yes21:17
pvojlmjlm: unsure yet.21:17
pvoif you mean for my test rig.21:17
pvoWe plan on doing nightly stress tests with +200 host machines21:17
johnpurwe should look at other test clusters, outside of Rackspace as well21:18
pvojust acquiring those machines has taken longer than expected21:18
pvojohnpur: ++21:18
sorenjohnpur: ++ indeed21:18
pvoI think this is what anso and anotherjesse were looking at.21:18
letterjjohnpur: I agree21:18
pvobut I could be wrong21:18
sorenOk, so *something* is finally happening. Great.21:19
justinsbIf we're all done on testing... We've had some potential issues come up with our XML support.  I'd like to import the CloudServers v1.0 XSD into lp:nova as a first step, and then I'm going to create a binding that makes use of it and see where we are.21:19
westmaaspvo: that was my impression as well21:19
sorenjustinsb: What are these potential issues?21:20
*** User859 has quit IRC21:21
justinsbsoren: We may be returning a string where we should be returning integers and vice versa21:21
*** User859 has joined #openstack-meeting21:21
justinsbsoren: And we may simply not be encoding things in the correct structure21:21
justinsbsoren: With the schema, we can easily check21:21
sorenjustinsb: So you'd use the XSD in the unit tests or something?21:21
justinsbsoren: We should be using the XSD in the unit tests, yes21:22
ttxjohnpur: re: exact time -- when I get up on the next day.21:22
justinsbsoren: But initially I will do it outside the unit tests21:22
sorenjustinsb: Makes sense. Are there any particular implications of this since you wanted to raise it here?21:22
justinsbsoren: Well, there's a potential copyright issue...21:22
johnpurttx: you are the gatekeeper, sounds good21:22
ttxjohnpur: which usually means Thursday, past 23:59 Hawaii time21:22
justinsbsoren: But other than that, just the timing wrt feature freeze.  And the general love of all things XML :-)21:23
johnpurttx: and you will listen to beggars who want just a few more minures?21:23
pvottx: so as long as it is thursday somewhere?21:23
ttxpvo: indeed21:23
ttxjohnpur: usually they are all alseep, but yes21:23
vishysoren: trying to get update from jesse on test hardware21:23
vishysoren: he's in another meeting21:23
sorenvishy: But of course. :)21:23
sorenA test rig that actually runs our reference platform would be rather nice.21:24
soren*cough* even.21:24
*** salvatore-orland has joined #openstack-meeting21:24
johnpuras i mentioned earlier, i am trying to recruit mtaylot to help pull togther all the test automation and deployments in some sort of coherent view21:24
pvosoren: what is your idea of the reference platform?21:25
*** diegoparrilla has joined #openstack-meeting21:25
sorenpvo: Ubuntu.21:25
pvowith what hypervisor?21:25
pvoand what api?21:25
pvoshouldn't that all be defined if it is a reference?21:25
sorenKVM, probably, since our reference platform actually supports that.21:25
ttxdid anyone mention that we have an election underway ? If you didn't receive a PPB voting link, you should probably complain to Stephen21:26
sorenThe API is hardly part of the reference platform.21:26
johnpursoren|pvo: we need to discuss this21:26
sorenttx: Noone mentioned this, no. I forgot.21:26
jaypipessoren: ++ on a test rig. been saying that for, well, quite some time now. ++ on integrating anso's Jenkins stuff into the main OpenStack Hudson platform too...21:26
pvono disagreement we need testing. getting enough hardware provisioned and available was a challenge.21:27
jaypipespvo: understood entirely. but I can still bitch. :P21:27
ttxjaypipes: during travel I hacked up the HTML version of my review list. will post soon. Loks nice. Prepare the kissing.21:27
sorenI got fed up waiting and bought three servers myself.21:27
jaypipesttx: will do. cheers mate.21:27
* kpepple_ also pulled a soren and bought his own set of servers21:27
johnpursoren: i like the dedication! talk to me about HW21:28
* jaypipes was also considering it..21:28
vishysoren: update from jesse: 6 servers, ipmi works, we are working on a basic jenkins vlan kvm setup21:28
ttxkpepple_: a "soren" is actually a hole-in-one on a par 521:28
vishysoren: we will need help doing a similar setup for xen etc.21:28
sorenttx: I don't love you anymore.21:28
pvojaypipes: for the record, we bought small pcs to do local dev/testing21:29
sorenvishy: 6 servers sounds like a fantastic start.21:29
* kpepple_ scribbles down "soren" definition for his upcoming Austin golf outing21:29
jaypipespvo: "we" is ozone?21:29
ttxsoren: nothing like private jokes.21:29
sorenttx: Nothing at all.21:29
pvojaypipes: we = ozone, sorry21:29
jaypipespvo: k. cool. not being a Rack employee (contractor), I'm not sure what the steps would be to do the same... I'll chat with johnpur offline about it.21:30
vishysoren: we will also document and publish config and setup so it is easy for people to define new reference archs and tests21:30
vishythe initial plan is just to fire it off post trunk merge21:30
sorenvishy: Yeah. It would be great if there was a straight forward way people could volunteer hardware for testing and magic would just happen.21:31
pvovishy: westmaas and tr3buchet are working on tests as well. are you all cordinating?21:31
westmaaswe are not yet21:31
vishyand have perhaps a separate ppa trunk_tested21:31
pvocoordinating, even21:31
vishythat only includes revs that pass all the tests21:31
jaypipesvishy: I've been monitoring the Anso Jenkins setup for a few weeks now... haven't seen the main vagrant_smoketests job complete successfully for a couple weeks.... you guys on that?21:31
westmaasI keep contacting names as I hear them, vishy I will work with you to make sure we aren't deuplicating21:31
sorenI'd be happy to talk more about testing, but perhaps we should call it a meeting?21:32
tr3bucheta few of us coordinated on the phone yesterday, but not all21:32
vishyjaypipes: yes we managed to break everything right before I went on vacation21:32
soren...and let people who care slightly less about this go?21:32
jaypipesvishy: ah, cool. I'm good at that, too. :)21:32
vishyjaypipes: it is currently being fixed21:32
pvosoren: soon as this is finished?21:32
sorenpvo: Sure.21:32
sorenDoes anyone else have anything to talk about?21:32
User859Please review : https://code.launchpad.net/~ntt-pf-lab/nova/flatmanager-ipv6/+merge/53064 and https://code.launchpad.net/~tpatil/nova/bug704300/+merge/5333321:33
ttx+& ^21:33
ttx+1 even21:33
* soren blinks21:33
*** openstack changes topic to "Openstack Meetings: http://wiki.openstack.org/Meetings | Minutes: http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/"21:33
openstackMeeting ended Tue Mar 22 21:33:31 2011 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)21:33
openstackMinutes:        http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-03-22-21.00.html21:33
openstackMinutes (text): http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-03-22-21.00.txt21:33
openstackLog:            http://eavesdrop.openstack.org/meetings/openstack-meeting/2011/openstack-meeting.2011-03-22-21.00.log.html21:33
*** nelson has left #openstack-meeting21:33
*** johan_ has left #openstack-meeting21:33
*** Vek has left #openstack-meeting21:33
* soren thanks ttx for the pre-baked script21:33
*** creiht has left #openstack-meeting21:33
*** User859 has quit IRC21:33
sorenThat was easy enough :)21:33
ttxsoren: np21:34
*** User859 has joined #openstack-meeting21:34
*** User859 has quit IRC21:34
sorenHow's London?21:34
sorenOr did you make it to LEeds?21:34
sorenI forget.21:34
*** jarrod has left #openstack-meeting21:34
johnpursoren: nice job21:34
sorenjohnpur: It was all ttx. I just copied and pasted :)21:34
sorenjohnpur: Well, most of it :)21:34
jcookemanVery nice weather in Leeds right now.21:34
johnpurand ttx couldn't stay away :)21:34
sorenpvo: here or in #openstack?21:35
* johnpur pvo is eay21:35
sorenIt would be great if mtaylor was here.21:35
*** Ryan_Lane has left #openstack-meeting21:35
*** letterj has left #openstack-meeting21:36
westmaasvishy: are you guys working on actual tests or just on making testing easier?21:36
vishywestmaas: we haven't added new tests beyond smoketests21:36
johnpursoren: mtaylor is inches away from joining our team21:36
sorenjohnpur: Awesome!21:36
sorenpvo: What are your tests based on?21:37
pvoso it feels we have at least 4 different pushes for integration21:37
pvo*today* we have cloudservers api tests21:37
pvosome OSapi tests for nova21:37
ttxsoren: Leeds.21:37
pvothat westmaas has been coordinating21:37
pvoand tr3buchet has a framework that we were going to adapt21:38
pvoto handle some of the scale testing21:38
sorenWhere is the code for this?21:38
pvowe had a talk yesterday about where to put it21:38
westmaassoren: its in a branch now21:38
pvodoes it make sense to put into nova trunk?21:38
pvoso everyone can run the same tests or keep it separate?21:38
westmaaswhich it?21:38
westmaasthe set of tests?21:38
tr3buchetperfect place i think21:38
sorenpvo: Yes.21:39
pvowhichever suite we think21:39
westmaaspvo: I think so.21:39
tr3buchetframework + set of tests for all to use21:39
sorenpvo: Right next to the other smoketests.21:39
pvook, good.21:39
pvoso, the next question I'd ask is who is leading the testing work?21:39
sorenPreferably runnable in an identical fashion21:39
pvowe have little pockets of sporadic work21:39
pvosoren: well, there is the rub.21:40
westmaasI have been coordinating the work with qa and sounds like some ozone people have been added21:40
sorenI've been doing a bunch of testing and hardly any feature work this cycle, but I'm hardware challenged.21:40
pvoI'm not sure how we want to handle tests for all hypervisors and apis.21:40
pvowe obviously have xenserver hypervisors provisoned so that would be our focus, but I understand the need to test the rest of the code.21:41
*** danwent has quit IRC21:41
pvosoren: do you see this as a jenkins config that knows how to call out to the different rigs to test each stack?21:41
sorenWe could just get this stuff running and try to generalise it as we move forward.21:41
pvosoren: I would agree, but your tests are going to be ec2 and kvm?21:42
pvowestmaas's are going to be osapi and xenserver21:42
vishywe were hoping to generalize the testing config so that it could blow away and reprovision21:42
sorenpvo: They're EC2 now because that was all that was even remotely stable when I started this effort.21:42
sorenpvo: ...and it's all the smoketests tests.21:42
pvosure, no blame21:42
sorenpvo: I'd love to test the OS api, too.21:42
vishyso it could install ubuntu/kvm, run tests21:43
sorenpvo: I just don't. Yet.21:43
pvovishy: we wipe the machines down between tests, but don't reprovision21:43
pvosoren: right on21:43
jlmjlmCan we configure an instance to support multiple APIs, and test by trying them both out?21:43
vishythen install redhat/xen, run tests21:43
vishyboth apis automatically run, so we should be able to run both sets of tests against both21:43
vishythen we'll at least know what doesn't work against different configs21:43
sorenI don't reprovision the boxes, I just remove everything pertaining to nova and install it again.21:44
pvoI'm looking for a volunteer to handle the work coordination21:44
*** ewanmellor has quit IRC21:44
pvoI guess it isn't ewan21:44
vishysoren: I think we have to reprovision ultimately21:44
sorenI think it should be someone at Rackspace. The great wall around you guys makes it impossible for me to do any of this.21:44
tr3bucheti agree vishy21:44
pvoyou are one of us21:44
pvo: )21:44
vishysoren: no vpn?21:45
sorenvishy: I can't see hardware through VPN.21:45
pvosoren: none of us can *see* the hw21:45
pvoits all in other DCs... not sure what you mean by that.21:45
sorenpvo: I just don't feel like guessing what the right question to ask are when you guys are sitting around with all the answers at hand.21:46
tr3buchetsoren. i'm "inside" how may i be of service?21:46
sorenA volunteer appears!21:46
pvosoren: I don't think I have any answers at hand... I just go ask a bunch of questions21:46
westmaashaha, if tr3buchet didn't actually mean to volunteer, I'm happy to coordinate as well21:47
tr3bucheti just want to help21:47
sorenpvo: You don't honestly believe that I have as much easy access to information as you do?21:47
tr3buchetthis is necessary21:47
tr3buchetthe sooner the better21:47
uvirtbotwestmaas: Error: "^" is not a valid command.21:47
*** cynb has joined #openstack-meeting21:47
tr3buchetif i'm capable i'll do whatever is necessary (giddyup!)21:48
westmaaspvo: what exactly do you want someone to lead?  the creation of a set of tests as well as integration into jenkins?21:48
pvowestmaas: what I think we need is someone to determine what tests get run, how the framework looks and setup and configuration of the jenkins instances and workers21:49
sorenvishy: Reprovisioning from scratch would be "ideal". I just don't think it adds much real value compared to what I do.21:49
*** troytoman is now known as troytoman-away21:50
sorenvishy: I use packages. Exclusively. The uninstall mechanics are *very* well tested.21:50
vishysoren: well if we need to install xen on the boxes, we need to go from scratch21:50
sorenvishy: True.21:50
*** reldan has quit IRC21:50
westmaasalright, I am absolutely happy to coordinate that.  sounds like tr3buchet wants to help out in any way as well.21:51
tr3buchetand provisioning should be just as automated as tests21:51
sorenOk, so we have an n-dimensional test matrix.21:51
sorenOne dimension is hypervisor.21:52
vishyyes, and it should all be documented, so someone can replicate the setup21:52
sorenAnother is API.21:52
justinsbvishy: If you're looking at installing from scratch, you guys might want to check out the notsovm stuff I hacked up yesterday.  Beginnings of a bare-metal cloud.21:52
sorenAnotther is base OS.21:52
vishyanother is network model21:52
sorenAnother is network model.21:52
sorenHeheh :)21:52
* vishy is working on his psychic powers21:52
jlmjlmia32 vs amd64, eg21:53
westmaasprobably x86 vs ...21:53
sorenUser database backend.21:53
sorenVolume backend.21:53
sorenThis is complicated.21:53
vishythis is why we need 2 or 3 recommended configs21:53
sorenI agree.21:53
westmaasright, and we can expand as needed21:54
tr3buchetand then if you want to use another, test it yourself?21:54
westmaasonce we have those down21:54
vishywe can test those, then if people want to add a new config, we have instructions for creating new configs and how to create a test cluster21:54
sorenSupporting a bunch of different stuff is cool, but when people ask for recommendations, I want to be able to give a clear answer.21:54
tr3buchetyes i agree21:54
tr3bucheti worry we can still break things for configs which aren't a part of the automated tests21:54
*** jk0 has left #openstack-meeting21:55
westmaastr3buchet: what do you mean exactly?21:55
tr3buchetwell say we have a recommended config that we test, on specific network model/os/db etc.21:55
_cerberus_That's it's possible for everything to pass on our "blessed" options and still blow up on the unblessed ones21:55
westmaasyeah, I think I got it.21:55
tr3buchetif that's the gateway which code to trunk has to pass through21:55
westmaasdefinitely possible, but this will be a ton better21:56
tr3buchetthen that code could be breaking other untested configs21:56
soren_cerberus_: That's why I want it to be easy for people to volunteer test environments to this.21:56
tr3bucheti agree it will be much better21:56
_cerberus_soren: completely agree21:56
tr3buchetyep that's perfect21:56
soren_cerberus_: They get to choose which config they think is cool, it gets automatically tested.21:56
tr3buchetif no one is willing to test a certain config we can't guarantee it works21:56
_cerberus_Sounds brilliant21:57
jlmjlmYeah, that's always possible.  But unless we want to test all 64 cases that come out of our matrix, what's to be done?21:57
sorenIf someone *actually* cares about a particular config, giving up a single host to test it against should be possible.21:57
_cerberus_Well, I think to sorens point, we still need the "blessed" versions. No matter what, we'll ensure that those work. You can run yours through the configurator and let us know, but we're not going to check for you21:57
soren_cerberus_: Right.21:57
tr3bucheti like it21:57
justinsbBlessed configurations (if carefully chosen) still achieve good coverage of configurations21:58
ttxremember this is not a discussion channel... you should continue this on #openstack for everyone to see :)21:59
sorenI see what you're saying, but I somehow doubt blessed configurations will be chosen like that.21:59
*** jcookeman has left #openstack-meeting22:00
sorenBut there's certainly some wisdom to be taken from an approach like that.22:00
justinsbsoren: Well, I think it's possible to choose the first few arbitrarily.  Then we can pick a few more just to achieve pairwise coverage22:00
sorenjustinsb: I just don't think we'll bless some combination of stuff to achieve good coverage. We'll choose them, probably, because they're already in good shape, and there's likely going to be a lot of duplication.22:02
*** _0x44 has joined #openstack-meeting22:02
sorenI don't, for instance, think we'll bless any configuration that uses AoE.22:02
justinsbI think that's the right place to start22:02
sorenWell, not "bless" as in "recommend" at least.22:02
justinsbThe point is we do have a solution to the combinatorial explosion22:02
westmaasok, so me and tr3buchet will work together and try to move forward on this?22:04
sorenFor at start, though, I think choosing a few blessed combos is the way forward.22:04
westmaasbut obviously getting buy-in from the whole community22:04
sorenRAckspace has one already: OpenStack API, XenServer, Debian.22:05
tr3bucheti think we should definitely choose the openstack API22:05
sorenAnother popular one is EC2 API, KVM, Ubuntu.22:05
*** cynb has left #openstack-meeting22:06
tr3buchetshould we not recommend/bless the openstack api?22:06
sorenBut I believe EC2 support is important. I'm going to keep testing it.22:06
soren..and as such, it'll continue to be in good shape.22:06
tr3buchetso we'll recommend them equally22:07
jlmjlmSeeing as an instance supports multiple APIs, I'm not sure they should be considered part of the configuration22:07
tr3buchetso there are two combos right there22:07
tr3buchetlet's start with thsoe22:07
jlmjlmso much as different tests of that configuration.22:07
*** eday has left #openstack-meeting22:07
sorenWhat I'm saying is: Yes, we shoudl recommend the OpenStack API, but in a way that doesn't suggest that the EC2 API is going away or is unmaintained.22:07
sorentr3buchet: Yes.22:08
justinsbI'd lobby for the OpenStack API to be blessed with the more open platform22:08
westmaastr3buchet: agreed22:08
justinsbi.e. KVM22:08
tr3buchetanyone against having a 3rd combo?22:08
sorenMy experience so far is that just doing one single config on one single box has unearthed enough bugs to keep me busy for months.22:09
westmaasalso, jlmjlm makes a good point.  as far as testing the two APIs its really just that there are no tests for the OS API22:09
justinsbThere are the beginnings of tests for the OS API, but they haven't been merged :-(22:10
westmaasthose tests will be merged as soon as we have a minimum reasonable merge22:10
sorenLots of combos is fine, but I think we'll be plenty busy dealing with our findings with veyr simple configs/very few combos.22:10
justinsbI think we should actually start with one combo22:10
justinsbWhatever we can get up and running today22:10
justinsbSo if Xen means reimaging, then no Xen22:11
justinsbIf the OS API isn't complete yet, then no OS API22:11
tr3buchetmakes sense22:11
tr3buchetso to start, sounds like it'll be ec2 kvm ubuntu22:11
tr3buchetas the os api tests get more complete22:12
tr3buchetwe can add another22:12
justinsbIdeally we would document the process of building a test box so that everyone can do it22:12
justinsbAnd we have a head start for config #222:12
tr3buchetjustinsb: i don't think that's ideal i thin it's a requirement22:12
sorenIn my setup, Jenkins sets everything up. It just needs a box with Ubuntu on it and an ssh user with sudo privs.22:12
justinsbsoren: Nice.  I've got one of those sitting idle that I can reimage at will...22:13
sorenThen it logs in and does magic.22:13
justinsbsoren: Is the script in your repo somewhere?22:13
sorenlp:~linux2go/nova/jenkins-config IIRC22:13
sorenThe install.?.sh ones.22:13
*** johnpur has left #openstack-meeting22:13
*** chuck_ has quit IRC22:25
*** jlmjlm has left #openstack-meeting22:46
*** westmaas1 has joined #openstack-meeting22:49
*** old_devel has left #openstack-meeting22:52
*** nixpert has quit IRC22:55
*** chuck_ has joined #openstack-meeting22:58
*** pvo has quit IRC23:03
*** diegoparrilla has left #openstack-meeting23:04
*** chuck_ has quit IRC23:08
*** dendrobates is now known as dendro-afk23:14
*** dragondm has quit IRC23:22
*** danwent has joined #openstack-meeting23:29
*** chuck_ has joined #openstack-meeting23:40
*** chuck_ has quit IRC23:51

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