Wednesday, 2010-10-20

masumotokHello, I appreciate if somebody give me an reply.00:07
masumotokIn today's meetings, I saw "go/not go meeting will be held on Thursday, same time".00:07
masumotokThat means "Austin Release may achieved or postponed, that is meeting aganda". correct?00:07
*** silassewell has joined #openstack00:09
*** cloudmeat1 has joined #openstack00:12
*** AimanA is now known as HouseAway00:14
grizzletoothmasumotok: I believe you are correct.00:14
*** cloudmeat has quit IRC00:14
masumotokgrizzletooth: Thanks!00:15
*** grizzletooth has quit IRC00:35
*** grizzletooth has joined #openstack00:35
*** grizzletooth has joined #openstack00:36
*** sanjib has quit IRC00:38
*** roamin9 has joined #openstack00:39
mtaylordendrobates: pong00:39
*** maplebed has quit IRC00:45
*** gundlach has quit IRC00:45
*** niks has quit IRC00:46
*** jc_smith has quit IRC00:55
*** daleolds has quit IRC00:55
*** cloudmeat1 has quit IRC00:55
*** gustavomzw has quit IRC01:08
notmynamethis could be an interesting problem in the days to come
burrisnotmyname, it'll be good practice for the impending switchover to perl 601:19
notmynameburris: right after we're all running on ipv6?01:19
burrisipv6 might actually happen01:19
notmynameya, I keep hearing rumors. more and more dire all the time ;-)01:20
*** wreese_ has joined #openstack01:33
*** creiht_ has joined #openstack01:34
*** schisamo has quit IRC01:34
*** rbergero1 has joined #openstack01:34
*** rbergero1 is now known as rbergero01:35
*** glange_ has joined #openstack01:38
*** _FeliXdk has joined #openstack01:38
*** kevnfx has joined #openstack01:38
*** Kami_ has quit IRC01:38
*** wreese has quit IRC01:38
*** creiht has quit IRC01:38
*** soren has quit IRC01:38
*** thom has quit IRC01:38
*** ioso has quit IRC01:38
*** glange has quit IRC01:38
*** rbergeron has quit IRC01:38
*** FeliXdk has quit IRC01:38
*** thom_ has joined #openstack01:38
*** soren_ has joined #openstack01:38
*** ioso has joined #openstack01:38
*** Kdecherf has quit IRC01:38
*** sandywalsh has quit IRC01:38
*** ar1 has quit IRC01:38
*** localhost has quit IRC01:38
*** odyi has quit IRC01:38
*** anotherjesse has quit IRC01:38
*** devcamcar has quit IRC01:38
*** jtimberman has quit IRC01:38
*** ttx has quit IRC01:38
*** dendrobates has quit IRC01:38
*** btorch_ has quit IRC01:38
*** pandemicsyn has quit IRC01:38
*** kirkland has quit IRC01:38
*** jaypipes has quit IRC01:38
*** f4m8_ has quit IRC01:38
*** mfoemmel has quit IRC01:38
*** Kami_ has joined #openstack01:38
*** _FeliXdk is now known as FeliXdk01:38
*** rbergero is now known as rbergeron01:39
*** rbergeron has joined #openstack01:39
*** Kami_ is now known as Guest7518501:39
*** ArdRigh has quit IRC01:39
*** abecc has quit IRC01:41
*** Kdecherf has joined #openstack01:41
*** sandywalsh has joined #openstack01:41
*** ar1 has joined #openstack01:41
*** odyi has joined #openstack01:41
*** anotherjesse has joined #openstack01:41
*** devcamcar has joined #openstack01:41
*** jtimberman has joined #openstack01:41
*** ttx has joined #openstack01:41
*** dendrobates has joined #openstack01:41
*** btorch_ has joined #openstack01:41
*** pandemicsyn has joined #openstack01:41
*** kirkland has joined #openstack01:41
*** jaypipes has joined #openstack01:41
*** f4m8_ has joined #openstack01:41
*** mfoemmel has joined #openstack01:41
*** sets mode: +vv dendrobates pandemicsyn01:41
*** miclorb_ has quit IRC01:43
*** huz has joined #openstack01:43
*** localhost has joined #openstack01:45
*** miclorb has joined #openstack01:50
*** pvo has joined #openstack01:51
*** pvo has quit IRC01:51
*** pvo has joined #openstack01:51
*** ChanServ sets mode: +v pvo01:51
*** iammartian has joined #openstack01:53
*** pvo has quit IRC01:54
*** ArdRigh has joined #openstack01:55
*** daleolds has joined #openstack01:56
dendrobatesmtaylor: we have three merge requests that have been sitting for between 2-6 hours02:05
dendrobatesmtaylor: can you see why they are not being merged?02:05
dendrobatesmtaylor: they are holding up me spinning the RC02:06
*** Cybodog has quit IRC02:09
*** Cybodog has joined #openstack02:09
dendrobatesGaaaa!!!!!!!111  Hudson!02:17
*** BK_man has quit IRC02:21
*** herki has quit IRC02:23
*** daleolds has quit IRC02:28
*** soren_ has quit IRC02:33
*** soren_ has joined #openstack02:33
*** ChanServ sets mode: +v soren_02:33
*** soren_ is now known as soren02:33
* soren pats dendrobates on the head02:34
sorendendrobates: Hudson troubles?02:34
sorenErk. Too much red.02:35
sorenOk, I think I know what that's all about..02:35
* soren looks02:35
sorenYup, found it.02:37
sorendendrobates: It's all your fault, actually :)02:37
sorendendrobates: was missing both a commit message and a description.02:38
sorendendrobates: I added a commit message (copied from the commit in the proposed branch). It should be smooth sailing from then on.02:38
sorendendrobates: I'll see what's up with the other two once this one lands.02:39
*** dabo has quit IRC02:41
sorenOne of these days, I'm going to have to try to find out why it takes several mintes to go from "Completed test command: /var/lib/hudson/" to "Firing tarmac_post_commit hook".02:42
dendrobateswhat, no it wasn;t.02:46
dendrobatessomething screwed up during the push then02:46
sorenIf it was ever there, someone/something removed it.02:48
sorenI added it, it got merged. Vish's branch just landed, too.02:48
sorenNext up is Jay's.02:48
dendrobatesit would be nice if we had some error to look at when something happened02:49
sorenWe do.02:49
sorenGo to the failed build, look at the console output.02:49
dendrobatesI looked all over.   It's on the same box as the wiki right?02:49
*** joearnold has joined #openstack02:50
sorenI don't believe it is, no.02:50
sorenThat one is one of the ones that failed due to missing commit message on the merge proposal.02:50
sorenNope, it's not the same box.02:51
dendrobatesoh, a web console.  I was poking around the serve.02:51
sorenIt seems only Monty and I have logins on it.02:51
dendrobatesthere is a hudson instance running on it, but it seems to just be for the planet02:51
sorenYes, it's remote-controlled by the "real" hudson server.02:52
sorenJay's branch merged.02:52
sorenSo yay, we have a succesful build again.02:52
dendrobatesnow will the ppa build?02:53
sorenThat was unrelated, but yes.02:53
dendrobateswhat was that?02:53
sorenWell, I know why it failed *today*.02:53
sorenI hav eno idea why it seems to have been failing since the 6th.02:53
sorenLet me just check if that's really the case.02:54
sorenIt is.02:57
sorenNo idea why..02:57
*** silassewell has quit IRC03:03
mtaylordendrobates, soren hello03:07
*** pvo has joined #openstack03:09
*** pvo has quit IRC03:09
*** pvo has joined #openstack03:09
*** ChanServ sets mode: +v pvo03:09
mtaylorsoren: you seem to have found the fun with missing commit message and description03:09
sorenmtaylor: You have a messed up sense of humour :)03:10
mtaylorsoren: yes. this is definitely true03:12
*** pharkmillups has joined #openstack03:17
*** Robi__ is now known as Robi_03:20
*** joearnold has quit IRC03:36
*** pharkmillups has quit IRC03:38
*** pharkmillups has joined #openstack03:38
sorenIf you would do the honours, we can roll  atarball.03:38
dendrobatesI'm on it03:39
sorenHm... Hang on :)03:40
*** pjz_ is now known as pjz03:40
dendrobatesthe diff looks fine03:40
sorenIt's missing a little ibit.03:41
soren 2 seconds.03:41
dendrobateswhat is it missing?03:41
* soren hates hotel wifi03:41
sorenversion in docs03:41
dendrobatesIt's 2 lines I pushed to approved03:43
*** ambo has joined #openstack03:43
*** pvo has quit IRC03:44
dendrobateslet hudson do it's magic and we are ready to go.03:44
sorenThat's good, 'cause I'm about to cross my usefulness threshold.03:45
* soren goes to pass out.03:45
dendrobatesI am about past mine.  get some sleep03:45
sorenI promise.03:45
sorenYou too!03:45
sorenhappy RC.03:45
*** kashyapc has joined #openstack03:46
*** pharkmillups has quit IRC03:48
*** ambo has quit IRC03:49
*** Ryan_Lane has joined #openstack03:57
*** ambo has joined #openstack03:57
dendrobateswell they rc tarballs are up.  now I need to go pack.  I'll send the email tomorrow morning.04:02
*** sirp has quit IRC04:04
*** sirp has joined #openstack04:05
*** Ryan_Lane has quit IRC04:10
*** ambo has left #openstack04:12
*** ArdRigh has quit IRC04:25
*** joearnold has joined #openstack04:31
*** joearnold has joined #openstack04:35
*** Ryan_Lane has joined #openstack04:44
*** Ryan_Lane has quit IRC04:49
*** Ryan_Lane has joined #openstack04:54
*** johnbergoon has joined #openstack05:12
*** jtimberman has quit IRC05:13
*** jc_smith has joined #openstack05:18
*** kevnfx has quit IRC05:24
*** jtimberman has joined #openstack05:27
*** Ryan_Lane is now known as Ryan_Lane|away05:50
*** joearnold has quit IRC05:54
*** guigui has joined #openstack06:01
*** ArdRigh has joined #openstack06:01
*** sirp has quit IRC06:14
*** sirp has joined #openstack06:15
*** perestrelka has joined #openstack06:30
*** allsystemsarego has joined #openstack06:31
*** BK_man has joined #openstack06:31
*** roamin9 has quit IRC06:38
*** roamin9 has joined #openstack06:45
*** alekibango has quit IRC06:47
*** arthurc has joined #openstack06:48
*** befreax has joined #openstack06:54
*** befreax has left #openstack07:01
*** jc_smith has quit IRC07:02
*** omidhdl has joined #openstack07:08
*** roamin9 has quit IRC07:13
*** roamin9 has joined #openstack07:13
*** gaveen has joined #openstack07:15
*** gaveen has joined #openstack07:15
*** calavera has joined #openstack07:16
*** ibarrera has joined #openstack07:18
*** almaisan-away is now known as al-maisan07:20
*** alekibango has joined #openstack07:34
*** miclorb has quit IRC07:39
*** thom_ is now known as thom07:49
*** iammartian has quit IRC07:56
zykes-where's the rc tarball ?08:00
*** irahgel has joined #openstack08:34
*** krish has joined #openstack08:52
*** befreax has joined #openstack09:10
*** krish has quit IRC09:33
*** ArdRigh has quit IRC09:33
*** huz has quit IRC09:34
*** smithpg1002 has joined #openstack09:39
*** Guest75185 is now known as Kami_09:43
*** jonwood has joined #openstack09:53
*** jakedahn has joined #openstack10:07
*** roamin9 has quit IRC10:09
*** jakedahn has joined #openstack10:20
*** omidhdl has quit IRC10:36
*** kashyapc has quit IRC10:53
*** omidhdl has joined #openstack11:06
dendrobateszykes-: there is a download link there11:13
*** gustavomzw has joined #openstack11:24
*** cloudmeat has joined #openstack11:26
zykes-and that works on maverick ?11:26
*** dendrobates is now known as dendro-afk11:32
*** Dweezahr has quit IRC11:44
*** Dweezahr has joined #openstack11:44
*** perestrelka has quit IRC11:51
*** perestrelka has joined #openstack11:51
*** littleidea has joined #openstack11:55
*** littleidea has quit IRC12:00
zykes-what's the difference on and ?12:03
*** befreax has quit IRC12:11
*** westmaas1 has joined #openstack12:18
*** befreax has joined #openstack12:19
*** silassewell has joined #openstack12:21
*** kakoni has joined #openstack12:23
kakoniHi guys. So promised to come back with swift-bench results when running more tuned cluster.12:24
*** RexMundi has quit IRC12:28
*** dizz|away has quit IRC12:30
kakoniSo test setup, ubuntu 10.10 on 1 proxy machine, 6 storage nodes. average perf about 350 writes / sec, and read somewhere in the 1000 range.12:32
zykes-why does ony want to get rid of redis ?12:37
zykes-in Nova12:38
*** kashyapc has joined #openstack12:43
*** ctennis has quit IRC12:47
*** kakoni has quit IRC12:48
*** pvo has joined #openstack12:51
*** ChanServ sets mode: +v pvo12:51
*** befreax has quit IRC12:52
*** littleidea has joined #openstack12:54
*** befreax has joined #openstack12:58
*** ctennis has joined #openstack13:02
*** ctennis has joined #openstack13:02
*** littleidea has quit IRC13:03
*** ddumitriu has joined #openstack13:07
*** kashyapc has quit IRC13:12
*** ddumitriu has quit IRC13:12
*** kashyapc has joined #openstack13:15
*** matclayton has joined #openstack13:16
*** gaveen has quit IRC13:19
zykes-hmm, when could there be in place a framework for writing a new driver ?13:22
*** littleidea has joined #openstack13:24
*** littleidea has quit IRC13:29
*** ppetraki has joined #openstack13:30
*** gundlach has joined #openstack13:31
*** gaveen has joined #openstack13:32
*** jkakar has quit IRC13:35
*** jkakar has joined #openstack13:36
*** omidhdl has left #openstack13:47
*** befreax has quit IRC13:51
*** abecc has joined #openstack13:57
*** abecc has joined #openstack13:57
*** Podilarius has joined #openstack14:06
*** rnirmal has joined #openstack14:08
*** BK_man has quit IRC14:09
*** schisamo has joined #openstack14:11
gundlachjaypipes: ping14:15
*** herki has joined #openstack14:19
jaypipesgundlach: pong14:21
gundlachhowdy, have you made the new test-overhaul branch?  i don't want to duplicate your work but would love to convert some tests to nosetests.14:22
jaypipesgundlach: can you give me a few more hours?  I've been working on it..14:23
*** rnirmal has quit IRC14:23
*** kevnfx has joined #openstack14:23
gundlachsure.  could you tell me the name of a unittest that you haven't started converting, so i can take care of that and then cherry-pick it into your branch later?14:23
jaypipesgundlach: sure thing :) take any of them below network_unittest (in the import list in
jaypipesgundlach: I've been working on a method to reset the test environment using fixtures and haven't converted any more tests14:24
gundlachjaypipes: what do you need to reset beyond the setUp and tearDown methods?  (and are you aware of module-level setUp and tearDown support in nose?)14:25
jaypipesgundlach: yes... here is what I'm working on:
*** pvo has quit IRC14:28
jaypipesgundlach: not tested at all...a work in progress, but I think you can see where I'm going with it.14:28
gundlachjaypipes: ah, i see.14:29
*** kevnfx has quit IRC14:29
*** rnirmal has joined #openstack14:30
*** littleidea has joined #openstack14:34
creiht_interesting.... eventlet is getting a zeromq hub14:36
*** silassewell has quit IRC14:37
ttxinteresting indeed14:40
*** creiht_ is now known as creiht14:41
*** ChanServ sets mode: +v creiht14:41
*** pharkmillups has joined #openstack14:45
*** Ryan_Lane|away is now known as Ryan_Lane14:49
*** pvo has joined #openstack14:51
*** pvo has quit IRC14:51
*** pvo has joined #openstack14:51
*** ChanServ sets mode: +v pvo14:51
*** ar1 has quit IRC14:51
*** gondoi has joined #openstack14:56
*** piken has joined #openstack14:58
pikenCan nova be setup for redundancy on the cloud controller?14:59
jonwoodpiken: I don't think there's any reason you couldn't just set up two cloud controllers. They share a database and message queue, so I don't think they'd break each other.15:00
jonwoodThen either put them behind a shared IP with failover, or point some of your nodes at one, and some at the other.15:01
pikenalso has anyone here actively used libcloud and boto with nova?15:09
*** HouseAway is now known as AimanA15:09
*** f4m8_ has quit IRC15:16
*** jdarcy has joined #openstack15:16
*** dendro-afk is now known as dendrobates15:16
*** kakoni has joined #openstack15:17
dendrobateshello folks15:18
dendrobatessitting in an airport trying to get the RC email out to the team.15:19
dendrobatesthings were delayed last night when I broke the merge env and couldn't fix it.15:20
dendrobatesanyone out there care to modify the nova install docs for the RC?15:22
*** f4m8_ has joined #openstack15:25
*** silassewell has joined #openstack15:25
*** jakedahn has quit IRC15:26
annegentledendrobates: sure, was planning to update today15:30
*** calavera has quit IRC15:35
creihtmtaylor: did the swift debian stuff get updated?15:38
creihtlooks like they did15:38
*** littleidea has quit IRC15:39
*** cloudmeat1 has joined #openstack15:44
*** cloudmeat has quit IRC15:45
*** metcalfc has joined #openstack15:46
dendrobatesannegentle:  as soon as we can.  basically we need instruction on how to install from a tarball.15:52
annegentledendrobates: ayup15:52
dendrobatesannegentle: are you in the office?15:53
annegentledendrobates: yeah, sure am15:53
dendrobatesannegentle: see you in a bit, then.  We can discuss it in the RW15:53
*** dizz has joined #openstack15:54
annegentledendrobates: sounds good15:54
dendrobatesannegentle: though I don;t know if I remember how to talk to people face to face.15:54
*** schisamo has quit IRC15:55
annegentledendrobates: bwah ha15:55
*** khussein has joined #openstack15:56
notmynamedendrobates: just saw your email. curious about what you mean by installing it on maverick rather than lucid. I know we have packages for maverick, but we run on (and dev for) lucid15:58
*** khussein_ has joined #openstack15:59
zykes-dendrobates: can't one just use the packages on lucid ?16:02
zykes-instead of tarball16:02
*** khussein has quit IRC16:03
*** khussein_ is now known as khussein16:03
*** kevnfx has joined #openstack16:03
letterjdendrobates:  In your email about the release candidate packages you say: "I will be providing links to an Ubuntu package archive later today, if16:03
letterjyou want to install onto Ubuntu Maverick."16:03
letterjdendrobates: Will different packages be needed for Lucid16:04
creihtI see a recurring theme here :)16:04
*** matclayton has quit IRC16:05
*** perestrelka has quit IRC16:07
*** guigui has quit IRC16:07
*** perestrelka has joined #openstack16:07
*** perestrelka has joined #openstack16:08
*** dendrobates is now known as dendro-afk16:11
*** daleolds has joined #openstack16:13
*** kevnfx has quit IRC16:18
*** dizz has quit IRC16:21
*** kevnfx has joined #openstack16:22
mtaylorcreiht, letterj we should be uploading source packages to the ppa for both lucid and maverick16:25
mtaylorcreiht: oh - yeah!16:26
mtaylorcreiht: the debian stuff should be up to date and good to go16:26
creihtmtaylor: yeah it looks good so far, and thanks!16:26
*** kakoni has quit IRC16:27
mtaylorcreiht: w00t16:27
*** johnpur has joined #openstack16:28
*** ChanServ sets mode: +v johnpur16:28
*** kevnfx has quit IRC16:30
*** maplebed has joined #openstack16:33
*** joearnold has joined #openstack16:35
*** rnirmal has quit IRC16:35
*** rnirmal has joined #openstack16:38
*** dendro-afk is now known as dendrobates16:45
*** khussein has quit IRC16:46
jaypipeshehe, Ewan Mellor now has a new (and cooler) last name.
jaypipesannegentle: a few things to fix up on the Nova docs...see notes in code review.  I don't blame you for not knowing these things...they change so darn often :)16:49
*** silassewell has quit IRC16:51
*** DubLo7 has joined #openstack16:53
*** krish has joined #openstack16:59
jaypipeseday: what's up with this blbueprint?
jaypipeseday: looks like it can be deleted?17:04
edayjaypipes: this was one of the blueprints I converted over from an etherpad, I think Vish made it originally17:05
edaythis was months ago17:05
jaypipeseday: delete it? :)17:05
jaypipeseday: want to give gundlach's a once over?17:06
*** daleolds has quit IRC17:08
*** mdomsch has joined #openstack17:09
*** kashyapc has quit IRC17:12
*** dendrobates is now known as dendro-afk17:12
edayvishy: any more details on that blueprint above jay is asking about? it was on the etherpad you had a few months back, if you can remember...17:14
vishyi think that was about making integration tests work in fake mode17:15
vishymeaning that you could run multiple workers and communicate in a test17:15
*** silassewell has joined #openstack17:16
edayso part of the integration testing overhaul that is needed17:16
*** al-maisan is now known as almaisan-away17:17
*** dpackard has joined #openstack17:23
dpackardhi all17:23
dpackardI am looking to manage a small cluster connected via gigabit ethernet17:24
edayjaypipes, gundlach: approved, but not marking as approved for merge since I'm not sure if this should go into the release since the RC has been cut17:25
dpackardIt needs to be lightweight, as the processors are 600 MHz and have 256MB ram17:25
gundlacheday: right -- i asked dendrobates to status=approve it if he wants to17:25
edaygundlach: ok, cool17:25
alekibangodpackard: how many computers?17:25
alekibangodpackard: today most people are busy a lot, as tomorrow is a release of nova17:25
dpackardalekibango: 8 computers17:26
dpackardthat's right! oct 21st17:26
alekibangodpackard: nova is not created to run on 256 MB computers, 16+ GB would be more appropriate :)17:26
dpackardthat's what I was afraid of..17:26
alekibangoyou will need one compter (or more) with sql17:26
*** joearnold has quit IRC17:27
dpackardyes, that is possible17:27
jonwooddpackard: It might work using UML17:27
alekibangodatabase, and on others, os will take 60-70 mb, nova also will take few mb, 128 mb will be there for virtual machines? thats pretty low17:27
alekibangojonwood: i didnt use uml for long time, so i cant compare it17:28
alekibangowhat purpose it should serve?17:28
alekibangodpackard: ^17:28
dpackardIt is a platform for embedded systems (robots)17:28
alekibangoif 2 virtual machines with 70 mb is ok for you :)17:28
alekibangodpackard: i never measured lowest requirements :)17:29
dpackardthe devices are Overo Gumstix, tied together on an Overo Stagecoach17:29
dpackardReally.. I think I am looking for something that doesn't require virtualization17:29
alekibangodpackard: it would be better to buy 2 normal cheap computers  and give them some ram17:29
jarrodi have a rack of dual quad-core dells w/ 16gig waiting for openstack17:29
alekibangojarrod: the same here :)17:30
dpackardalong the lines of the (now dead) openmosix project17:30
alekibangobut i have 8cores and not dell17:30
jarrodalek: this week?17:30
alekibangojarrod: asap17:30
jonwooddpackard: If you're not looking for virtualization, Nova isn't what you're looking for.17:30
jarrodwhen is it released17:30
dpackardalekibango: it is for visual processing in robotics17:31
dpackardjonwood: thanks :)17:31
jonwooddpackard: What were you hoping it would do?17:31
dpackardjonwood: I was hoping it would provide some basic load-balancing features on real compute nodes17:32
*** kashyapc has joined #openstack17:32
alekibangodpackard: buying 2 new servers might be cheaper :)17:32
alekibangoeven 'servers'17:32
dpackardalekibango: hehe, this is for robotics :)17:33
*** joearnold has joined #openstack17:33
alekibangoif they run small guests, thats fine17:33
jonwooddpackard: Yup, Nova's not that, and I'm afraid it's completely out of my field, so I'm not sure what would do it.17:33
alekibangodpackard: nova enables you to run many virtual hosts on a number of servers17:33
alekibangoand helps to manage it17:34
*** pharkmillups has quit IRC17:34
alekibangovirtual hosts can be small - for whatever reasons used17:34
dpackardalekibango: ok :) I think I will have to continue looking17:34
dpackardjonwood: thanks for your input!17:34
alekibangobut servers should not be that small :) even desktop computer my 2 year old daughter is using is having 2 GB RAM17:35
dpackardalekibango: goodluck with the release tomorrow!17:35
alekibangodpackard: heh, i am not the one responsible, but thanks :)17:35
alekibangodpackard: my atom desktop computer would be better server than yours 8 :)17:36
dpackardhahaha :p17:36
jonwoodalekibango: It would make an awful robot though.17:36
dpackardalekibango: precisely  ;)17:37
dpackardthanks guys!17:38
alekibangoi loved robotics when i was younger :)17:38
alekibangoit got me to computers really17:38
dpackardVery cool :) check out if you want to get some hobby/research platforms17:39
alekibangomaybe for my kids :)17:40
dpackardgotta run :) thanks again17:40
*** dpackard has quit IRC17:40
vishyfound a nasty bug17:42
*** perestrelka has quit IRC17:43
*** sophiap has joined #openstack17:43
vishyin fact a couple of them17:43
*** perestrelka has joined #openstack17:43
*** burris has quit IRC17:44
*** schisamo has joined #openstack17:47
*** rnirmal has quit IRC18:01
mtaylorjaypipes: I got that going for drizzle workitems via hudson ... I'm gonna spin it up for openstack via hudson now18:01
*** rnirmal has joined #openstack18:01
*** pharkmillups has joined #openstack18:03
*** Ryan_Lane is now known as Ryan_Lane|food18:03
*** metcalfc has quit IRC18:05
*** ibarrera has quit IRC18:06
jaypipesmtaylor: well played.18:10
jaypipesvishy: ya? where?18:10
vishyif there is a bunch of stuff in the queue when a worker starts18:11
vishyit does very badly18:11
*** khussein has joined #openstack18:11
vishybecause the service doesn't have a manager set yet18:11
jaypipesvishy: ah...18:12
vishyalso just got bit by dnsmasq pids wrapping18:12
jaypipesgundlach: got the test fixture reset environment working :)18:12
vishythere was a todo in there18:12
vishyso i just todid it18:12
jaypipesvishy: :) nice.18:13
*** jc_smith has joined #openstack18:13
*** daleolds has joined #openstack18:19
*** piken has quit IRC18:20
creihtannegentle: btw, in your source directory, you should be able to run "python build_sphinx" to build the docs locally18:22
creihtthen point your browser to  doc/build/html/index.html18:22
annegentlecreiht: great! trying it now.18:22
annegentlehm, invalid command, and when I look at I don't see a build_sphinx line18:24
creihtannegentle: then you probably don't have sphinx installed18:24
creihtwhat os are you running?18:24
annegentleok, yeah that's right. I'm on MacOS18:25
annegentleso apt-get won't work, right?18:25
creihtannegentle: if you type "easy_install" does it print out stuff, or show command not found?18:26
gundlachjaypipes: oh, lovely :)  send me a branch link when you have one18:26
annegentlelooks like easy_install is installed18:26
creihtthen try18:26
creihtsudo easy_install -U sphinx18:26
annegentlecool, it's downloading18:27
annegentlegeez what would I do without you guys helping me :)18:27
*** pharkmillups has quit IRC18:30
*** terje has joined #openstack18:33
*** alekibango has quit IRC18:37
*** pharkmillups has joined #openstack18:37
*** almaisan-away is now known as al-maisan18:37
*** silassewell has quit IRC18:38
*** Ryan_Lane|food is now known as Ryan_Lane18:39
*** burris has joined #openstack18:40
uvirtbotNew bug: #664073 in swift "swift-log-stats-collector can run over itself" [Medium,New]
*** khussein_ has joined #openstack18:47
*** alekibango has joined #openstack18:47
*** joearnold has quit IRC18:49
*** rnirmal has quit IRC18:49
*** rnirmal has joined #openstack18:50
*** khussein has quit IRC18:51
*** khussein_ is now known as khussein18:51
*** kakoni has joined #openstack18:54
*** ArdRigh has joined #openstack18:58
*** dendro-afk is now known as dendrobates19:01
*** perestrelka has quit IRC19:05
kakonihi guys. so hows the refrence swift cluster doc/setup coming up?19:05
*** perestrelka has joined #openstack19:05
*** rlucio has joined #openstack19:09
*** kashyapc has quit IRC19:10
*** sanjib has joined #openstack19:12
letterjmtaylor: I built packages from trunk and have the that code up and running in our lab.  It looks great!  Lots more testing needs to be done.19:21
alekibangokakoni: one server  for tests or multinode setup?19:22
kakonialekibango: multi-node setup19:23
alekibangobest we have atm is:
*** silassewell has joined #openstack19:24
alekibangoplease contribut there :)19:24
alekibangokakoni: see also wiki.openstack.org19:25
*** sirp has quit IRC19:27
dendrobatesmtaylor: are you around?19:35
alekibangojaypipes: we have response on the hijacking allegation  blog19:36
*** piken has joined #openstack19:39
*** sirp has joined #openstack19:40
*** DubLo7 has quit IRC19:41
jaypipesalekibango: interesting response. I agree with him on both of his points.19:42
alekibangoyes he is not dumb19:43
alekibangojaypipes: i believe its possible to correct somehow our course :) which we should do19:43
alekibangothe standardization is very important part of this19:44
alekibangojaypipes:  this is also interesting reading
alekibangogot me thinking a bit :)19:45
alekibangostill i believe in holy fork   :)19:46
pikenis Austin still being released stable tomorrow?19:47
alekibangoi think opencore-izing  will not help... that they should move to service oriented business - thats right economic layer -- linux distros are not (and should not try) to create intrustries19:47
alekibangopiken: prolly19:47
pikenI will keep my fingers crossed as we are trying to get moved from Euca to it and upper management just likes to see official releases.19:48
alekibangopiken: austin nova release is really initial19:48
jaypipesalekibango: yeah, ready that yesterday.  I don't really think Mark is all about opencore... I think Bradley is over-stating a few things..19:49
alekibangomight be, but still interesting read...19:49
pikenalekibango: even initial it is more refined then euca. lol19:50
alekibangopiken: hehe, :))19:50
jaypipespiken: understood.  we're trying our best.  AFAIK, we're on track to release tomorrow.  vishy, do you think those bugs you found are critical and worth holding hte release up for?19:50
vishyone is critical19:50
vishyi'll propose19:50
alekibangobug fixes  can always bubble into stable branch... :)19:50
kakoniand then swift and rsync. I guess the default max conns set to 2,4 and 8 (for objects). perhaps something that one could tune?19:52
*** sirp has joined #openstack19:54
*** khussein has quit IRC19:55
vishyjaypipes: i'm just trying to figure out if i can write a test case19:55
*** rnirmal has quit IRC19:55
*** khussein has joined #openstack19:55
*** rnirmal has joined #openstack19:55
creihtkakoni: waiting on hardware for the reference setup19:56
*** annegentle has left #openstack19:56
creihtkakoni: what was your other question in reference to replication?19:57
*** kakoni has quit IRC19:59
*** westmaas1 has quit IRC20:03
*** sirp has quit IRC20:04
*** kakoni has joined #openstack20:07
*** mtaylor has quit IRC20:08
kakonicreiht: other question; any recommendations for the rsync setup? Go with the values set in the default confs, or should we tune them?20:08
creihtfor a production cluster, with a lot of drives, the default is pretty good20:08
creihtat least a good starting point20:08
creihtkakoni: you had 2 drives per machine right?20:09
kakonicreiht: btw, so ran new swift test today. it was 1 proxy + 5 storage nodes (+1 swift-bench generator) (with new hardware). for 1-byte inserts we saw figures around 300-400 puts per sec20:09
kakonicreiht: not any more;) initially this was raid0 setup, now theres 4 drives per machine.20:10
creihtahh ok20:10
creihtI would use some smaller valuse then20:10
*** annegentle has joined #openstack20:10
creihtand that rate sounds pretty slow20:11
creihtbut it can depend on so many things20:11
kakoniwell better than the earlier 100 per sec, right;)20:11
*** sirp has joined #openstack20:11
creihtkakoni: some things to check to see where the bottle neck is20:12
creiht1.) check the cpu usage of the bench script... (it will only use one core).  If it is using close to all cpu, I would try running another instance20:13
creihtkakoni: btw, what do you have the workers option set to in the proxy, account, container, and object servers?20:13
creiht(if it isn't set, it defaults to 1, which probably isn't ideal)20:14
kakoniworkers are typically set to cores * 220:14
creihtkakoni: is that what they are set to in your config?20:15
kakoniyeah, so for instance that newer test done today took place in 4 core machines, so worker value was set to 820:15
creihtso then20:16
creiht2.) Check the cpu usage on the proxies and and storage nodes to see if something is maxing out on cpu20:17
creiht3.) check the iowait on the storage node to see if disk io is what is limiting20:18
kakoni2) Atleast in the proxy nodes cpu usage is minimal.20:18
creiht4.) normally you would check the network usage on the interfaces, but for that test, you shouldn't be saturating the network20:18
kakoni3) yep will do that20:18
creihtkakoni: what concurrency are you running the benchmark at? the default is 10, if that isn't set, keep bumping it up until requests/s stop going up20:19
*** krish has quit IRC20:20
kakoniyeah I cranked it up till about 150. That was the "breakeven";) point20:21
creihtkakoni: did you check 1.)?20:21
kakoniwell no, I wasnt actively monitoring bench machine cpu. something to put to my note list;)20:22
creihtyeah the benchmark is just single process, so if you max it out on cpu, then I recommend starting doing multiple bench processes20:23
creihtI need to add a capability for it to do it automatically20:23
*** pharkmillups has quit IRC20:23
*** littleidea has joined #openstack20:23
creiht5.) Check /var/log/syslog to see if any errors are showing up20:26
kakoni5) well usually there is always something;) connection timeouts etc..20:28
creihtkakoni: .5 second timeout?20:29
creihtsome of the timeouts are a bit aggressive20:29
*** joearnold has joined #openstack20:29
creihtthough on a small network, I imagine that shouldn't be a huge issue20:30
*** al-maisan is now known as almaisan-away20:30
*** burris has quit IRC20:31
kakonicreiht: yep. thats the one.20:31
creihtkakoni: are the connection timeouts to a single server, or do they appear to be accross the board?20:31
creihtkakoni: and are you running 1.0.2 or 1.1.0rc?20:32
kakoniwell;) latest trunk20:32
creihtok good20:32
kakoniso i guess thats 1.1.020:32
creihtI'm also going to try running similar tests in our lab with the swift-bench20:33
*** kevnfx has joined #openstack20:33
uvirtbotNew bug: #664146 in nova/austin "boto version should be 1.9b not 1.9b1 in pip-requires" [Critical,In progress]
*** mtaylor has joined #openstack20:36
*** ChanServ sets mode: +v mtaylor20:36
creihtkakoni: and how many objects are you putting in for the test?20:38
kakonicreiht: typically 1000020:39
kakonicreiht: but ok, ill ran some more tests tomorrow, will watch more closely for the cpu/iowait behaviour.20:41
kakoniokii, bye20:43
*** kakoni has quit IRC20:43
dendrobatescan someone review that please20:49
uvirtbotNew bug: #664150 in nova "Existing Messages in the queue break the workers." [High,In progress]
*** khussein has quit IRC20:52
dendrobatesjaypipes eday soren mtaylor: around?20:53
mtaylordendrobates: always?20:54
mtaylordendrobates: looking20:54
dendrobatesmtaylor: thx20:54
jaypipesdendrobates: y20:54
mtaylordendrobates: done20:54
dendrobatesjaypipes: just wanted a review mtaylor helped me out20:55
jaypipesdendrobates: k20:55
mtaylorit's what I do :)20:55
*** sophiap has quit IRC20:55
uvirtbotNew bug: #664153 in nova "Nova-network fails to start dnsmasq if a pid has been reused" [Medium,In progress]
dendrobatesmtaylor: quick question for a debian dev?20:57
mtaylordendrobates: yup20:57
dendrobateswould 2010.1~rc-0ubuntu4 be superseded by 2010.1-0ubuntu120:57
dendrobatesI always mess that up, if it is not simple20:57
*** joearnold has quit IRC20:58
*** ctennis has quit IRC20:58
mtaylordendrobates: no. the ~ only applies in the debian revision20:58
mtaylordendrobates: so, 2010.1-0ubuntu4~rc would be superseeded by 2010.1-0ubuntu420:58
*** joearnold has joined #openstack20:59
dendrobatesok, that will work thanks.20:59
mtaylordendrobates: debian packaging does not really have good structures for dealing with pre-releases ... since those aren't supposed to go in the archive anyway :)20:59
*** rlucio has quit IRC20:59
mtaylordendrobates: so - the problem you may run in to doing what I pasted...20:59
mtaylordendrobates: is that both 2010.1-0ubuntu4~rc and 2010.1-0ubuntu1 will expect a 2010.1.orig.tar.gz ... and the various archives will barf if you try to have two different tarballs called 2010.1.orig.tar.gz21:00
vishyproposed two bugfixes21:00
dendrobatesmtaylor: they would be in different archives21:01
mtaylorthen in that case it should be fine21:01
dendrobatesthe RC will be in a ppa21:01
dendrobatesthx again21:01
mtaylordendrobates: in general though we should probably come up with a 'better' strategy for dealing with what we're trying to do here for the future21:01
mtaylordendrobates: any time21:01
mtaylordendrobates: btw - you can do: dpkg --compare-versions  2010.1~rc-0ubuntu4 gt 2010.1-0ubuntu1 && echo "2010.1~rc-0ubuntu4 is greater"21:03
mtaylorif you ever want to test version math21:03
dendrobatesmtaylor: thanks21:03
*** mtaylor has quit IRC21:04
*** jdarcy has quit IRC21:06
*** allsystemsarego has quit IRC21:09
*** arthurc has quit IRC21:11
*** ctennis has joined #openstack21:14
*** ctennis has joined #openstack21:14
*** kevnfx has quit IRC21:27
*** pharkmillups has joined #openstack21:28
jaypipesvishy, soren, gundlach: if you have 3 minutes, would be good to get your input on  Thanks in advance :)21:30
vishylooks good21:31
vishyoutstanding question: does the rpc.cast to compute move into create_and_run_instance?21:32
vishysame for volume and network calls21:33
*** burris has joined #openstack21:33
jaypipesvishy: ya, as much as possible would move to compute node.  (basically, cloud controller would send a request to create and start up an instance, and the compute node would create it in the datastore and fire off tasks to the relevant worker nodes.) I'm thinking of yielding the instance id back to the cloud controller as soon as the instance is created in the data store, and then firing off the async jobs to the worker no21:36
jaypipesdes after that...21:36
vishyyeah that sounds perfect21:36
vishyexactly what i was thinking21:36
jaypipesvishy: cool.  will add it to our list of stuff to chat about at the summit :)21:36
vishyit is so meta to have ComputeManager using rpc.cast to call itself on another node21:37
jaypipesvishy: hopefully I can have a demo branch done by that point.21:37
jaypipesvishy: hehe, indeed.21:37
vishyjaypipes: you saw my email about network state and where i think we should go with ip allocation?21:37
jaypipesvishy: yep, I did, and 100% agree with that direction21:38
vishyjapipes: cool, yeah i'd like to have basically no code in the api's21:38
vishythe one part that is a little tricky is the key generation21:38
vishywe probably need a crypto manager21:38
jaypipesvishy: the only issue I foresee is the current Rackspace API is synchronous for adding a server and returning its IP :(  It will block waiting until the IP is assigned..21:39
vishycuz it was happening in auth for a while21:39
vishyjaypipes: time to rewrite the spec :)21:39
jaypipesvishy: but that's RS' problem ;)21:39
jaypipesOS API v2 :)21:39
edayvishy, jaypipes: So, pushing data creation to the worker might not mix well with vishy's idea yesterday about keeping workers read-only :)21:40
*** theron_ has joined #openstack21:41
*** metcalfc has joined #openstack21:42
* creiht kicks himself for making swift-bench just put to one container21:43
edayvishy, jaypipes: I think before deciding what should be created where (or really changing anything) we should figure out the data model and how to secure and scale it. We might end up having the schedulers doing all the writing if we go with vishy's idea yesterday, or worker nodes doing their writing and pushing up events from there.21:44
*** rlucio has joined #openstack21:55
vishyeday: it isn't going to the worker, it still runs in the api process, the code just moves into compute_manager21:55
edayvishy: ok, when I read "as much as possible would move to compute node", I thought node == worker21:56
uvirtbotNew bug: #664182 in swift "swift-bench should put to multiple containers" [High,New]
edayvishy: but yes, all for putting creation code in the correct manager class :)21:56
vishyoh good point, i missed that21:57
vishyeday: yes it really depends on where the writing is happening, that piece of the code should do the db calls21:57
*** mdomsch has quit IRC22:03
*** littleidea has quit IRC22:09
*** miclorb has joined #openstack22:18
*** theron_ has quit IRC22:21
*** silassewell has quit IRC22:23
*** pvo has quit IRC22:25
*** abecc has quit IRC22:26
*** Podilarius has left #openstack22:28
*** aliguori has quit IRC22:29
*** abecc has joined #openstack22:34
creihtmuch better22:44
*** rnirmal has quit IRC22:45
jaypipeseday: sorry, had to step away for a phone call....22:45
jaypipeseday: yeah, I was only talking about the write to the data store being done  by the manager, not the API server.22:47
creihtso in the simple lab config now with the updated swift-bench, I get 1300 PUTs/s and 2600 GETs/s to one proxy22:47
*** johnpur has quit IRC22:53
*** maplebed has quit IRC22:57
*** ppetraki has quit IRC22:59
edayjaypipes: do you mean the manager class in the API server process, or from the nova-compute worker?23:00
*** joearnold has quit IRC23:00
jaypipeseday: well, the compute_manager is not used in the API server process (it is called via rpc.cast to the compute node)23:00
*** joearnold has joined #openstack23:02
edayjaypipes: yeah, but it could be used on both ends. For example, both the openstack and ec2 api servers should call a common compute class/method to create data/send call/whatever23:02
*** dendrobates is now known as dendro-afk23:05
*** gundlach has quit IRC23:06
*** hazmat has quit IRC23:11
*** sophiap has joined #openstack23:11
vishyeday: I have a patch in for review23:14
cdub_does nova include any kind of guest agent?23:15
vishyif you have a second23:15
vishycdub_: no23:15
*** Ryan_Lane is now known as Ryan_Lane|away23:16
cdub_vishy: is there a plan for one, or considered outside the scope of nova?23:16
vishycdub_: yes, there is a blueprint for it i think, didn't make it into austin23:16
cdub_vishy: have a blueprint pointer?23:17
cdub_vishy: thank you23:18
vishycdub_: np23:18
*** jonwood has quit IRC23:18
*** pharkmillups has quit IRC23:19
*** gaveen has quit IRC23:19
cdub_hmm, XenStore...ok23:20
*** pothos has quit IRC23:30
*** pothos has joined #openstack23:30
*** gaveen has joined #openstack23:32
*** pvo has joined #openstack23:32
*** pvo has quit IRC23:32
*** pvo has joined #openstack23:32
*** ChanServ sets mode: +v pvo23:32
*** sophiap has quit IRC23:33
*** joearnol_ has joined #openstack23:34
*** joearnold has quit IRC23:37
*** rlucio has quit IRC23:41
*** gustavomzw has left #openstack23:43
*** joearnol_ has quit IRC23:49
*** terje_ has joined #openstack23:50
*** AimanA is now known as HouseAway23:50
*** sophiap has joined #openstack23:51
*** terje has quit IRC23:54
*** irahgel has left #openstack23:59

Generated by 2.14.0 by Marius Gedminas - find it at!