Friday, 2011-04-15

*** matiu has quit IRC00:21
*** matiu has joined #openstack00:28
*** matiu has joined #openstack00:28
notmynamettx: swift 1.3.0 approved00:35
*** winston-d has joined #openstack00:38
*** throughnothing has joined #openstack00:46
openstackjenkinsProject swift build #239: SUCCESS in 27 sec:
openstackjenkinsTarmac: Final Cactus versioning. Please accept when you sign off the release.00:47
*** adjohn has joined #openstack00:52
*** grape has joined #openstack00:58
*** JuanPerez has joined #openstack01:05
*** bcwaldon has quit IRC01:08
*** zigo-_- has quit IRC01:13
*** matiu has quit IRC01:14
*** dendro-afk is now known as dendrobates01:14
*** Ryan_Lane has quit IRC01:16
*** matiu has joined #openstack01:21
*** shentonfreude has joined #openstack01:29
*** hagarth has quit IRC01:35
*** mray has joined #openstack01:37
*** bcwaldon has joined #openstack01:44
*** mray has quit IRC01:57
*** rchavik has quit IRC01:59
*** zul has quit IRC02:05
*** zul has joined #openstack02:05
uvirtbotNew bug: #761324 in nova "I start a instance,but for a moment ,it shutdown" [Undecided,New]
*** eskp has quit IRC02:13
*** eskp has joined #openstack02:23
*** j05h has quit IRC02:32
*** grapex has joined #openstack02:38
*** j05h has joined #openstack02:49
openstackjenkinsProject nova build #818: SUCCESS in 2 min 28 sec:
openstackjenkinsTarmac: Final versioning for Cactus02:54
*** dendrobates is now known as dendro-afk02:57
*** dendro-afk is now known as dendrobates03:08
*** koolhead17 has quit IRC03:14
*** hggdh has quit IRC03:32
*** hadrian has quit IRC03:35
*** dendrobates is now known as dendro-afk03:40
*** joearnold has joined #openstack03:43
*** eskp has quit IRC03:44
*** smeier00 has joined #openstack03:45
*** bcwaldon has quit IRC03:56
*** gaveen has joined #openstack04:01
*** gaveen has joined #openstack04:01
*** joearnold has quit IRC04:01
*** eskp has joined #openstack04:05
*** joearnold has joined #openstack04:05
*** adiantum has quit IRC04:13
*** bcwaldon has joined #openstack04:16
*** adiantum has joined #openstack04:27
*** joearnold has quit IRC04:27
*** matiu has quit IRC04:33
*** eskp has quit IRC04:36
*** bcwaldon has quit IRC04:41
*** f4m8_ is now known as f4m804:43
*** throughnothing has quit IRC04:45
*** throughnothing has joined #openstack04:46
*** eskp has joined #openstack04:46
*** throughnothing has quit IRC04:47
*** guynaor has joined #openstack04:51
*** guynaor has left #openstack04:51
*** throughnothing has joined #openstack04:53
*** santhosh has joined #openstack04:53
*** mihgen has joined #openstack04:56
*** elambert_ has joined #openstack04:58
*** elambert has quit IRC04:58
*** elambert_ is now known as elambert04:58
*** rchavik has joined #openstack05:18
*** ccustine has quit IRC05:23
*** CloudChris has joined #openstack05:25
*** mihgen has quit IRC05:26
*** santhosh has quit IRC05:29
*** santhosh has joined #openstack05:34
*** elambert has quit IRC05:37
*** smeier00 has quit IRC05:49
*** jeffjapan has quit IRC05:49
*** kashyap has joined #openstack05:57
*** rsaidan has joined #openstack06:01
*** mihgen has joined #openstack06:02
*** mgoldmann has joined #openstack06:04
*** mgoldmann has joined #openstack06:04
zykes-anyone here from midokura ?06:08
adjohnzykes-: Ype06:12
*** jeffjapan has joined #openstack06:12
zykes-adjohn: does midostack use openvswitch ?06:15
adjohnzykes-: we have a virtual networking platform (midonet) which uses some of openvswitch06:15
zykes-ok, do you support san ?06:16
*** zaitcev has quit IRC06:16
*** herki has quit IRC06:18
adjohnWe support L2 switches, so in theory you could run FC or similar protocols on a virtual L2 switch06:19
adjohnOther than that, we don't have support for those protocols, but they could be added..06:20
*** santhosh has quit IRC06:22
zykes-adjohn: how does mido do storage then ? we have shared storage in our company so.06:24
adjohnOur distro also provides a block storage service which can run on commodity hardware (similar to ebs).. Though we are open to adding san support for specific customers needs.06:25
*** winston-d has quit IRC06:26
*** winston-d has joined #openstack06:28
winston-dBK_man : hi, do you have what libvirt.xml.template Nova compute is using?  there's one under /usr/share/nova and another one in /usr/lib/python/site-packages/nova/virt06:29
*** adjohn_ has joined #openstack06:34
*** daveiw has joined #openstack06:35
*** adjohn has quit IRC06:36
*** adjohn_ is now known as adjohn06:36
*** notmyname has quit IRC06:40
*** notmyname has joined #openstack06:41
*** ChanServ sets mode: +v notmyname06:41
*** shentonfreude has quit IRC06:43
*** notmyname has quit IRC06:46
*** notmyname has joined #openstack06:46
*** ChanServ sets mode: +v notmyname06:46
*** openstack has joined #openstack06:54
*** ttx has joined #openstack06:55
*** ttx has joined #openstack06:55
*** zenmatt has quit IRC06:57
*** antenagora has joined #openstack06:58
*** rcc has joined #openstack07:06
*** jstinson has quit IRC07:20
*** Gaelfr has joined #openstack07:21
*** lurkaboo is now known as purpaboo07:22
*** bkkrw has joined #openstack07:24
*** lionel has quit IRC07:27
*** lionel has joined #openstack07:28
*** gaveen has quit IRC07:36
ttxsoren: ping me when around07:39
*** nacx has joined #openstack07:42
winston-dHugoKuo_ : around?07:51
sorenttx: o/07:53
ttxsoren: hey, I'm about to tag release, you agree this needs to be done before branching cactus out ?07:54
sorenttx: If you're talking about bzr tags it's not very important, really. We can move those around at will. We shouldn't! But we could.07:55
*** allsystemsarego has joined #openstack07:57
*** CloudChris has left #openstack07:57
* ttx grabs a coffee and will upload release after that.08:02
*** fayce has joined #openstack08:02
*** fayce_ has joined #openstack08:03
*** fayce has quit IRC08:03
*** fayce_ is now known as fayce08:03
*** Gaelfr has quit IRC08:05
*** uksysadmin has joined #openstack08:17
zykes-ttx: do you know if there's been any interest for OS from Scandinavia ?08:18
zykes-like Bahnhof or some of the others.08:18
*** alekibango has quit IRC08:22
sorenzykes-: What's Bahnhof?08:23
soren(Other than a train station in German)08:23
zykes-soren: .se isp08:23
zykes-i guess some norwegian providers that could use OpenStack would be something like Basefarm / active2408:25
*** Gaelfr has joined #openstack08:26
ttxsoren: you can branch out lp:nova/cactus and push Final=False to lp:nova.08:27
ttxsoren: if you can document the proces for doing so at, would be great08:28
*** koolhead11 has joined #openstack08:29
zykes-i thought maybe you would know ttx ;)08:31
ttxzykes-: sorry, no.08:31
ttxI'm not the Scandinavian type.08:31
zykes-heh, soren?08:32
ttxlet's do some bugmail noise08:34
sorenzykes-: I don't know of anyone working with OpenStack in Scandinavia, no.08:37
zykes-would be fun to change that08:37
sorenVery mush so.08:38
sorenMuch, even.08:38
sorenSo much to do, so little time.08:38
ttxmush mush.08:38
*** mihgen has quit IRC08:38
*** ramkrsna has joined #openstack08:40
*** 77CAAF90J has joined #openstack08:52
sorenttx: Did you create a diablo series yet?08:57
ttxsoren: let me check08:57
ttxyes, the series is created08:57
sorenOk. lp:~hudson-openstack/nova/cactus created based on current trunk.08:58
sorenDo you have the LP magic covered?08:58
ttxdefien LP magic08:58
winston-dBK_man : is "injection of network settings" implemented in your Nova-2011.1 RHEL package?08:59
ttxsoren: I did the uploads, if that's what you mean08:59
winston-dBK_man : or do I have to apply additional patches?  I didn't find any related information on
sorenttx: I was more referring to the shuffling around of development series, pointing branches for the various series to the right places, etc.09:01
sorenAnyway. Coffee. Must. Have. Coffee.09:01
*** guigui has joined #openstack09:01
ttxsoren: no I didn't09:01
*** flopflip has joined #openstack09:02
*** flopflip_ has quit IRC09:03
*** dodeeric has joined #openstack09:06
zykes-you on a bug fix race ttx ?09:09
ttxon a bug releasing race, yes09:09
zykes-oh, why ?09:10
ttx(cactus is out, but don't tell anyone)09:10
zykes-hmm, why all the mails then about new bugs?09:10
ttxNot new bugs09:10
ttxFixCommitted -> FixReleased changes09:10
*** adjohn has quit IRC09:12
*** jeffjapan has quit IRC09:13
sorenttx: Ok.09:20
RichiHwhat's the status wrt release? iirc, this might/should happen today?09:21
ttxRichiH: don't tell anyone, but it's released already.09:21
RichiHcan i get images, already?09:21
sorenttx: I've set diablo to point to trunk and cactus to point to cactus.09:22
ttxsoren: ok, did you push the Final=False versioining in ?09:22
sorenttx: No, not yet.09:22
sorenttx: Shall I go and do the same for glance and swift as well?09:22
ttxsoren: I think you can do Glance.09:22
*** perestrelka has quit IRC09:23
soren?!? Glance is confusing.09:24
zykes-why ? :p09:25
RichiHin somewhat related news, two of us (not me :/) will be in santa clara in two(?) weeks. it seems we might be betting on openstack in a, for us, big-ish way. as such, i am wondering if there are any plans for a community council or similar where people with a strong interest can get involved. all of this is subject to openstack being a good fit for us, of course. but at least from what i saw up to now, i suspect this to be the case09:25
ttxsoren: yes, there was some strange cactus vs. trunk issues09:25
sorenWEll, for one, the glance trunk is owned by glance-core.09:25
RichiHttx: thanks09:25
ttxsoren: the reason is that jaypipes opened cactus early09:26
ttx(under glance-core)09:26
RichiH(also, i am wearing my openstack t-shirt today in anticipation of the release(thanks soren))09:26
ttxthen I think when lp:glance was pointed to it, it retained its glance-core ownership09:26
zykes-RichiH: us who?09:27
RichiHgerman ISP09:27
ttxsoren: so nothing prevents galnce-core peeps to actually bypass the merge control09:27
sorenttx: Nothing at all.09:27
sorenI'm fixing this up..09:27
*** alekibango has joined #openstack09:29
sorenErk, this is confusing.09:29
ttxok, it's officially out09:30
RichiHwhat is the current minimum hardware recommended to get my own openstack cloud going?09:35
RichiH should note this as well, imo09:35
*** adiantum has quit IRC09:35
ttxsoren: lp:nova still points to cactus09:35
alekibangoRichiH: that really depends on how you want to use it09:35
RichiHalekibango: for now, i just want it to run in test mode09:36
sorenttx: I know. Haven't fixed that part yet.09:36
sorenttx: Will do now.09:36
alekibango4-64 GB ram should do09:36
ttxsoren: preparing Final=False versioning09:36
RichiHalekibango: i would scavenge a few machines and get going09:36
alekibangoeven less if your virtual hosts will be small09:36
alekibangoor if you will want to use swift09:36
sorenttx: I set Bexar to "obsolete".09:38
alekibangoso cactus is out?09:38
alekibangocongrats :)09:38
alekibangoplease put appropriate urls in topic :)09:39
sorenttx: You disagree?09:39
ttxsoren: no, just feels strange :)09:39
* soren would argue that Bexar has been obsole for months :)09:40
sorenI think my keyboard is lossy.09:40
alekibangosoren:  +109:40
alekibangoi hope cactus will actually work :)09:40
sorenalekibango: I know 100% certin that come things to work.09:41
sorenLossy keyboard again.09:41
alekibangoor fingres09:42
sorenNope, they're all there.09:42
sorenI just counted.09:42
ttxneeds to be accepted before we open the floodgates09:43
*** rsaidan has quit IRC09:45
sorenI wonder if there's anything I need to do on Jenkins..09:47
ttxwe'll soon see.09:50
ttxsoren: you can mark lp:nova/cactus "mature", maybe09:56
sorenttx: Good point.09:59
*** NovSlave has joined #openstack10:01
alekibangoannegentle: it would be nice to include some real world config examples10:01
NovSlaveHi again guys10:02
alekibangointo docs, so people will be able to have something to start with10:02
NovSlavevishy ? here?10:02
sorenNovSlave: It's 3 AM in Vishy-land.10:03
sorenttx: Oh, dear.10:04
sorenttx: Seems our docs are not being updated.10:04
NovSlaveok no problem may you ll be able to help!10:04
alekibangohehe... soren its in your best interest to make installation of openstack easier :)10:05
sorenalekibango: That's all I do. All day :)10:05
alekibangosoren: i know... :)10:05
alekibangoits pretty hard job10:05
NovSlaveok thx just 2sec so i can set up myself to provide you with my logs10:05
sorenalekibango: I've had worse.10:06
alekibangosoren: some config examples would be helpfull.. sometimes docs and wiki are totally wrong10:06
alekibangoso having real world examples - with all touched config files... would help10:06
sorenalekibango: I used to have to make it easy to install something that I knew didn't work. And there was no way for me to fix it. If Openstack has problems, I can go and fix them.10:06
*** rsaidan has joined #openstack10:07
alekibangosoren: good point10:07
alekibangoreminds me sun tzu... ordering your general a task which army is not able to do is dangerous even for emperors10:07
ttxsoren: could you explain the impact of "docs not being updated" ?10:08
koolhead11hello all10:08
alekibangosoren: which docs?10:08
sorenttx: Ok, so apparantly, it's been quite a while since the docs on were updated. That's usually done from a Jenkins job, but something changed (in Nova), so the job didn't really work.10:08
ttxoh, ok10:09
ttxanything that doesn't mean I need to issue an emergency 2011.2.1 is fine by me.10:09
* ttx wants a sane weekend.10:09
ttxbefore the 3-week traveling frenzy.10:10
sorenre traveling next week, too?10:12
ttxI'll be leaving home early next weekend, so that's my last sane weekend for a while.10:12
*** guigui has quit IRC10:14
BK_manwinston-d: hi10:15
sorenttx: The job that updates the docs is the nova job. It triggers a bunch of downstream jobs that I really don't want to run.10:16
BK_manwinston-d: network injection is implemented for RHEL guests (RHEL-like)10:16
sorenttx: Hm..10:16
sorenttx: I guess I could disable the downstream jobs for a little bit.10:16
BK_manwinston-d: any of those who have /etc/sysconfig/network-scripts/ifcfg-ethX10:16
BK_manwinston-d: libvirt.xml is specified in /etc/nova/nova.conf: --libvirt_xml_template=/usr/share/nova/libvirt.xml.template10:17
ttxsoren: if you accept the open-diablo branch it will trigger "nova"10:18
ttxsoren: with the adequate bunch of downstream jobs ?10:19
sorenttx: right, but my concern is snapshotting the cactus final docs.10:19
ttxah, ok.10:19
sorenttx: ..but i don't have a current copy of them.10:19
* ttx lunches and lets soren sort the mess.10:19
sorenttx: Bah, I'll get it sorted somehow.10:19
soren...and then approve the open-diablo thing.10:19
* soren puts on his ninja suit10:19
*** shermanboyd has quit IRC10:21
NovSlavei m starting a new oneserver installation brb in 1 hour10:28
* soren -> lunch10:32
openstackjenkinsProject nova build #819: SUCCESS in 2 min 26 sec:
openstackjenkinsTarmac: Diablo versioning.10:39
*** DigitalFlux has quit IRC10:40
*** DigitalFlux has joined #openstack10:41
*** arun__ has joined #openstack10:55
*** arun_ has quit IRC10:56
sorenttx: Oh, crap.11:06
*** perestrelka has joined #openstack11:07
sorenttx: Failed to copy cactus release to the release ppa.11:08
sorenttx: Failed as in forgot.11:08
uvirtbotNew bug: #761623 in nova "Json response don't meet specification for addresses and metadata." [Undecided,New]
*** 77CAAF90J has quit IRC11:14
*** mihgen_ has joined #openstack11:15
*** ctennis has quit IRC11:19
*** adiantum has joined #openstack11:27
*** ctennis has joined #openstack11:31
*** ctennis has joined #openstack11:31
uvirtbotNew bug: #761652 in nova "get addresses servers/id/ips doesn't have handler" [Undecided,New]
ttxsoren: you can still copy it -- the new version isn't built yet.11:41
sorenttx: I tried.11:41
sorenttx: Couldn't.11:41
sorenThe copy packages thing doesn't show it.11:41
ttxalso you need to get rid of the gamma2 hack11:41
sorenOh, crap.11:41
ttxmaybe just delete the *king packages11:41
ttxbefore someone upgrades to them.11:42
* soren does so.11:42
ttxmight just make the previous ones available, who knows.11:42
sorenThey suffer from the same problem, really.11:44
sorenWell, not the exact same, but a very related one:11:44
sorenTheir package version says "~gamma2"11:44
ttxsoren: looks like you're in for a manual upload.11:44
onlanyis there any information available about Zones?11:45
sorenRemoved the ~gamma2 hack.11:45
onlanyi would like to know how many Zones can one compute node manage?11:46
onlanyand how many vms can be in one Zone?11:46
ttxsoren: the 2011.3~gamma2 packages are still pending at
ttxnot sure how to get rid of those11:48
*** arun__ is now known as arun_11:48
*** arun_ has joined #openstack11:49
daboonlany: It's the other way around: a zone can have multiple compute nodes, and each compute node can manage multiple VMs.11:50
*** rchavik has quit IRC11:51
daboonlany: see
sorenttx: What really?11:56
sorenttx: Hm..11:56
ttxsoren: maybe it wil scrap them once built, maybe11:56
sorenttx: I'll delete them as soon as I can.11:56
ttxsoren: document the process so that we don't make the same errors next time :)11:57
sorenHOLY CRAP!11:57
onlanythanks dabo11:58
*** rchavik has joined #openstack11:58
sorenPPA build queue: 16223 jobs (eight days)11:58
ttxyour builds are 12min away11:58
sorenMy current hypothesis is that the build will fail.11:59
soren..because the source package is gone.11:59
daboonlany: zones are simply logical groupings of services. They will usually be based on physical groupings, but that's not essential. For my testing, I have 9 nested zones on a single box!12:00
*** bcwaldon has joined #openstack12:00
*** nagyz has joined #openstack12:02
onlanywow :)12:04
*** aimon has joined #openstack12:05
onlanyi have to try out this too :)12:05
daboonlany: Oh, it's a lot of fun to manage!
onlanyu know about existing commercial solutions based on openstack, except cloudscaling?12:08
daboonlany: I don't really keep up on that side of things. It's in production at NASA, but I'm not sure they count as 'commercial'.12:09
*** adiantum has quit IRC12:11
*** drico has quit IRC12:11
zykes-onlany: Midokura?12:12
onlanythanks a lot12:13
*** h0cin has joined #openstack12:18
*** paltman has quit IRC12:25
ttxsoren: ok, the orphaned builds failed.12:25
*** paltman has joined #openstack12:25
sorenCrisis averted.12:25
sorenWindow of opportunity shattered.12:25
* soren exhales for the first time in an hour12:25
ttxsoren: still need to manually push to release PPA, and probably regen a trunk PPA build12:27
*** zigo-_- has joined #openstack12:27
sorenttx: I already did all of that. I just needed to be on top of this.12:27
sorenttx: Check it out:12:27
sorenFailed to build. I must suck.12:28
ttxdepwait on glance12:28
sorenI got it.12:29
zigo-_-soren & ttx: I got all working on my Debian test server now.12:29
sorenzigo-_-: With kvm?12:30
zigo-_-My issue was the uec-publish-tarball stuff totally broken...12:30
zigo-_-It should be written in BOLD letters on the doc that we should not use uec-publish-tarball.12:30
ttxsoren: might want to copy glance packages over from trunk ppa to release ppa, before my new version hits ?12:30
sorenzigo-_-: I use it *all* the time.12:30
sorenttx: Already did.12:30
zigo-_-I got stuck durring a week, thinking it was something else.12:30
ttxzigo-_-: uec-publish-tarball is fine... though it might rely on some specific version of euca2ools12:31
zigo-_-soren: Which version do you use then?12:31
sorenzigo-_-: Of what?12:31
sorenzigo-_-: uec-publish-tarball? Whichever is in Ubuntu.12:31
zigo-_-From cloud-utils, right?12:31
ttxzigo-_-: yes12:31
sorenzigo-_-: Sounds about right.12:31
zigo-_-Ok, I'll give it another try then.12:32
alekibangoit might just work now :)12:32
sorenzigo-_-: It's part of my automated test setup.12:32
sorenzigo-_-: It works on Lucid, Maverick, and Natty.12:32
zigo-_-Ok, thanks for the info.12:32
sorenSure thing.12:32
zigo-_-What's the set dead line for Cactus?12:33
sorenYesterday evening.12:33
sorenIt's released.12:33
zigo-_-So, we're now working on Diablo, then?12:34
sorenThat's the plan.12:34
ttxsoren: re: "Already did." -- can't see them in ?12:35
zigo-_-So, you can pull my patches for Debian now? :)12:35
zigo-_-What's the status?12:35
sorenttx: Sorry, I must have gotten confused somewhere. I have them ready to upload. I'll nuke the ones labeled "gamma2".12:35
zulis diablo ready yet? huh huh is it? is it? how about now? ;)12:36
zigo-_-zul: No, not ready, it STARTS development, as Cactus is released.12:36
*** ryker has joined #openstack12:38
*** dendro-afk is now known as dendrobates12:39
*** NovSlave has quit IRC12:42
zigo-_-soren: Does the unit tests work in the released Cactus?12:43
sorenzigo-_-: Sure.12:44
sorenzigo-_-: We run them on every single commit.12:44
zigo-_-I'll try with your orig.tar.gz then.12:45
*** hadrian has joined #openstack12:46
*** bcwaldon has quit IRC12:53
*** j05h has quit IRC12:54
*** aliguori has joined #openstack12:55
gholtsoren: Are the older packages of Swift supposed to have disappeared immediately when Cactus went live?12:55
sorengholt: I don't believe we've touched swift *at all*.12:56
sorenttx: ?12:56
gholtI don't really see older packages of nova either; maybe I just don't know where to look.12:56
ttxno.. except from releasing the tarballs on LP12:56
sorenttx: Oh, you did that?12:56
ttxgholt: what are you looking for exactly ?12:56
sorenI didn't know.12:56
gholtMore curious really. Trying to figure out things "work".12:57
ttxthey are all at
*** ppetraki has joined #openstack12:57
sorengholt: Can you be a tad more specific? :)12:57
ttxgholt: depends on what you call "older packages" obviously12:58
gholtThe tar balls yeah, but I thought we provided a form of packages for Ubuntu.12:58
ttxah, that I haven't touched12:58
gholtSo, if somebody was using them for their prod environment (which would be crazy maybe, but...) are they all of the sudden going to have new stuff if they kick a box today? 'Cause that'd be bad.12:59
sorengholt: We most certainly do.12:59
ttxsoren: when I said "cactus is released" I meant swift+glance+nova.12:59
sorengholt: Not yet, as we haven't touched Swift at all.12:59
gholtOkay, this page looks weird then?
sorengholt: But the current idea is that the PPA's aren't labeled "bexar", "cactus", and "diablo", but rather "trunk" and "release".13:00
ttxgholt: oh, that's the Natty packages. Apparently the ubuntu folks got busy13:00
sorenThe idea being that "release" should be well tested and fit for upgrades as well.13:00
ttxsoren: zul uploaded 1.3.0 to natty.13:00
gholtYou're going to have to have at least a release - 1 as well.13:00
sorengholt: How would that help?13:00
sorengholt: So today, they'd upgrade from austin to bexar.13:01
sorengholt: How's that useful?13:01
gholtWell, I'd keep all the releases, but you said you'd only keep trunk and release.13:01
zigo-_-soren: Where should I take cactus from? The PPA, or is it in the official Natty repo?13:01
sorenzigo-_-: It's not in natty yet.13:01
ttxsoren: not so sure of that.13:01
sorenttx: orly?13:02
*** ryker has quit IRC13:02
sorenttx: Oh, I see.13:02
sorenCool beans.13:02
sorenDarn it, I need to run for... an hours and fifteen minutes.13:02
gholtAh, found 'em:
ttxgholt: you mean, people should have a way to install, from PPA, the 1.2.0 release.13:03
ttxeven once the 1.3.0 release is out13:03
gholtOh yeah, definitely.13:03
gholtOr are you teasing me?13:03
ttxgholt: right, our basic release/trunk PPA layout doesn't allow for that13:04
ttxno, not teasing13:04
gholtWell, they can get them from the archive I found, that's good enough.13:04
ttxsomeone that enables release PPA currently says "give me the last stable release"13:04
*** alekibango has quit IRC13:05
gholtYeah, I think that's okay as long as they can get the old ones from somewhere.13:05
gholtI just couldn't figure out where the old ones were, but I know now:
*** Zangetsue has quit IRC13:05
ttxgholt: that's the trick -- I'm not sure that PPA will indefinitely contain Bexar.13:05
*** alekibango has joined #openstack13:06
ttxnothing in its description says so13:06
gholtIndefinite isn't a problem, but for at least a while would be good.13:06
*** mray has joined #openstack13:06
gholtMost folks that run production environments probably make their own packages to be honest.13:06
gholtBut if some crazed person uses the ppas, they aren't going to be able to instantly upgrade a huge cluster on release day.13:07
ttxgholt: mtaylor may upload new versions on that one at any time. The only way to be sure is to have a "Bexar" PPA and copy thoise packages to it13:07
gholtAnd their probably going to have to kick some boxes before they are ready to upgrade everything, so they'll need the code they're running on the rest of the cluster.13:07
zigo-_-ttx: Is this cactus?
ttxgholt: i guess we could have a Release-1 PPA.13:07
zigo-_-I'm not used with your naming schemes...13:08
ttxcactus release is 2011.2 for Nova, yes13:08
zigo-_-Merci! :)13:08
*** ryker has joined #openstack13:08
gholtttx: I'm not sure it's an issue, we can manually upload stuff to the archive, I'll bug notmyname/team on it later this morning and see what they want.13:09
ttxgholt: note that you can access old versions by looking at "superseded" packages13:09
ttxin a PPA13:09
gholtOh cool, where do you see that? Or you just mean from external packaging tools?13:10
ttxgholt: example:
ttx1791 old packages :)13:11
*** f4m8 is now known as f4m8_13:11
*** hggdh has joined #openstack13:11
gholtOkay cool, that's a lot of old stuff. :)13:14
* gholt bookmarks
*** kbringard has joined #openstack13:17
gholtWe also don't seem to provide packaging for Lucid, the latest stable Ubuntu, but I guess that really isn't an issue since, like I said, most prod folks probably won't use them anyway, heh.13:18
kbringardhappy release day friends13:18
*** fayce has quit IRC13:20
*** kbringard has quit IRC13:20
*** kbringard has joined #openstack13:20
*** bcwaldon has joined #openstack13:22
nagyzI've entended both and in the compute node to add a new call13:26
nagyzand it works fine13:26
nagyzexcept, that the resulting string is escaped.. :)13:27
*** HugoKuo_ has quit IRC13:28
uksysadminhey - I've just installed nova from nova-core/trunk and I get a bug with creating a keypair after I've already created that keypair13:30
uksysadminkevinj@nova-api:~$ euca-add-keypair --debug openstack13:31
uksysadminUnknownError: An unknown error has occurred. Please try your request again.13:31
uksysadminknown bug or should I raise an issue? [or someone point out what I'm doing wrong (apart from running the same command again and again! ;-))13:31
doudeHi all, when I start a VPN on a project (in VLANnetwork mode), the cloudpipe instance didn't take the first IP of the subnet (in my case I use the command 'nova-manage vpn run p1'. How the cloudpipe instance requests specially the firts IP of the subnet ?13:32
uksysadmin(deleting the keypair and creating it again works, btw)13:33
kbringarduksysadmin: what do you see in your nova-api.log?13:33
kbringardany errors?13:34
kbringard /var/log/nova/nova-api.log to be exact13:34
kbringardassuming that's where you've set logging to do13:35
uksysadminadd keypair (when exists) nova-api.log
uksysadminyeah its logging tehre13:36
uksysadminjust seems an unhandled exception - no biggie.13:36
kbringarduksysadmin: it seems to think the keypair name already exists13:36
kbringardDuplicate: The key_pair openstack already exists13:37
uksysadminyeah that's correct13:37
uksysadminits the output that isn't nice13:37
kbringardwhich would explain why removing it and readding it works13:37
kbringardoh, ok13:37
*** dendrobates is now known as dendro-afk13:37
kbringardso you're asking about the UnownError output13:37
kbringardnot why it's not working13:37
uksysadminI'll raise a bug - no showstopped in the grand scheme of things - just think it can be handled better13:37
kbringardgot it, sorry, I was wrong about what you were asking13:38
nagyzwhere is returning the data thru rabbitmq from a method call in on any service is handled?13:40
kbringardsorry nagyz, I don't know13:44
*** JuanPerez has quit IRC13:44
kbringardbut I know it sucks to ask a question and get no reply at all, so I figured I'd let you know I saw it :-D13:44
nagyzthanks :)13:44
nagyzI've extended the serialization metadata to include adding xmls from strings13:45
nagyzand it works fine13:45
nagyz...except the string (which contains the xml) returned by an internal API call gets escaped while it travels thru rabbitmq13:45
*** gondoi has joined #openstack13:47
*** dspano has joined #openstack13:48
uvirtbotNew bug: #761774 in nova "Adding a keypair that already exists produces an Unknown error" [Undecided,New]
ccookehmm. How possible would it be to install Windows from an ISO in a VM on openstack?13:52
*** ramkrsna has quit IRC13:52
ccooke(I'm currently evaluating openstack here and it's nice enough to manage VMs for myself, so I wonder if I could get away with using it for the windows VM I have to keep around for work)13:52
dsockwellccooke: i haven't actually done it, but there is reason to be hopeful, considering the hypervisors used13:54
*** iammartian has left #openstack13:54
ccookeI guess I could set it up as a non-managed kvm vm and just import it later13:55
kbringardccooke: yea, or install it in like virtualbox and then convert it to qcow213:55
kbringard(which is basically what you said, but sounds easier to me, because I'm lazy) :-D13:56
kbringardso, I've got an odd one13:58
kbringardthe file is there13:59
kbringardwhen I run the qemu command manually, I get the same error13:59
kbringardbut, here is the weird part13:59
doudeDid you test this functionnality ? cloudpipe ?13:59
kbringardwhen I do like, euca-run-instances -n 6 with that image id13:59
uksysadminright - an eon ago (month or so) I used to be able to assign a floating address so I could access my openstack instances as part of my network using: sudo nova-manage floating create hostname IP/CIDR14:00
kbringardit works on the one compute node I have that is the samy physical host as the glance api/registry server (glance is using file backing)14:00
kbringardbut the 2 other compute nodes, it fails14:00
kbringardI've even tried copying the file over manually (both via curl to the glance API, and scp)14:00
uksysadminThat creates entries in my nova.floating_ip table but when I associate with an instance I get ApiError: ApiError: Address (IP) is not allocated14:00
kbringarduksysadmin: if you're using VLAN mode, make sure the IP has a project associated with it14:01
uksysadmine.g. euca-associate-address -i i-00000002
*** DogWater has joined #openstack14:01
*** bcwaldon has quit IRC14:01
* uksysadmin goes off to check my configs14:01
DogWateranyone looked this over?
kbringardthat's how mine looks in the db14:02
uksysadminhmmm, yeah my project_id rows are null14:03
kbringarduksysadmin: I'd just do a quick update to those tables14:04
uksysadminI need to shoot [cheers for help so far kbringard] so I'll have a look at what I'm doing wrong14:04
uksysadminwill do14:04
kbringardI don't know if you're doing anything wrong14:04
kbringardI thought I recalled it working awhile back, and then recently I had to update it manually14:04
kbringardperhaps there was a change to the nova-manage script14:04
uksysadminyeah - if its not something I can set on the command line, I'll assume its a bug14:05
kbringardseems reasonable to me :-D14:05
uksysadminconfirm that's working now after updating the table14:05
kbringardoh, on that previous weirdness about the images, the checksum in the glance server matches what is on the remote servers , so it doesn't seem like the image is getting munged14:05
* kbringard scratches his head14:06
kbringardoh, and you're most welcome uksysadmin14:06
kbringardI've gotten so much help here, it's the least I can do to give back when I can14:06
*** Zangetsue has joined #openstack14:07
*** mahadev has joined #openstack14:08
DogWaterIs it really common in Openstack clouds to use non-raided local storage to store VMs?14:10
DogWaterI was a tiny bit surprised to read that in the Dell/Rackspace bootstrapping document14:10
kbringardnot sure, last time I read that was like, 3 months ago14:11
kbringardlet me go back and reskim it14:11
* soren returns14:11
kbringardwb soren14:11
DogWaterI'm just asking what you folks actually do in real life, rather than what Dell is proposing people do.14:11
kbringardI've got my VM instances themselves running on local disk, which is hardware raid 114:12
kbringardand I've got my images on an nfs mounted volume14:12
kbringardthe backing for the images is raided, etc14:13
DogWaterIt seems like that document is written for service providers and hyperscale clouds in a service provider environment it seems a bit risky to store customer data in JBOD14:13
creihtDogWater: I think it all depends on how much you want to protect yourself from failure14:13
alekibangoDogWater: some companies who do cloud ready applications do not want raid14:14
kbringardif you're using swift, then in theory, that is where you have the fault tolerance for your data stores14:14
*** Tongpow has joined #openstack14:14
alekibangoas its cheaper to live without one and they are ready to accept some problems of the node14:14
alekibangothey have hundreds nodes ready14:15
kbringardand VMs are considered volitile... if you have data you want to save then you should be putting it on a volume, which would probably be backed by swift14:15
creihtswift is indeed redundant, and can be used with glance to store your base images, but can't be used for local storage or volumes14:15
alekibangono, swift is cold storage14:15
kbringardoh right, because of lvm14:15
kbringardyea yea, my bad14:15
alekibangoDogWater: if you want redundancy, look at nova-volume more closely14:15
alekibangoand on sheepdog or rbd14:15
alekibangoi am betting on sheepdog14:15
*** Tongpow has left #openstack14:16
kbringardwhat I meant was, you snap your volume and store it back to swift (or sheepdog, or whatever)14:16
alekibangowhich should be production ready (stable) in may14:16
alekibangoyes, snap it14:16
alekibangoyou just do backups its safe enough for most people14:16
DogWaterWell, I was just trying to spec out a reference design for a cloud and I was planning on doing the whole "enterprise SAN" thing which is why I was a bit surprised that Dell suggested storing the data on non-raided disks14:16
sorenttx: Ok, status? Where are we at?14:16
*** grapex has quit IRC14:16
sorenttx: We should set up an etherpad with stuff that needs doing.14:16
doudeHow the cloudpipe instance gets the first IP of the subnet in VLAN network mode ?14:17
alekibangoDogWater: i know how you feel, but its hard really to provide those, as nova can be used 1000 ways14:17
* creiht goes to look at the disks14:17
creihterm docs14:17
alekibangokbringard: look here for some more examples
kbringardalekibango: thanks, I'll check it out14:17
alekibango(i do not endorse the distro, i didnt try it,.... byt there are some images you might like)14:17
DogWaterin fact a lot of the "software cloud" solutions require using centralized storage (such as OnApp.. etc)14:17
DogWaterSo if anything I am more confused by reading that document than ever =)14:18
alekibangoDogWater: with nova, you can avoid central point of failure14:18
alekibangobut its not easy14:18
kbringardDogWater: haha, yea14:18
alekibangoas easy as it should be14:18
alekibangoDogWater: it will be easy in  6 months imho14:18
alekibangofor now its only for the brave... who has many computers14:18
alekibangoand experience14:18
kbringardalekibango: +114:19
*** daveiw has quit IRC14:19
alekibangobut its the most open,powerfull, scalabe, adoptable  cloud solution14:19
alekibangothose things also mean its very hard to get it running :)14:20
DogWaterSo nova is a compute platform/hypervisor?14:20
alekibangonova manages virtual hosts in a sane way14:20
DogWateror it's the front-end to the hypervisor14:20
creihtDogWater: in the dell doc, it seems that they give a pretty good story for both sides14:20
alekibangokind of frontend in your sentence14:20
creihtuse raid for local storage if you want that extra durability on that node14:21
creihtor don't to get better performance/cost in general and build your apps on the cloud knowing that your vm's are etherial(sp?)14:21
alekibangoyou can even have diskless nova nodes -- running virtual servers having disks in SAN14:21
DogWatercreiht: I see if that node goes down though, there is no HA?14:21
creihtfor a user's vm, no14:21
creihtit is the same for ec214:21
*** dendro-afk is now known as dendrobates14:22
*** chuck_ has joined #openstack14:22
DogWaterand people are fine with that?14:22
alekibangoDogWater: you can have disks on network using sheepdog14:22
DogWaterI guess that goes back to the 'scaling out' thing, where customers have apps on multiple nodes.14:22
*** zul has quit IRC14:22
alekibangoor RBD (ceph)14:22
DogWaterload balancing, etc14:22
creihtDogWater: right, if they are just using one cloud instnace, then they are not really using the cloud :)14:22
alekibangoits more like virtual hosting14:23
alekibangocreiht: but there are such uses...14:23
kbringardso, since people seem to be awake now, I've got a kind of head scratcher14:23
creihtthings like ceph and sheepdog are interesting, but you really pay the price in performance14:23
kbringardif anyone is interested14:23
creihtdsockwell: thank you :)14:23
*** reldan has joined #openstack14:23
*** j05h has joined #openstack14:23
blamarkbringard: bring it14:23
kbringardok, so14:24
alekibangocreiht: do you think its really not worth?  sheepdog is (almost) as fast as local nfs14:24
creihtand nfs is blazing fast! :)14:24
alekibangoand can do some tricks with volumes also14:24
DogWatercreiht: I guess the big downside with centralized storage is going to be scalability, and cost.. performance is generally ok?14:24
creihtDogWater: centralized storage as in a san?14:24
DogWatercreiht: have you messed with gluster+native NFS at all? I was fairly impressed but it does add some overhead.14:24
creihtDogWater: I haven't messed with that actually14:25
kbringardI've got an image that I uploaded into glance, when I launch it it works fine when it runs on the compute node that physically is the same host as the glance server, but when it goes to any other compute nodes, I get qemu-img errors14:25
DogWatercreiht: yes, storing the VMs centrally in a san, such as equallogic or DIY iSCSI14:25
alekibangoDogWater: i tested gluster... i like it but its slow14:25
dsockwellmy plan was to store VMs on an OpenSolaris/OpenIndiana box14:25
creihtDogWater: I think it depends on how big your cluster is going to get14:25
kbringardbut, I checked, and the file is in _base14:25
creihtor what your plans are14:25
kbringardand it has the same md5sum as what glance thinks it should have14:25
dsockwelland it would have worked if not for my cheap FC switch with a bad PSU14:25
blamarkbringard: What errors?14:25
kbringardis one14:26
creihtDogWater: traditional san is doable, but is expensive, and is only going to scale so far14:26
kbringardI've also gotten an error about info chardev as well14:26
kbringardbut it's not doing it any more14:26
DogWaterdsockwell: there are some interesting raid controller caching things hitting right now.. LSI/dell cachecade, adaptecs SSD/SAS cache14:26
kbringardthese same compute nodes work fine with all the other images I've uploaded14:26
kbringardso it doesn't seem to be a problem with the compute nodes themselves14:26
*** openstackjenkins has quit IRC14:26
kbringardand I've tried reupping the "broken" image14:26
creihtDogWater: yeah, My group will be playing with that soon :)14:27
kbringardbut it does the same thing14:27
alekibangocreiht: problem is, most people do nto have plans14:27
dsockwellDogWater: i have 2 SSDs set up as logs atm14:27
alekibangothey just want to 'get cloudy'14:27
alekibangoand they need some examples14:27
alekibangoso they can get idea what they really want14:27
*** openstackjenkins has joined #openstack14:27
alekibangothose examples (use cases) should be well documented14:27
creihtDogWater: and we are looking at making something around an iscsi stack14:27
alekibangoand even all example config files should be published somehow14:27
dsockwelli haven't done a lot of testing (stupid silkworm) but they seem to soak up IOs pretty well14:27
creihtI should have more to talk about it at the design summit14:27
alekibangomany peolpe have no idea what to do or why14:27
kbringardblamar: I've also tried copying the file over manually (via both curl from the glance api directly, and scping it)14:27
blamarkbringard: Permissions issue?14:28
kbringarddon't think so, it has the same perms as all the other images in there (that work)14:28
*** johnpur has joined #openstack14:28
*** ChanServ sets mode: +v johnpur14:28
kbringardand, when I run the qemu-img command manually as root, it tells me the same thing14:28
*** j05h has quit IRC14:28
kbringardsame versions of everything I've checked on all the machines14:30
kbringardblamar: weird, right?14:30
DogWaterwhen deploying load balancing for a cloud would you use physical load balancers or VMs?14:30
DogWaterseems like there would be a benefit to having physical load balancing upstream from the 'cloud'14:31
*** zenmatt has joined #openstack14:31
blamarkbringard: yeah, thinking...14:31
zigo-_-Sorry, wrong window.14:32
kbringardthat's the other error I was getting14:33
kbringardbut that one seems to have stopped14:33
*** kirshil has joined #openstack14:33
*** kakella has joined #openstack14:34
kbringardI upgraded this morning to the latest ppa, so perhaps that's what resolved that one error14:34
*** kakella has left #openstack14:34
*** kakella has joined #openstack14:34
*** kakella has left #openstack14:34
*** Gaelfr has quit IRC14:34
kirshilWhere is possible to get most up-to-date doc on nova compute, in particulare the description on 8773 api?14:35
*** j05h has joined #openstack14:35
*** koolhead11 has quit IRC14:36
*** reldan has quit IRC14:36
annegentlekirshil: working on Cactus doc copy right now, and a new landing page at What specifically is 8773 api?14:37
*** shentonfreude has joined #openstack14:38
*** grapex has joined #openstack14:38
*** Gaelfr has joined #openstack14:39
kbringardblamar: one additional bit of info: I "snapped" this from another running instance14:40
blamarkbringard, the compute node?14:40
kbringardby which I mean, I took the disk, disk.local, etc14:40
blamarkbringard, ah14:40
kbringardand copied it elsewhere then uploaded it into glance14:40
kbringardbut, I would have thought that if it was an issue with the image itself14:40
kbringardthen it wouldn't work anywhere14:40
kbringardbut it works on that one compute node (which incidentally isn't the one I took it from originally)14:41
kbringardit's got my mind boggled14:41
*** grapex has quit IRC14:43
sorenmtaylor: Paging dr. Taylor.14:43
kbringardblamar: and now it's gone back to the chardev error14:44
nagyzsoren, do you see any reason why the answer from an api call gets escaped? I wouldn't call that seamless rpc :-)14:44
sorennagyz: Can you be more specific?14:45
sorennagyz: This isn't still about the libvirt xml stuff, is it?14:45
nagyzif I add a new method to compute/ and, and call it thru rabbitmq, and lets say the method returns "<x>a</x>", it gets espaced on the recieving end14:46
nagyzand yes, it's still about that.14:46
nagyzI've managed to add the returning XML into the DOM nicely14:46
sorenThat's exactly what should happen.14:46
sorenYou're returning a string that happens to contain XML.14:46
uvirtbotNew bug: #761827 in nova "Release floating IP before disassociate it" [Undecided,New]
nagyzand every string gets escaped that goes thru rabbitmq?14:47
sorenYou're not extending the reponse XML with more XML. You're encapsulating a string that happens to be XML.14:47
sorennagyz: This has nothing to do with rabbitmq.14:47
nagyznote, we're only talking about API calls. Noone said anything about any other XML.14:47
nagyzit isn't the same escape problem as yesterday, I've managed to fix that :)14:48
nagyzlet me rephrase my problem14:48
*** elambert has joined #openstack14:49
nagyzI've added this to
nagyzok :)14:49
nagyzits only two lines, but I'll make a more complete example that way14:49
*** RJD22 is now known as RJD22|away14:50
*** rnirmal has joined #openstack14:52
nagyzthe logs sais: <a>123<\/a>14:52
nagyzshould that happen?14:53
nagyzthat's why I said rabbitmq is involved, or at least the process that does the rpc14:54
sorenI would expect things to be unescaped by then.14:54
nagyzany pointers where should I look in the code?14:55
dspanoThe link to the cactus release notes on is broken. It links to this URL: http://http//
*** RJD22|away is now known as RJD2214:55
*** j05h1 has joined #openstack14:56
*** jkoelker has joined #openstack14:56
ccookehmm. Deleting a network with nova-manage14:57
ttxdspano: indeed14:58
ccookeit says it failed, but I'm not even sure how to identify the network to the command - there's no usage information14:58
*** dragondm has joined #openstack14:58
ccookeah, got it14:58
mtaylorsoren: ola14:58
*** j05h has quit IRC14:58
*** grapex has joined #openstack14:59
*** ryker has quit IRC14:59
ttxnotmyname: if you accept you should be all set to start working on diablo/1.415:00
*** ryker has joined #openstack15:00
ccookeSo, I've deleted the project the network is associated with, but nova-manage complains the network needs to be disassociated with the project before it can be deleted...15:00
notmynamettx: done15:01
ttxThe Diablo floodgates are now officially open.15:03
kbringardttx: cool, I have a merge proposal :-)15:04
ttxkbringard: cool, I need some sleep15:04
kbringardhaha, go sleep amigo15:04
kbringardgreat work on cactus15:04
johnpurttx: Awesome job man!15:04
ttxjohnpur, kbringard thanks!15:05
annegentlettx: thanks for all the hard work!15:05
*** shentonfreude has quit IRC15:07
*** jonkelly has joined #openstack15:08
*** shentonfreude has joined #openstack15:09
*** uksysadmin has quit IRC15:11
*** shentonfreude has quit IRC15:13
nagyzhave a nice weekend, all15:14
*** nagyz has quit IRC15:14
*** reldan has joined #openstack15:15
*** dmshelton has joined #openstack15:15
*** shentonfreude has joined #openstack15:15
openstackjenkinsProject swift build #240: SUCCESS in 28 sec:
openstackjenkinsTarmac: Diablo versioning (1.4-dev).15:17
*** kirshil has quit IRC15:18
*** kirshil has joined #openstack15:19
kirshilannegentle: 8773 is a port which is used by default by EC2 style API. Actually I'm mostly interested in libcloud - EC2-Openstack  API15:20
kirshilAlso why the API both 8773 and 8774 are not SSL-encoded?15:22
*** Gaelfr has quit IRC15:26
*** enigma1 has joined #openstack15:37
jk0can I get a quick look at please? it's just a one-liner :)15:37
zigo-_-ttx: Cactus ships a novamanage.1 instead of a nova-manage.1 ... :)15:37
zigo-_-Very deceptive ... :)15:38
ttxzigo-_-: if only you would have looked into it three days ago15:39
zigo-_-ttx: It took me a long time to find this out (I just did...)15:40
*** rchavik has quit IRC15:40
*** chuck_ is now known as zul15:42
*** jonkelly has quit IRC15:42
*** zul has quit IRC15:42
*** zul has joined #openstack15:42
ttxzigo-_-: fortunately, nothing that can't be solved at packaging level.15:42
* ttx calls it a day, a week and a release.15:43
*** fayce has joined #openstack15:44
ttxtime to drink and celebrate.15:44
*** j05h has joined #openstack15:46
*** koolhead17 has joined #openstack15:46
*** j05h1 has quit IRC15:48
*** j05h has quit IRC15:48
zigo-_-ttx: Guess what I'm currently doing... :)15:48
*** bkkrw has quit IRC15:48
openstackjenkinsProject nova build #820: SUCCESS in 2 min 30 sec:
openstackjenkinsTarmac: Only poll for instance states that compute should care about.15:49
jaypipesttx: have a beer for me :)15:50
*** dendrobates is now known as dendro-afk15:50
*** fayce has quit IRC15:53
*** j05h has joined #openstack15:54
*** fysa has joined #openstack15:55
kbringardanyone have any idea what causes this error:15:55
kbringard(nova.compute.manager): TRACE:15:55
kbringardError: operation failed: failed to retrieve chardev info in qemu with 'info chardev'15:55
*** dragondm has quit IRC15:56
*** stewart has joined #openstack15:56
*** mihgen_ has quit IRC15:57
*** duffman_ has quit IRC15:58
*** duffman_ has joined #openstack15:58
*** j05h has quit IRC15:58
*** j05h has joined #openstack15:58
dspanoWhat flag in nova would I have to set to run libvirt under software emulation for testing on a PC I have setup?16:00
*** j05h has joined #openstack16:00
dspanoI'm getting this error when I try to start instances: Instance '1' failed to spawn. Is virtualization enabled in the BIOS?16:00
kbringardwhat's it say in the trace?16:01
kim0dspano: perhaps in nova.conf → --libvirt_type=qemu ?16:02
*** gondoi_ has joined #openstack16:03
*** maplebed has joined #openstack16:03
openstackjenkinsProject nova build #821: SUCCESS in 2 min 29 sec:
openstackjenkinsTarmac: Removed the unused self.interfaces_xml variable.16:04
*** gondoi_ has quit IRC16:04
*** gondoi_ has joined #openstack16:04
dspanoI'll give that a try.16:05
dspanoThe trace is kind of big.16:05
kbringarddspano: put it into paste.openstack.org16:07
BK_manCactus RPMs for RHEL6.0 are out:
*** mahadev has quit IRC16:10
kbringardBK_man: nice16:11
kbringardgood work sir16:11
uvirtbotNew bug: #761913 in openstack-dashboard "python tools/ errors with python2.7" [Undecided,New]
*** glenc_ has joined #openstack16:16
*** ryker has quit IRC16:17
*** hggdh has quit IRC16:17
*** ryker has joined #openstack16:18
*** eday has quit IRC16:19
*** glenc has quit IRC16:19
*** redbo has quit IRC16:19
*** eday has joined #openstack16:20
*** hggdh has joined #openstack16:20
*** redbo has joined #openstack16:20
*** BK_man has quit IRC16:23
*** BK_man has joined #openstack16:26
*** mahadev has joined #openstack16:28
*** nacx has quit IRC16:30
*** HugoKuo has joined #openstack16:30
colinnichis there any way to influence the Last-Modified header returned from swift?16:31
notmynamecolinnich: not yet. we've talked about the ability to change it (ie keep it separate from the put timestamp), but I don't think any code's been written for that yet16:32
*** kashyap has quit IRC16:32
*** dragondm has joined #openstack16:32
colinnichnotmyname: ok, ta16:32
notmynamewell, no other way than to upload at a different time ;-)16:32
notmynamestep one: invent a time machine :-)16:32
colinnichnotmyname: in 2007? :-)16:32
*** MarkAtwood has joined #openstack16:34
colinnichnotmyname: I take it X-Object-Meta-Mtime is only used by st?16:34
notmynamemaybe (ie I'm not sure)? it's a standard metadata header, so anyone can set/read it16:35
colinnichnotmyname: If I download the file, it ends up with a correct modification time, and that meta header is set to 2007, so I guess that's how st does it16:35
notmynameya, st sets it as the actual mtime of the file16:36
*** joearnold has joined #openstack16:36
colinnichnotmyname: ok16:36
* zigo-_- got a lintian clean nova package! :)16:36
*** world_weapon has joined #openstack16:37
zigo-_-Just finished fixing all remaining issues ...16:37
*** KnuckleSangwich has joined #openstack16:38
KnuckleSangwichAnyone know the current status of the S3 api compatibility layer in Swift? Is it still experimental?16:39
zigo-_-Ok, I'll wait for that, try to fix it, upload and then finally ... sleep ! :)16:39
*** fysa has quit IRC16:39
zigo-_-Ooops, wrong window again.16:40
Nick0laHello all.  Silly question (another one)...16:41
Nick0laif I use flat dhcp in my set up, does nova act as the DHCP server, or do I have to configure my own?16:42
patri0tNick0la: Dnsmasq will do that16:42
*** fysa has joined #openstack16:42
patri0tand nova handles that part (start, config dnsmasq)16:43
kbringardpatri0t: I've also submitted a patch to allow you to set q dnsmasq_config_file in the event that you want to fine tune the dnsmasq settings16:44
kbringardas an aside16:44
dspanokbringard: Sorry I just got dogpiled by a bunch of people in my office. I just put it on paste.openstack.org16:44
kbringarddspano: no worries16:44
dspanokbringard: It16:44
kbringardwhat's the url?16:44
kbringardLockFailed: failed to create /usr/lib/pymodules/python2.6/cld01.Dummy-1-249816:44
kbringardthere's your problem16:44
patri0tkbringard: nice16:45
kbringardsorry, was looking it up16:45
kbringardI have mine set to16:45
kbringard /var/lib/nova/tmp16:45
*** chuck_ has joined #openstack16:45
BK_mankbringard: are you using my build? :)16:45
*** zul has quit IRC16:45
*** chuck_ is now known as zul16:46
kbringardBK_man: don't think so, I'm running ubuntu16:46
dspanokbringard: Thanks a lot. I'm going to learn python because of this project. :)16:46
uvirtbotNew bug: #761947 in nova "create server response json doesn't contain progress" [Undecided,New]
kbringarddspano: if you put --lock_path=/var/lib/nova/tmp in your nova.conf16:46
kbringardthat should resolve it16:46
BK_mankbringard: that mean that I have the same for RHEL :)16:46
kbringardor any path that the user nova has write access to16:46
kbringardBK_man: I'm probably going to use your build once Cent6 gets released16:47
kbringardat least to test it out16:47
kbringardman, it's cold, I need typing gloves16:47
*** colinnich has left #openstack16:47
kbringardBK_man: I used that path because it gets created with nova16:48
kbringardin truth I'm not sure why it's not set as the default16:48
Nick0lapatri0t: is there any documentation anywhere on how nova hands out addresses?  Lease times, ability to hand out static IP's based on MAC, etc...?16:48
dspanokbringard: That did the trick.16:48
kbringarddspano: cool16:48
Nick0laI've use dhcpd frequently.  Never used dnsmasq16:49
BK_mankbringard: from my point of view casual users should NOT use source code to test nova. Thay should use packages. So, it's a packaging issue16:49
kbringardNick0la: generally you'll want to let dnsmasq handle it... but I think the leases are written out to /var/lib/nova/networks/16:49
kbringardI don't know that there is formal documentation on it, but the quick and dirty is that16:50
kbringardit writes out the info about the instance to the lease file16:50
kbringardand then the instance get's the correct info when it does a dhcpdiscover16:50
kbringardI think it defaults to just the mac address16:50
kbringard120s lease time16:50
Nick0laokay. i'm setting up a test/proof of concept for my company.16:51
kbringardand the instance id as rthe hostname16:51
Nick0lato offer rackspace-like offerings to our customers.16:51
kbringardone thing that doesn't get set is the domain name (so like, hostname -f won't work)16:51
Nick0laand we traditionally manage IP space very very carefully.  :-)16:51
Nick0laokay.  and it will respect that the first address in the range is the gateway?16:52
kbringardNick0la: it pulls a free IP from the fixed_ips table in the nova DB16:52
*** purpaboo is now known as lurkaboo16:52
kbringardcheck the fixed_ips table in the DB16:52
kbringardit's got the netmask, broadcast, gateway16:52
Nick0laone last question...16:53
kbringardalthough, I've only used VLAN mode... but I believe it is the same for the FlatDHCP mode16:53
Nick0lahehe I was going to use vlans, except we are an ISP, so the doleing out of vlans is strictly prohibited unless it passes through our guru first.16:54
*** fysa has quit IRC16:54
Nick0laflatdhcp seemed the best alternative16:54
kbringardyea, I think it works mostly the same way, I was just giving you the caveat :-D16:54
Nick0lahow do I add multiple IP ranges?  On one directive in nova.conf, or multiple directives?16:55
kbringardyou mean to be handed out by DHCP?16:55
Nick0la--fixed_range=,    etc16:56
kbringardhmmm, well, in VLAN mode, when you add a new network using nova-manage16:56
Nick0laI have a small range for testing, but when it goes live, I'll need to be able to add more.16:56
kbringardit inserts all the IPs into the fixed_ips table16:56
kbringardI don't think the fixed_range gets used by nova16:56
Nick0la?  Really?16:57
*** adiantum has joined #openstack16:57
kbringardsomeone with more intimate knowledge of it may correct me16:57
Nick0laI thought that was the purpose of the fixed_range directive..16:57
kbringardit is if you're using pure dnsmasq16:57
kbringardfor dynamic assignment16:57
kbringardbut nova uses static-dynamic assignment16:58
*** gondoi_ has quit IRC16:58
kbringardit dynamically generates the leases file16:58
kbringardwhich causes dnsmasq to statically assign the IP to the corresponding MAC addy16:58
kbringardyou'll want to look at the nova-manage network command16:58
kbringardI think it's like16:59
Nick0lahmmm  okay.16:59
*** gondoi_ has joined #openstack16:59
kbringardnova-manage network create CIDR number_of_networks number_of_ips17:00
kbringardso like17:00
kbringardnova-manage create 1 25617:00
kbringardwould create an entry in the networks table for
kbringardit would make it just one network17:00
kbringardnd would put 256 addresses in the fixed_ips table17:00
*** gondoi_ has quit IRC17:00
*** vvuksan has joined #openstack17:01
kbringardthen, you'll need to go into the fixed_ips table and manually set reserved = 1 on any IPs that are used by other stuff (like routers, etc)17:01
kbringardso nova won't assign those17:01
kbringardthen finally, in the networks table17:01
vvuksanlink to compute release note => http://http//
vvuksanis broken17:01
kbringardmake sure everything is happy in the networks table17:01
kbringardNick0la: getting networks setup initially still requires some work, in my experience sadly17:02
vvuksanactually link on the page says http// instead of http://17:02
kbringardvvuksan: howdy friend17:02
vvuksankbringard: howdy :-)17:02
Nick0laso if I were to use 2 128  it would create 2 networks ( and with 128 addresses each?17:02
kbringardin theory, yes17:02
Nick0laokay.  I was wondering why I needed to put in the number of networks and addresses.  That all seemed pretty self evident with the CIDR.17:03
kbringardI think it has to do with out it splits up the addresses to add them to the fixed_ips table17:04
Nick0la...and again...the number of IP addresses seems pretty self evident, as well.17:04
*** fysa has joined #openstack17:04
Nick0la10.1.1.128 is only gonna have 126 hosts available to it.  Basic networking.  I'm wondering if there is another reason it needs to be explicitely set on the commandline?17:05
kbringardwell, it breaks it up automagically based on the number of IPs you ask for, so you may have to fiddle with that number a bit17:06
*** MarkAtwood has left #openstack17:06
*** Ryan_Lane has joined #openstack17:08
Nick0lahmmm.  I'm still missing something, I think.  What does it do with the IP's that it doesn't allocate?17:08
Nick0lafor example: 1 12817:08
Nick0lawhat happens to all the extra IPs?17:08
kbringardnothing, they just never get allocated17:09
kbringardbecause they're not in the fixed_ips table17:09
kbringardthe fixed_ips table and the networks table are gospel to nova17:09
kbringardthe nova-manage tool is just a way to get data in there quickly17:09
kbringardbut ultimately it only cares about what's in those 2 tables, when it comes to allocating IPs and configuring the network17:09
Nick0laokay.  Would it break it terribly to manipulate that database directly?   should it go through controls/sanity checking first?  Is the database accessed for every request or does nova read and cache the contents?17:10
vvuksankbringard: since you are now Nova expert :-) you can perhaps help me17:10
kbringardpersonally, I use the nova-manage tool to put the skeleton info in there17:11
Nick0la(sorry for the 2000 questions -- I just wanna understand it better)17:11
kbringardNick0la: no worries, I like to help, but I don't always know everything, so sorry if I don't do a good job explaining it :-/17:11
kbringardbut yea, I create the basic info with nova-manage and then I'll go in and modify it manually to make it fit my environment17:11
kbringardvvuksan: sure thing boss, what's up?17:12
Nick0lakbringard: We can't all be experts on everything.  :-)17:12
Nick0laonly my 3 year son knows everything.  Or so he claims.17:12
kbringardlol, yea, same with my daughter17:12
vvuksanNick0la: he's right :-)17:12
vvuksanalthough in my case I'm informed that my help is not needed as she know everything :-)17:13
Nick0lamay I run a network design past you?  for sanity?  I'm very very very new to nova, and I don't wanna build it wrong.  (and I have a deadline of getting a working system up by the end of June)17:13
vvuksankbringard: anyways I'm trying to solve my problem where I don't control my network infra and don't have access to VLANs17:14
kbringardvvuksan: lol, back in that boat eh?17:14
Nick0lavvuksan:  you and me too!17:14
vvuksankbringard: Say I have a host machine and with it a set of IPs17:14
vvuksan8 IPs17:14
vvuksanalthough I can get more17:14
*** dendro-afk is now known as dendrobates17:14
vvuksanso the way I did it which works to an extent17:15
winston-dBK_man : hi17:15
vvuksanand btw this is all for internal use17:15
BK_manwinston-d: hi17:15
vvuksanno public IPs involved17:15
vvuksanso on each node I set up compute, network17:15
vvuksaneach node has it's own network range ie. 192.168.x.0/2417:15
*** blamar has quit IRC17:16
vvuksanthat is because you can have only one nova-network per network range17:16
winston-dyou mentioned that 'network-injection' is there for RHEL like guest.  what does that mean?  I have similar issue with
vvuksanso that works17:16
*** blamar has joined #openstack17:16
vvuksanthen what I have done is precreated NAT rules17:16
kbringardvvuksan: why don't you just use one nova-network machine and create 8 different networks on it17:17
kbringardto simplify your routing paths?17:17
vvuksanbasically on each node I built <=> , <=>
vvuksankbringard: how does that help me though since say two nodes only share eth0 interfaces17:17
kbringardor is that because each interface is hardwired to a single subnet?17:17
zigo-_-Oups, intended for channel... :)17:17
vvuksani mean eth0 are on the same 10.0.x.x network17:18
kbringardzigo-_-: you're having misfire issues today dude :-p17:18
zigo-_-My unit tests are failing here on my test server. We believe this is because of missing build-depends. alekibango got it to work on his Debian, and we managed to make a diff of packages he got installed: Does any of the core team dev know what package in this list are missing from my test server???17:18
kbringardvvuksan: ah, ok, then ignore my previous comment :-D17:18
alekibangoits not working yet, just tests passed and packages built17:18
zigo-_-kbringard: I'd like to be able to use a better client than BitchX, but I get logged out all the time if I do from my box.17:18
vvuksankbringard: the question is how can I limit number of guests that will run on aparticular node17:18
zigo-_-So I just IRC from shell on a server ...17:19
alekibangozigo-_-: show me your log...17:19
vvuksansince e.g. I got only .1-.8 that is attached to host1, .9-.15 to host217:19
vvuksankbringard: am I making sense ?17:19
kbringardvvuksan: hmmm, that's a good question... I would say projects are a good way to go... but then you run into VLAN issues17:19
zigo-_-alekibango: Hang on...17:19
zigo-_-Then BitchX isn't great...17:19
kbringardzigo-_-: try irssi17:20
*** freeflying has quit IRC17:20
alekibangoyes, irssi is very good17:20
BK_manwinston-d: on what version?17:21
alekibangoi consider using it as this kde application is taking huge ammounts of memory17:21
zigo-_-It doesn't have color, hilighting, multi-window, shortcuts, etc.17:21
zigo-_-In fact, I should just setup an IRC bouncer...17:21
*** freeflying has joined #openstack17:21
winston-dBK_man: i'm using your package, 2011.1.1-417:21
BK_manwinston-d: consider upgrading to Cactus17:22
zigo-_-test_get_foxnsocks ERROR17:22
BK_manwinston-d: I released packages couple hours ago17:22
Nick0lavvuksan:  May I ask?  Why a different network range for each host machine?  wouldn't you have the same ranges over all the hosts so if you have to move a guest from one node to another, it won't have to change IP addresses?17:22
zigo-_-test_bad_login_both_bad ERROR17:22
*** antenagora has quit IRC17:22
Nick0lavvuksan:  I am so not experienced with nova, so I hope that is not a silly question....17:23
vvuksanNick0la: it's not a silly question. Problem is that what I'm trying to do is perhaps not something Nova was designed for17:23
vvuksanNick0la: so I'm trying to work around it17:24
Nick0lavvuksan: ahhh.  okay.  :-)  What are you trying to accomplish?17:24
* Nick0la is curious17:24
zigo-_-Is it fine to run the unit tests on a server that is already running nova ?17:24
vvuksani don't manage my network infrastructure. Only machines that run on top of it17:24
vvuksanit's all internal private stuff17:24
vvuksanso I can get any number of IPs but getting my own VLAN would be tough17:25
vvuksanplus there are other logistical issues17:25
winston-dBK_man : do you mean that your Bexar release doesn't support 'network injection'?  I'd love to upgrade the Cactus but I prefer fixed the issues I met with Bexar first.17:25
vvuksanso I figured that I could simply on each host run my own private network in e.g. 192.168.x.0/24 range then NAT to it17:25
vvuksanfrom the public interface17:25
Nick0lavvuksan: I understand.17:26
vvuksanthat way it's fairly "clean"17:26
BK_manwinston-d: Bexar version is supporting network injection too17:26
vvuksani have concept in place17:26
*** Gaelfr has joined #openstack17:26
vvuksantrouble is I need a way to limit how many VMs get kicked of on the machine17:26
BK_manwinston-d: you can setup it in MySQL db (networks table)17:26
winston-dBK_man : do I have to do anything to enable that or it was already default?17:26
vvuksani would e.g. add static NAT mapping17:26
Nick0lavvuksan: I still do not undertand the need for a differnt subnet on each host.  :-)17:26
winston-dBK_man : ok, i'm listening.17:27
Nick0laI am in the same boat as you.  I cannot use VLAN's.17:27
BK_manwinston-d: there is a column and you should set it to 1 for your particular network. It's off by default17:27
vvuksanNick0la: you could bridge directly on eth017:27
vvuksanhowever that is fraught with peril17:27
Nick0laour network guru would have me fired instantly if I even thought seriously about adding a vlan without 3 weeks of change control.17:27
vvuksancause you need to specify a default gw17:28
Nick0laI am planning on using dhcp and bridges to do my network.17:28
vvuksanand nova-network unless you are really careful will try to assume the default gw IP17:28
Nick0laI have a single range of external IP's that have been granted to me (after much wheedling and no small amount of bribery)17:29
winston-dBK_man : which column?17:29
Nick0laand I'm gonna let all my hosts use that range and let DHCP dish them out.17:29
BK_manwinston-d: could you please dump your iptables and ebtables config and sent it to me for an analysis?17:29
winston-dBK_man : oh i see, the 'injected'17:29
vvuksanNick0la: you don't have any other DHCP servers on that network segment ?17:29
Nick0la...that's the plan, anyways.17:29
*** reldan has quit IRC17:29
BK_manwinston-d: we are unable to reproduce dnsmasq bug in our lab17:29
vvuksanNick0la: trouble is I can't use DHCP on the public network17:30
vvuksansince there is one there already17:30
Nick0lathe IP range I was given is in a specific VLAN set up for me specifically for this project.  Same with the IP's, so I have complete control over DHCP.17:30
*** Gaelfr has quit IRC17:30
vvuksanthat's good17:30
Nick0lavvuksan:  ugh!  that sucks.17:30
winston-dBK_man : sure. but how can i make sure the problem i met is caused by 'dnsmasq'?  what iptable/ebtables config do you want ? the one on compute node or on network node?17:31 can control dhcp for your 192.blah addressing, yes?17:31
vvuksansince I run my own "walled" off bridge :-)17:32
Nick0laso...why do you not give all the nodes access to the entire /24?17:32
vvuksanyou can't17:33
Nick0lait is a nova limitation? or just the way you wanna do it?17:33
vvuksanit's a networking limitation17:33
Nick0laah.  okay.17:33
vvuksanunless I span my own virtual network17:33
vvuksanacross all my compute nodes17:33
Nick0laI understand now.17:34
Nick0layah...that could be..problematic.17:34
vvuksanunfortunately it's a tricky issue :-(17:34
vvuksanperhaps not soo tricky17:34
*** bcwaldon has joined #openstack17:34
vvuksanwhat is tricky is telling Nova to dish out only X IPs in range17:34
vvuksani will have to play with it17:34
Nick0laI was curious because isn't the benefit of nova that if one host dies, the other hosts can pick up the slack?17:35
*** mahadev_ has joined #openstack17:35
Nick0laand if the hosts each hand out different IP ranges, it would be difficult to migrate a machine to a new host.17:35
*** mahadev has quit IRC17:35
Nick0lamigrate a 'guest', sorry.17:35
Nick0lamy clients are most certainly gonna require a static IP address.17:36
Nick0laso making sure their IP moves with the guest is very important to me.17:36
*** reldan has joined #openstack17:38
Nick0lawhat does this directive mean:  FAKE_subdomain=ec217:38
BK_manwinston-d: from both if you can17:39
kbringardvvuksan: sorry, got pulled away17:39
vvuksankbringard: no worries17:39
kbringarddo you know if you're network gear passes like, vlan017:39
kbringardas "untagged"17:39
winston-dHey guys.  It took quite long for Nova to do 'networking' for a newly created instance here.  Is that normal?  How long does your Nova take?17:39
kbringardmaybe you could create projects17:39
kbringardassign them all to the same "untagged" vlan17:39
kbringardthen you just launch instances in each project17:40
kbringardsince IPs are allocated to projects17:40
winston-dBK_man : OK.  I've just altered the database and am trying to launch a new instance to see if networking issue is gone.17:40
BK_manwinston-d: I need to go home, it's late evening in Russia. Please send you logs to abrindeyev@griddynamics.com17:40
kbringardso you basically have one project per host17:40
vvuksankbringard: hmm17:40
kbringardand when that host is out of IPs, if more try to get launched in that project, they'll just get stuck in networking17:41
* BK_man off to home17:41
*** rds__ has joined #openstack17:41
winston-dBK_man : ok.  It's already 1:40 in the morning here. :) I'll send the logs to you.17:41
*** elasticdog has quit IRC17:41
vvuksankbringard: looking17:41
*** BK_man has quit IRC17:41
vvuksankbringard: you got a link to projects docs ?17:43
*** kashyap has joined #openstack17:44
kbringardthat's the admin manual17:44
vvuksanthat may not cut it for me17:44
vvuksansince projects are associated with users17:45
kbringardI think you can have users in multiple projects17:45
kbringardthe only thing is, you have to add :project_name at the end of the API key17:45
kbringardwhen making your call17:45
*** reldan has quit IRC17:45
kbringardso you'd probably need to write some middleware to manage that17:46
kbringardI'll keep thinking too... that's just one idea that comes to mind17:47
kbringardnot ideal I know :-/17:47
*** bcwaldon_ has joined #openstack17:49
*** adiantum has quit IRC17:49
dspanoCan you only allocate floating ips with the VLAN manager?17:49
dspanoI'm using flatDHCP.17:49
kbringardyou should be able to allocate floating to flatdhcp17:50
*** hagarth has joined #openstack17:52
*** mramige has joined #openstack17:53
*** mgoldmann has quit IRC17:54
*** mgoldmann has joined #openstack17:54
Nick0lahmmm....what does this directive mean:  --routing_source_ip17:55
Nick0la(in the nova config)17:55
*** elasticdog has joined #openstack17:55
*** elasticdog has joined #openstack17:56
Nick0laoops...nevermind...I just found the comprehensive list...17:59
winston-dkbringard : how long does your Nova take to boot a new instance?17:59
vvuksanwinston-d: mine come up quickly17:59
vvuksan30 seconds17:59
Nick0lahrm...nevermind my 'nevermind'.  --routing_source_ip shown in an example is not on the comprehensive list.....18:01
winston-dvvuksan : wow, really?  it takes almost 3 mintues for a guest with 5GB image18:01
winston-dvvuksan : and from both euca-describe-instance and nova-compute log, 'networking' stage took almost half of the time18:02
*** mahadev_ has quit IRC18:02
winston-dvvuksan : scheduling is quick, networking takes forever, launching is OK. altogether, it takes forever to boot a instance.18:03
vvuksandoesnt' seem right18:03
vvuksanwhat you could do is18:03
*** mahadev has joined #openstack18:03
vvuksanwatch the console log for the instance18:03
vvuksanto see what it's doing18:03
winston-dwell, by the time console log appears, it already passes the 'networking' stage.18:04
winston-d'networking' is compute node building up VLAN interface, bridge and stuff, then setting up IPTALBES rules, etc.18:06
vvuksani'm not using VLAN18:08
vvuksanso I may not be seeing the issue you are seeing :-/18:08
winston-dthat's possible. hmmm18:08
*** stewart has quit IRC18:10
*** stewart has joined #openstack18:11
dspanokbringard: Weird. I'm getting this error. File "/usr/lib/pymodules/python2.6/nova/db/sqlalchemy/", line 444, in floating_ip_allocate_address raise db.NoMoreAddresses()18:11
*** syah_ has quit IRC18:12
*** rcc has quit IRC18:15
*** syah has joined #openstack18:16
uvirtbotNew bug: #762047 in nova "Several modules don't pass pep8" [Undecided,New]
zigo-_-FYI: I have uploaded the Cactus release for Debian at (deb openstack main)18:17
zigo-_-That's only Nova for the moment.18:17
rykerif anyone going to the conference needs a ride from the SF airport to their hotel, I will be arriving at SFO on the 25th @10:45am.  Just email me to set something up: john.m.alberts -at- gmail.com18:20
*** antenagora has joined #openstack18:24
*** clauden has joined #openstack18:25
*** stewart has quit IRC18:35
*** zigo-_- has quit IRC18:38
*** joearnold has quit IRC18:38
Nick0lamy photography gear shipped this afternoon!18:39
Nick0lawrong window...sorry18:40
*** mahadev has quit IRC18:40
*** mahadev has joined #openstack18:41
kpepplecan someone tell Stephen Spector that his link to the Cactus release announcements in his Community Weekly Newsletter is 404 ?18:44
kbringarddspano: do you have a project associated with the floating ip?18:44
kbringarddid we talk about this already? My brain is mush today18:44
dspanokbringard: I figured it out. I was being a dummy. I thought the hostname argument was for injecting the hostname in to the instance.18:45
kbringardhehe, not being a dummy, it's not super clear18:45
dspanokbringard: Once I changed the hostname to cld01, it worked. Is there a way to inject the hostname?18:45
kbringarddspano: naw, I think the idea is that the floating_ip is an "external" IP that is managed in your out of band (to nova) DNS18:46
uvirtbotNew bug: #762071 in nova "Several modules use "== None" instead of "is None"" [Undecided,New]
*** elambert has quit IRC18:52
*** joshfng has joined #openstack18:52
dspanokbringard: Do you know why it would raise db.NoMoreAddresses here: /usr/lib/pymodules/python2.6/nova/db/sqlalchemy/", line 620, in fixed_ip_associate_pool18:57
kbringarddspano: in my experience that happens when they're not associated with the project you're authenticating as...18:58
kbringardif you're not using projects then it may be a bug :-/18:58
*** reldan has joined #openstack18:59
*** joshfng has quit IRC19:00
kbringarddoes anyone know if the XML returned changes format if there are more than X number of instances running?19:01
kbringardfor euca-describe-instances?19:01
kbringardI'm seeing a weird error where my tests start failing if there are more than a certain number (which I have not pinned down yet)19:01
annegentle+1 for Gina Trapani being a Sphinx promoter:
*** dragondm has quit IRC19:04
*** Ryan_Lane is now known as Ryan_Lane|food19:10
*** photron has joined #openstack19:10
*** dragondm has joined #openstack19:14
*** jkoelker has quit IRC19:14
*** kashyap has quit IRC19:15
*** allsystemsarego has quit IRC19:22
dspanokbringard: I got it to work. I needed to delete the network that the script created and create a new one for my test project.19:25
dspanokbringard: Thanks for you help man.19:26
kbringardah, good work and np19:26
*** dodeeric has quit IRC19:27
*** camm has quit IRC19:27
kbringardfor anyone with kids and an i* device:
*** mgoldmann has quit IRC19:28
kbringardor, I guess if you just want to brush up on your basic arithmetic19:29
vvuksankbringard: btw I recommend
kbringardah, I don't have an iPad yet19:31
kbringardbut I'll bookmark that for when I do :-)19:31
alekibangokbringard: it seems to be very easy19:31
alekibangomaybe for my almost 3 yo daughter :)19:32
alekibango5yo can count better19:32
kbringardhehe, yea, I've a 2.5 year old and a 5m old19:32
kbringardso they're perfect for my kid19:32
kbringardI should have said "young" kids19:32
alekibangostill, no eyedevice19:32
kbringardhehe, you're probably better off19:33
alekibangohow far away from head do you hold it when talking?19:33
kbringardok, so dspano got me thinking... what should go in the host section of the floating_ips?19:34
alekibangokbringard: i will try to make my sons (5,7)  write similar games...19:34
kbringardis that the host where the network controller lives?19:34
kbringardalekibango: hah, I'll beta test if they do19:34
alekibangovery nice images btw19:34
alekibangokbringard: try gvrng19:36
alekibangomy kids love that.. and gcompris... tuxpaint...  and (heh) openttd and warzone210019:36
alekibangoeven the 5yo can play that!19:37
*** kakella has joined #openstack19:39
*** kakella has left #openstack19:39
*** rsaidan has quit IRC19:40
*** rsaidan has joined #openstack19:41
dspanoTo whomever put up the new documentation. Thank you.19:46
kbringardthat's probably annegentle19:46
kbringardshe is the awesome keeper of the docs19:47
*** ctennis has quit IRC19:49
*** Zangetsue has quit IRC19:49
*** Zangetsue has joined #openstack19:50
*** joearnold has joined #openstack19:52
*** MarkusT has joined #openstack19:53
MarkusTI can't seem to find an simple tutorial on how to modify (update + minor changes) an existing image. Is it really that complicated? My main problem involves regular updates of the instances, how do you do that easily?19:54
kbringardMarkusT: that's something I'm struggling with myself19:56
kbringardwhat I've resorted to doing is19:56
kbringardlaunch the instance19:56
kbringardmake the changes19:56
kbringardlog into the compute node (the host the instance was running on)19:56
kbringardshutdown -h the VM (from within the VM)19:57
kbringardthen cd /var/lib/nova/instances/$instance_id19:57
kbringardbundle it (if you use euca-bundle, etc) or glance-upload it19:57
kbringardthen launch it19:58
kbringardthe new one19:58
*** BK_man has joined #openstack19:58
kbringardnot the cleanest, but it works19:58
kbringardthen you can virsh start $instance_id and it should fire the VM back up19:58
kbringardor you can euca-terminate-instances it19:58
* annegentle shakes fist at wiki spammers19:59
*** iammartian has joined #openstack19:59
MarkusTkbringard: I'm an idiot. I tried "that" (at least thought so), but just copied the original image instead of the instance image. Will test it (and report back), thanks!19:59
annegentledspano: no, thank you for noticing! :)19:59
kbringardMarkusT: make sure you're uploading disk, and not disk.local19:59
kbringardalso make sure you link the kernel and ramdisk back up (if necessary)19:59
iammartianN<, already deleted it20:00
*** brd_from_italy has joined #openstack20:00
kbringardgenerally I just use the same kernel I launched it with originally, unless a kernel update is part of what you're trying to update20:00
kbringardin which case it's a bit more complicated20:00
*** ctennis has joined #openstack20:10
kbringardiammartian: how am I going to find out about how nova can help me with payday loans now?20:12
*** koolhead17 has quit IRC20:18
*** grapex1 has joined #openstack20:29
*** grapex has quit IRC20:32
*** jkoelker has joined #openstack20:33
*** bcwaldon_ has quit IRC20:34
*** bcwaldon has quit IRC20:34
MarkusTkbringard: Might be a dumb question, but if I run "euca-upload-bundle -b mybucket -m /tmp/disk.manifest.xml" it seems to work fine, but doesn't show up on describe-images. What am I doing wrong? :-)20:34
kbringarddid you euca-register?20:35
kbringardwhen you upload it should give you an id20:35
kbringardthen you have to euca-register $id20:35
kbringardbefore it'll show up in describe-images20:35
*** dendrobates is now known as dendro-afk20:52
MarkusTkbringard: Thanks, I did not register it.  A (last) one: I'm now able to start the instance, but can't log in (and can't get any console output). nova-compute.log says: "ignoring error injecting data into image ami-[...] (Mapped device was not found (we can only inject raw disk images): /dev/mapper/nbd15p1". Any ideas what is causing this?20:53
vvuksanyou can ignore that20:54
MarkusTvvuksan: O.k., but I still can't log in :-)20:54
kbringardyea, at this point it get's tricky20:54
kbringardwhat network mode are you using?20:54
MarkusTkbringard: FlatDHCP20:54
kbringardah, OK20:55
vishyuksysadmin: euca-allocate-address before euca-associate-address20:55
kbringardMarkusT: is your image set to get it's network info via DHCP (and can you ping the IP nova allocated it?)20:55
MarkusTkbringard: I was using a Ubuntu 10.04 image and only updated it. I'm not aware of any changes, which might cause it to fail now. The original image still works fine.20:56
kbringardcan you ping it?20:57
vishyMarkusT: injection isn't expected to work for whole disk images20:57
kbringardvishy is the man, if anyone can help you, it's him20:57
MarkusTkbringard: Thanks for helping me this far!20:58
kbringardhehe, you're welcome20:58
kbringardI have to go to a meeting, I'll bbl20:58
kbringardI would say it's probably one of 2 things... either it's not getting an IP20:58
kbringardor the credentials aren't being pulled correctly20:59
MarkusTvishy: I'm just trying to figure out how to update (and modify) an image. kbringard helped me to copy (and republish) the running instane, I just can't ssh back into the modified image20:59
kbringardyou can also check the compute node it started on, and look for the VNC port that was assigned to this instance (it starts at 5900 and goes up one for every instance)20:59
vishyhow did you update it? bundle and upload?21:00
kbringardand VNC to the instance to see if it hung somewhere and didn't start21:00
*** dendro-afk is now known as dendrobates21:00
MarkusTvishy: Yes. bundle, upload, register, run21:00
MarkusTvishy: Or complete: Run original instance, update, shutdown -h, bundle, upload, register, run21:01
vishywas the original image a whole disk image?21:01
vishybecacuse it sounds like it is trying to boot without a kernel and ramdisk21:01
vishyare you using Glance? or local image service21:01
MarkusTIt's lucid-server-uec-amd64.img, so: no. I wgetted it and uploaded it to mybucket.21:02
kbringardseems likely that it's not linked to the original kernel and ramdisk, like vishy said21:02
MarkusTHow do I link it?21:02
vishyyou didn't give it a kernel and ramdisk?21:03
vishywell easiest way is to specify them all at once with nova-manage combined_register21:03
vishybut you could upload the others and manually edit the json21:03
MarkusTvishy: I just bundled the "disk" image of the updated instances. I was under the impression not to loose kernel and ramdisk, since it's just a modified disk image.21:04
vishywhere did you get this "disk" image?21:05
*** dmshelton has left #openstack21:06
uvirtbotNew bug: #762182 in nova "Instance_type_id refactor broke XS Resizes" [Undecided,New]
MarkusTvishy:  /var/lib/nova/instances/$instance_id from the (updated) (halted) instance.21:08
vishyyeah that won't work21:08
vishyunless you are in --nouse_cow_images_mode21:09
vishyif you want to snag the modified image from outside21:09
kbringardvishy: ok, sorry, that's my bad21:09
*** antenagora has quit IRC21:10
vishyyou need to snapshot it with qemu-img, then use the qemu-img snapshot convert command, and when you upload it, you need to specify the kernel and ramdisk id21:10
vishywow it is too bad snapshotting for kvm didn't make it in because this would be really easy :)21:10
MarkusTvishy: kbringard uses it :-). I'm just trying to find an easy way to modify and update. I got the suggestion for "--nouse_cow" which didn't work, does --nouse_cow_images_mode work with bexxar? Is there an easy tutorial on the issue? I tried dozens, but each time it fails :-)21:11
kbringardvishy: I was able to glance-upload the raw disk image from instances21:11
kbringardwhen I saw "raw disk image" i mean the qcow221:11
vishysorry it is --nouse_cow_images21:11
vishykbringard: sounds dangerous.  Might work if the backing file happens to be there already21:11
kbringardyea, I was having weird issues with it21:12
kbringardthe image would run fine on the compute node that had the glance server running on it21:12
kbringardbut not on any of the other compute nodes21:12
kbringardbut, it's a new image_id, so the backing image changed21:12
kbringardit was very odd21:12
kbringardso I'll just chalk it up to "that doesn't work" and find a different way to do this21:13
kbringardsorry MarkusT, didn't mean to give you bad info :-)21:13
MarkusTvishy: Do I need --nouse_cow_images when using qemu-img?21:13
MarkusTkbringard: No problem. :-)21:13
*** johnpur has quit IRC21:13
vishyMarkusT: no21:13
*** drico has joined #openstack21:13
vishykbringard: you're going to have chained backing images in that case21:14
kbringardah, that would explain the problem I was having21:14
vishykbringard: so the snapshotting feature which does this essentially does the following:21:14
*** elambert has joined #openstack21:16
*** h0cin has quit IRC21:18
vishyqemu-img snapshot -c snap <file>21:19
*** photron has quit IRC21:19
*** ppetraki has quit IRC21:21
vishyqemu-img convert -s snap <new_file>21:21
vishywill give you a combined image21:22
vishyyou need to use qemu-common 14 which is in the ppa21:22
*** Zangetsue_ has joined #openstack21:25
MarkusTvishy: I'm using ppa:nova-core/release, there's no qemu-common (as far as I an see). Which ppa are you referring to?21:26
MarkusTvishy: Can I mix both? (qemu-common from trunk and the rest release?). I'd love to have a stable system, since trunk did already cost me weeks of fiddling around so far :-)21:28
vishyright now they should be equivalent.  I don't know why qemu wasn't copied into release21:28
vishylet me look21:28
*** stewart has joined #openstack21:29
*** Zangetsue has quit IRC21:29
*** Zangetsue_ is now known as Zangetsue21:29
vishyshould be fine for the moment21:30
vishymaybe I'll see if there is a reason soren didn't copy stuff over21:30
kbringardvishy: awesome, thanks21:30
*** hggdh has quit IRC21:30
*** hggdh has joined #openstack21:31
*** drico has quit IRC21:31
*** kbringard has quit IRC21:34
*** keekz_ is now known as keekz21:36
*** ppetraki has joined #openstack21:37
*** antenagora has joined #openstack21:37
*** antenagora_ has joined #openstack21:37
*** antenagora has quit IRC21:37
*** antenagora_ is now known as antenagora21:37
MarkusTvishy:  I don't get it. Everytime I run "qemu-img convert -s snap output" it shows me the help page (no error message). Am I missing something. What I've done so far: Run instance, shutdown -h, cd instance directory, snapshot -c snap disk, and then trying to convert. Right? (Thanks for far for your help! :-))21:38
*** Zangetsue has quit IRC21:38
vishyMarkusT: what version does qemu-img show21:38
*** Zangetsue has joined #openstack21:38
MarkusTvishy: qemu-img version 0.13.9121:39
vishyyou need 1421:39
*** vvuksan has quit IRC21:39
MarkusTvishy: Thought I got all files from the ppa. Will check again.21:39
*** MarkAtwood has joined #openstack21:41
vishyMarkusT: my bad21:42
*** dspano has quit IRC21:42
vishyMarkusT: it is 13.9121:42
*** ppetraki has quit IRC21:42
vishythe command is qemu-img convert -s snap disk <new_file>21:43
sorenvishy: "Copy stuff over"? to the release ppa, you mean?21:43
vishysoren: yeah, there is a bunch of stuff in the trunk ppa that isn't in the release ppa21:43
vishyqemu, suds, etc.21:44
sorenvishy: Yeah, we opened diablo to soon. Before we copied over the packages.21:44
*** drico has joined #openstack21:44
sorenvishy: I've been reassembling the pieces (fixing up the versioning in the process).21:44
*** bcwaldon has joined #openstack21:44
*** bcwaldon_ has joined #openstack21:44
sorenvishy: It's in nova-core/tmp right now. I'll probably copy everything over in a minute.21:44
vishysoren: ah ok21:45
*** stewart has quit IRC21:46
*** dragondm has quit IRC21:48
*** reldan has quit IRC21:54
*** bcwaldon has quit IRC21:59
*** bcwaldon_ has quit IRC21:59
sorenvishy: The deed is done.22:00
vishysoren: coolness22:00
sorenThe publisher runs every 5 minutes, so it'll be another minute-ish before it's apt-get'able from there.22:04
*** elambert has quit IRC22:06
*** antenagora has quit IRC22:07
*** drico has quit IRC22:09
MarkusTvishy: The new image fails to start (pending). compute log says: "Error: internal error process exited while connecting to monitor: char device redirected to /dev/pts/1" "Two devices with same boot index 0"22:09
vishyMarkusT: how did you upload the new image?22:10
MarkusTvishy: "euca-upload-bundle -b mybucket -m /tmp/<myoutputfilename>.manifest.xml"22:11
vishythis might be a heck of a lot easier22:12
vishy./nova-manage image image_register <path> admin T x86_64 ami ami <kernel_id> <ramdisk_id>22:12
*** reldan has joined #openstack22:12
MarkusTvishy: It seems I fucked up something else. Now no image is able to start (with the same error). Argl, need to check that first. :-)22:15
*** jheiss has quit IRC22:15
*** reldan has quit IRC22:17
MarkusTvishy: Any chance this might be related to updating kvm-qemu after nova was already installed? It's the only "bigger" change I'm able to identify in history22:18
vishyi would make sure libvirt-bin is updated22:19
*** jkoelker has quit IRC22:19
vishyand restart libvirt-bin22:19
vishyrestart nova-compute22:19
*** dragondm has joined #openstack22:19
*** dragondm has joined #openstack22:20
MarkusTI already restarted the whole system. With updated you mean install libvirt-bin from nova/trunk?22:20
*** mahadev_ has joined #openstack22:27
*** jfluhmann has quit IRC22:29
MarkusTvishy: Yeeaaaah! It works! I can't believe it finally works! :-) I needed to update libvirt-bin to trunk. So I now hold my breath I won't experience problems with that later on. But for now it works and I'm finally able to modify images. Thanks for all your help! I really appreciate it! :-)22:29
*** jheiss has joined #openstack22:29
vishyMarkusT: np, hopefully we'll have snapshotting support soon so you can do that all with one command :)22:30
*** mahadev has quit IRC22:31
*** MarkAtwood has quit IRC22:33
*** mray has quit IRC22:38
*** dendrobates is now known as dendro-afk22:45
_vinaydoes lc-tools work with openstack ?22:52
_vinayanyone know if there is a openstack driver in lc-tools ? .. thx22:52
*** mramige has quit IRC22:56
*** rnirmal has quit IRC23:00
*** Zangetsue has quit IRC23:00
*** iammartian has left #openstack23:01
*** aliguori has quit IRC23:05
*** arun_ has quit IRC23:09
*** mahadev_ has quit IRC23:13
*** mahadev has joined #openstack23:14
vishy_vinay: never heard of lc-tools so doubtful23:14
vishyah looks like it is libcloud23:14
vishysince openstack has an ec2-api, you should be able to get it to work with openstack, but it might require a bit of configuration23:16
*** clauden has quit IRC23:17
_vinayyes libcloud ... sorry23:19
_vinayvishy... I see. will keep that in mind when I will try it.... thanks23:20
mirrorbox_vinay: I'm the author of lc-tools. Libcloud probably will not work with your openstack right of the box23:22
mirrorbox_vinay: as hostname is hardcoded23:22
*** shentonfreude has quit IRC23:22
mirrorbox_vinay: but once you change it I think it will work most likely. Let me know if you need any help23:23
_vinaywell I have yet to look at libcloud ... but when you say hostname is hardcoded.. thats in libcloud right?23:32
_vinaymirrorbox: and how do I change the hostname?23:33
_vinayI will probably ping you once I start working on it23:33
mirrorbox_vinay: yeah, it's hardcoded in libcloud23:33
mirrorbox_vinay: find . -name in libcloud dir and you will see it23:34
*** rsaidan has quit IRC23:36
_vinaythere are 4 hostname variables in there23:40
_vinayso I guess just change all of them to point to the one of my nova installation ... right?23:40
mirrorbox_vinay: yeah, try that and see how it goes23:41
_vinaycool.. thanks mirrorbox for the info.. will try that and let you know23:41
*** Ryan_Lane|food has quit IRC23:45
*** reldan has joined #openstack23:50
*** world_weapon1 has joined #openstack23:51
*** j05h has quit IRC23:52
*** world_weapon has quit IRC23:53
*** j05h has joined #openstack23:56

Generated by 2.14.0 by Marius Gedminas - find it at!