Thursday, 2010-12-02

*** dubsquared has quit IRC00:09
*** rsampaio has joined #openstack00:13
*** dfg_ has quit IRC00:25
*** jc_smith has joined #openstack00:26
*** hggdh has quit IRC00:35
*** rsampaio_ has joined #openstack00:35
*** rsampaio has quit IRC00:36
*** maplebed has quit IRC00:36
*** DubLo7 has joined #openstack00:37
*** rsampaio_ has quit IRC00:40
*** hggdh has joined #openstack00:44
*** johnbergoon has quit IRC00:45
*** johnbergoon has joined #openstack00:46
*** rsampaio has joined #openstack00:49
*** rlucio has quit IRC00:56
*** dragondm has quit IRC00:58
*** Cybodog has quit IRC01:03
*** KyleM1 has quit IRC01:07
*** kevnfx has left #openstack01:10
*** ArdRigh has joined #openstack01:14
*** schisamo has quit IRC01:22
*** zaitcev has joined #openstack01:27
*** abecc has joined #openstack01:39
*** abecc has joined #openstack01:39
*** MarkAtwood has quit IRC01:39
*** metoikos has joined #openstack01:50
*** gaveen has quit IRC02:04
*** miclorb_ has quit IRC02:09
*** HugoKuo has joined #openstack02:11
*** jc_smith has quit IRC02:12
*** jimbaker has joined #openstack02:34
*** laurensell has joined #openstack02:36
*** miclorb has joined #openstack02:37
*** dubsquared has joined #openstack02:37
dubsquaredvishy:  you get a change to try out that bridge hack yet?02:37
vishyno not yet02:43
*** Ryan_Lane is now known as Ryan_Lane|away02:47
*** JordanRinke1 has joined #openstack02:58
JordanRinke1whoever admins the openstack mailing list there is probably a ton of emails pending admin or something similar from me02:59
*** kevnfx has joined #openstack03:03
*** kevnfx has left #openstack03:04
*** miclorb has quit IRC03:06
*** hazmat has quit IRC03:06
*** hazmat has joined #openstack03:15
*** laurensell has quit IRC03:21
*** miclorb_ has joined #openstack03:24
*** abecc has quit IRC03:27
*** sirp1 has quit IRC03:39
*** ramkrsna has joined #openstack03:43
*** hazmat has quit IRC03:43
*** jimbaker has quit IRC03:46
*** ramkrsna_ has joined #openstack03:48
*** maplebed has joined #openstack03:56
*** dendrobates is now known as dendro-afk04:01
*** ramkrsna_ has quit IRC04:04
*** sirp has joined #openstack04:09
*** maplebed has quit IRC04:17
*** jimbaker has joined #openstack04:18
*** ghaskins has joined #openstack04:31
ghaskinshi all, trying to walk through the install-tutorial and hitting this:04:33
ghaskinsany pointers appreciated04:33
*** kashyapc has quit IRC04:37
*** ArdRigh has quit IRC04:40
*** joearnold has quit IRC05:01
*** laurensell has joined #openstack05:05
*** rsampaio has quit IRC05:09
*** joearnold has joined #openstack05:09
*** eldarnugaev has joined #openstack05:11
*** kashyapc has joined #openstack05:13
*** laurensell has quit IRC05:21
*** eldarnugaev has quit IRC05:24
*** laurensell has joined #openstack05:30
*** joearnold has quit IRC05:39
*** eldarnugaev has joined #openstack05:47
*** joearnold has joined #openstack05:48
*** f4m8_ is now known as f4m805:48
*** metoikos has quit IRC05:52
*** sirp has quit IRC05:57
*** joearnold has quit IRC06:04
*** jimbaker has quit IRC06:18
*** jbaker has joined #openstack06:18
*** miclorb_ has quit IRC06:20
*** MarkAtwood has joined #openstack06:45
*** ramkrsna has quit IRC06:57
*** matiu has quit IRC07:12
*** miclorb_ has joined #openstack07:28
*** zaitcev has quit IRC07:43
*** ramkrsna has joined #openstack07:44
*** miclorb_ has quit IRC08:02
*** GasbaKid has joined #openstack08:11
*** arthurc has joined #openstack08:18
*** almaisan-away is now known as al-maisan08:21
*** calavera has joined #openstack08:30
*** ibarrera has joined #openstack08:40
*** eldarnugaev has quit IRC08:43
*** michaeldreamhost has quit IRC08:43
*** allsystemsarego has joined #openstack08:54
*** irahgel has joined #openstack09:22
*** eldarnugaev has joined #openstack09:38
*** miclorb_ has joined #openstack09:40
*** miclorb_ has quit IRC09:55
*** ar1 has quit IRC10:03
*** krish has joined #openstack10:05
*** krish has left #openstack10:05
*** MarkAtwood has quit IRC10:08
*** GasbaKid has quit IRC10:22
*** calavera_ has joined #openstack10:23
*** GasbaKid has joined #openstack10:24
*** 16SAARQT8 has joined #openstack10:38
*** calavera has quit IRC10:53
*** calavera_ is now known as calavera10:53
*** Cromulent has joined #openstack10:57
* soren lunches11:10
*** GasbaKid has quit IRC11:13
*** GasbaKid has joined #openstack11:13
*** Cromulent has quit IRC11:23
*** kashyapc has quit IRC11:40
*** KnightHacker has quit IRC11:54
*** KnightHacker has joined #openstack11:58
*** KnightHacker has joined #openstack12:00
*** dendro-afk is now known as dendrobates12:05
*** DubLo7 has quit IRC12:13
*** kashyapc has joined #openstack12:35
*** Cybodog has joined #openstack13:04
*** dendrobates is now known as dendro-afk13:10
*** gundlach has joined #openstack13:14
*** abecc has joined #openstack13:19
*** abecc has joined #openstack13:19
*** laurensell has joined #openstack13:25
*** Cromulent has joined #openstack13:28
*** laurensell has quit IRC13:29
*** ramkrsna has quit IRC13:35
*** GasbaKid has quit IRC13:37
*** matclayton has joined #openstack13:41
*** Cromulent has quit IRC13:44
*** ppetraki has joined #openstack13:49
*** aliguori has joined #openstack13:50
*** westmaas has joined #openstack13:54
*** gondoi has joined #openstack13:54
*** gondoi has quit IRC13:59
*** piken has joined #openstack14:00
ghaskinsrepost: any suggestions on whats wrong here:
sorenghaskins: You're missing the "nova-manage network create" step.14:01
ghaskinssoren: ah, thx!14:01
ghaskinswill try14:01
sorenghaskins: Somethng like this:14:01
*** gondoi has joined #openstack14:02
sorenghaskins: nova-manage network create 1 6414:02
ghaskinssoren: worked like a champ, thanx!14:03
sorenghaskins: np14:03
* ghaskins continues the tutorial14:03
*** rsampaio has joined #openstack14:06
ghaskinssoren: I assume this error means I probably have an incompatible version of boto installed?
*** Cromulent has joined #openstack14:09
*** hazmat has joined #openstack14:11
ghaskinsdoes anyone know what version of boto openstack has been known to work with14:12
ghaskinsi am running 2.0 atm14:12
ghaskinsacually 2.0b3 to be exact14:14
ghaskinswhat does Ubuntu ship?14:14
sorenBoto 2.0 itself is fine.14:16
soreneuca2ools with boto 2.0... Not so much.14:16
ghaskinssoren: ah, ok..ill downgrade the package to 1.9b then14:17
ghaskinsany particular version of euca2ools I should have, as well?14:17
ghaskinsi seem to be running euca2ools-1.3.114:19
*** Cromulent has quit IRC14:19
sorenghaskins: That should be fine.14:21
sorenghaskins: In your honour:  :)14:22
uvirtbotLaunchpad bug 684179 in nova "Having to run "nova-manage network create" is not easily discoverable" [Undecided,New]14:22
ghaskinssoren: re: euca, ok, re: bug, thx!14:22
sorenghaskins: You're far from the first person to stumble on that one. We should have fixed it long ago.14:23
uvirtbotNew bug: #684179 in nova "Having to run "nova-manage network create" is not easily discoverable" [Low,In progress]
*** jdarcy has joined #openstack14:47
*** f4m8 is now known as f4m8_14:48
*** GasbaKid has joined #openstack14:49
*** spectorclan has joined #openstack14:50
*** crago has joined #openstack14:51
*** crago has quit IRC14:54
*** crago has joined #openstack14:55
*** iammartian has quit IRC14:56
*** GasbaKid has quit IRC14:56
*** crago has quit IRC14:56
*** iammartian has joined #openstack15:01
*** Podilarius has joined #openstack15:05
*** gundlach has quit IRC15:06
*** dubsquared1 has joined #openstack15:11
annegentlesoren: do you run "nova-manage network create" directly before creating a project? I will add it to the install instructions (it's not in there)15:13
dubsquared1— /usr/bin/python /usr/bin/nova-manage user admin <user_name>15:15
dubsquared1—   /usr/bin/python /usr/bin/nova-manage project create <project_name> <user_name>15:15
dubsquared1—   /usr/bin/python /usr/bin/nova-manage network create <cidr> <num_networks> <total_ips>15:15
dubsquared1thats from my dox15:16
annegentlesoren: well, that step is not in the single install, but it is in the multi-server install steps15:16
dubsquared1— /usr/bin/python /usr/bin/nova-manage user admin admin15:17
dubsquared1—  /usr/bin/python /usr/bin/nova-manage project create demo admin15:17
dubsquared1— /usr/bin/python /usr/bin/nova-manage network create 1 23015:17
dubsquared1for example15:17
ttxannegentle: we should revamp the "install on Ubuntu" instructions based on the current PPA. I posted to that effect15:17
annegentlettx: thanks for that!15:17
sorenannegentle: It just needs to be run before nova-manage project zipfile.15:18
sorenannegentle: It doesn't depend on users already having been set up.15:18
*** gundlach has joined #openstack15:18
sorenannegentle: I've put in a patch that spits something like this out:15:18
sorenNo more networks available. If this is a new installation, you need15:19
sorento call something like this:15:19
sorennova-manage create network 10 6415:19
soren...if you run "nova-manage project zipfile" and it fails this way.15:19
soren..since that seems to be a common pattern.15:19
annegentlesoren: That's a much better fix than just doc, thanks for doing that.15:19
ttxannegentle: it leverages uec-publish-tarball which simplifies image registration15:20
annegentleis there any way to know what version of nova is installed on an existing installation?15:20
annegentlettx: oh that's good15:20
ttxannegentle: we should probably publish the ttylinux image under a uec-publish-tarball compatible form.15:20
annegentlettx: I appreciate you and soren both using blog entries to document fast-moving changes15:21
sorenannegentle: finding out what version you're on>  Not really. It depends on the install method you used, and if it's not from packages, there's really no way to do it.15:22
ttxannegentle: we should roll that back into doc, blogposts aren't great as reference doc.15:23
ttxannegentle: I just didn't want to add "yet another" install doc.15:23
annegentlettx: yup, that's what I appreciate most, that you intend for it to be rolled into doc :)15:24
ttxannegentle: you should define a few use cases (run from released packages, run from bleeding-edge packages, run from branch) and we should keep that up to date.15:25
annegentlesoren: re: version, ok15:25
ttxthough the matrix with deployment environments might become a bit too large15:26
annegentlettx: my current naming scheme for those three cases is NovaInstall/Austin, NovaInstall, InstallFromSource15:27
annegentlettx: now I need a new page for the VirtualBox items on the mailing list. Any suggestions?15:27
*** Podilarius has quit IRC15:28
annegentlettx: I think NovaInstall mixes different package gathering tho. Feel free to suggest a re-org as we go.15:29
ttxannegentle: i'm not sure this needs parallel instructions... maybe just a list of caveats and extra config steps for that kind of install ?15:29
ttx(the vbox stuff)15:29
annegentlettx: mostly just need a download location to post the links to the large iso or zip files15:30
ttxannegentle: we could also play a game of #Includes where we avoid duplicating info15:30
annegentlettx: love includes15:30
ttxannegentle: that makes it a bit less edit-friendly, but...15:31
annegentlettx: the edits we're seeing are light-touch, so I think (hope) people will leave macros alone. But I do know other tech writers have had to turn macros off when the wiki starts getting lots of edits.15:33
ttxI used includes heavily in
ttxI think for install instructions, that makes sense15:33
*** sirp has joined #openstack15:33
annegentlettx: I think for this particular item, install doc, it's time to do some more complex architecture to aid with maintenance15:33
annegentlettx: yes15:34
ttxannegentle: currently NovaInstall is a bit confusing with (1) an install from source (2) multimachine (3) Austin/10.10 instructions15:37
*** dfg_ has joined #openstack15:38
ttx(none of those covering the BleedingEdgePPA/10.10 case, which I think should be the recommended one.15:38
*** dendro-afk is now known as dendrobates15:38
annegentlettx: How about this? Have the wiki store just the bleeding edge and install from source cases, and the multimachine and Austin go to
*** johnpur has joined #openstack15:40
*** ChanServ sets mode: +v johnpur15:40
*** Podilarius has joined #openstack15:40
*** Cybo has joined #openstack15:40
ttxannegentle: makes sense to me. That way the docs in the release cover the installation instructions for the release :)15:40
*** schisamo has joined #openstack15:40
*** johnbergoon has joined #openstack15:41
ttxannegentle: also simpler to have a single use case per doc, even if that means playing with Includes for common content.15:42
ttxannegentle: not sure where would fall under that scheme though.15:43
ttxprobably in InstallFromSource.15:44
*** GasbaKid has joined #openstack15:45
annegentlettx: yeah, that sounds right to me.15:45
jaypipes_0x44, sirp, gundlach: if you've got a chance to peruse, that would be most appreciated.  Just want to know if I'm on the right track before I go further.15:46
gundlachjaypipes: saw that and starred it in my email.  i hope to be able to tackle it today.15:47
jaypipesgundlach: cheers :)15:47
sirpjaypipes: i looked at it some yesterday, and am looking at it right now15:47
jaypipessirp: sweet.15:48
sirpthe one thing that jumps out is the removal of parallax image uri, it def makes things simpler (and that's prob a good thing for now), but it's imcopatible (as far as i can tell) with the notion of federated registries...15:48
sirpperhaps we just keep it simple for bexar, cactus, and add a federated registries ticket for down-the-line15:48
sirpis that what you're thinking?15:48
sirp^^^ for jaypipes15:50
uvirtbotsirp: Error: "^^" is not a valid command.15:50
*** infernix has quit IRC15:51
jaypipessirp: I believe that is something to look into if and when a demand for that pops up...yes.15:51
jaypipessirp: after all, the Parallax registry has support for multiple backends...there is quite a bit of flexibility already there.15:52
jaypipessirp: and having a UUID identifier would essentially give us a federated registry ability for free...15:52
*** infernix has joined #openstack15:52
*** jordandev has joined #openstack15:53
jaypipessirp: since you could chain Parallax registries together as middleware... if one didn't find a UUID, simply pass the request on to another one.15:53
sirpjaypipes: with a uuid, who maintains the mapping of UUID -> (location, id)15:53
jaypipessirp: a parallax node.15:53
creihtwe need uuids as a service!15:53
jaypipescreiht: ++15:53
* creiht hopes jaypipes is joking as well :)15:54
jaypipescreiht: yes, dude.15:54
* jaypipes ushers creiht back into his hole...15:55
creihtjaypipes: sometimes it is hard to tell :)15:55
gundlachcreiht: just pay more attn to body language and tone15:56
jaypipescreiht: I know, I'm hard to read that way.  Kind of like a book by Sarah Palin.15:56
creihtrackerhacker: I think jaypipes would make for a good sarah palin photoshop :)15:56
rackerhackercreiht: ORLY15:57
rackerhackerwhy do you think of me when you think of sarah palin?15:57
creihtrackerhacker: I think of you when I think of photoshop :)15:57
_0x44jaypipes: Will star and look at it this evening.15:57
*** dendrobates is now known as dendro-afk16:03
*** kashyapc has joined #openstack16:03
*** trbs has joined #openstack16:04
*** dendro-afk is now known as dendrobates16:04
*** dragondm has joined #openstack16:08
* soren heads dinnerwards16:08
spectorclanDesign Summit 2011- new wiki page with current thinking on 2011 events at
spectorclanCurrent thinking is three events : US, Europe and Asia in 2011 with US in April ; Europe in Summer; Asia in Fall16:14
spectorclanCheck out thinking on US Date and Location - feedback is appreciated ; April 26-29, 2011 in San Francisco, CA16:14
*** befreax has joined #openstack16:15
*** laurensell has joined #openstack16:17
*** mdomsch has joined #openstack16:17
ttxspectorclan: 3 design summits in 2011 ? They are every 6 months currently. Planning to change that ?16:18
spectorclanttx: thinking is to have some type of mini-design summit in Europe during the summer; not a full 4 day event but maybe 1 or 2 days16:19
spectorclanwant to provide Europe an event in 201116:19
ttxspectorclan: ok, maybe not name it design summit then. rally or sprint, maybe16:19
spectorclanThe Asia event would be the main one with the US16:19
spectorclangood idea, will update16:20
ttxor miniconf or whatever.16:20
spectorclanttx: updated with Sprint for now, can work on name later16:20
spectorclanttx: thoughts on date?16:22
ttxspectorclan: not yet... I'd have to draft the D release schedule to see that it doesn't conflict with development milestones16:23
spectorclanttx: ok, no problem. thanks16:24
KnightHackercreiht: I can almost see your sarcasting smile .. but I finally got a chance to look at the existing auth implementation in swift and I see your arguments now :).16:26
KnightHackercreiht: I have a quick question if you don't mind. Is there a way that I can setup an "auth" server on a separate node than swift APIs?16:27
KnightHackerLike in the proposal, we have two models namely embedded and separated.16:28
creiht(that's how it works at RS now)16:28
KnightHackerI see. So, swift excepts some "auth" arguments in the WSGI environment. As long as these arguments are set, it doesn't matter where it came from.16:30
*** laurensell has left #openstack16:30
creihtit is all left as an excersize for the middleware16:31
KnightHackerLooking at this way .. the proposal is really just to "standardize" on the attributes that swift should expect to receive from the middleware.16:32
*** laurensell has joined #openstack16:32
*** Ryan_Lane|away is now known as Ryan_Lane16:33
KnightHackerIn the "separated" model, what happens if a request came directly to swift service itself instead of the "auth" service?16:36
KnightHackerWill it be redirected to Auth Service?16:37
creihtby separated, do you mean reverse proxy?16:37
KnightHackerI really mean running the auth service as an app on its own end point.16:38
creihtso the current scheme goes something like this16:38
creiht1.) use auths with auth service to get token16:39
creiht2.) user makes call to swift with the auth token16:39
creiht3.) the swift auth middleware makes a service call to auth to validate16:39
KnightHackerOK. is this middleware embedded in the swift service?16:40
creihtThat's how swift interacts currently with an autonomous auth system16:40
* KnightHacker thanks creiht for his help16:40
KnightHackerMy question is about the middleware that validates this token or whatever.16:41
creihtswift comes with a reference implementation of the auth middleware which interacts with the dev auth server in the same way16:41
KnightHackerYes, that's the implementation that I am looking at now.16:42
creihtAnd we have a separate auth middleware for the rack specific auth internally16:42
KnightHackerRight. This makes a ton of sense.16:42
*** calavera has quit IRC16:43
KnightHackerI guess my question is about separating the middleware from the swift service itself. For instance, can I deploy the middleware on a separate end-point than swift?16:43
KnightHackerI understand that this middleware is different than the auth service itself, which I like :).16:44
creihttheoretically yes16:44
KnightHackerBut can I deploy 5 nodes of that middleware without having to create 5 nodes of swift APIs?16:44
creihtoh well that is different16:45
creihtKnightHacker: perhaps a better question at this point, is what type of problem are you trying to solve here?16:46
JordanRinke1anyone here have the ability to admin the mailing list?16:46
KnightHackerscaling my authentication without having to scale my API servers.16:46
notmynameKnightHacker: your middleware could make a call to an external system (VIP in front of load balancers, etc) via whatever transport you like16:47
creihtso in this case the middleware is just a thin layer, what you would be scaling is the external auth service that it calls out to16:47
notmyname(that's exactly what the RAX auth system currently does)16:48
creihtThere isn't really much of anything to "scale" at the middleware layer16:48
KnightHackerWhat if an Operator decided to implement the full "authentication" mechanism in the middleware layer.16:49
*** MarkAtwood has joined #openstack16:49
KnightHackerThat way, the middleware won't be calling out to any external auth system.16:49
creihtthen that is their problem :)16:49
creihtwe can't prevent people from creating their own non-scalable systems :)16:49
KnightHackerhaha .. I was debating whether or not it is an edge case, while I am typing my previous message.16:49
notmynameit's more of a "bad idea" rather than "edge case"16:50
KnightHackerWell, we aren't recommending to them to create an extrenal Auth system either, are we?16:50
creihtIn most cases, I expect most places will either a.) already have an external auth system that they will integrate with, or b.) setting up something16:51
creihtKnightHacker: We at least recommend to *not* use the built in dev auth system :)16:51
notmynameit might not be a bad thing to have some contrib auth middlewares that support common stuff like LDAP or something16:52
notmynameas examples, not prod-ready16:52
creihtnotmyname: that basically already exists with xtoddx's auth implementation for nova16:53
Ryan_Laneweb SSO support would be even nicer ;)16:53
Ryan_Lanelike SAML16:53
KnightHackercreiht: notmyname: Thanks a bunch guys. I think I have a much better understanding now.16:53
creihtand there is an example of how to integrate with repoze.who in the docs16:53
creihtfeel free to ask questions anytime16:54
KnightHackerI appreciate it.16:54
KnightHackerAnd yes, I like the sample implementation for SAML, OAuth, ..etc.16:55
creihtpatches welcome :)16:55
zykes-what auth does swift support nwo ?16:55
KnightHackerYes, I have started to look at the code and exploring :).16:55
creihtnwo=New World Order?16:56
edayvishy, soren: Can you guys approve again? had some text conflicts and won't let me merge without approving those :)16:56
creihtzykes-: swift auth is handled via middleware, so while support for auth system X isn't built in, it shouldn't be difficult to incorporate16:56
notmynamezykes-: the devauth that swift ships with (not for prod use) closely matches (but isn't identical) to the rackspace cloud auth system16:57
KnightHackercreiht: what are you thoughts about authenticating "admins" or "ops" folks. For instance, what if I want to restrict the swift admin operations to specific accounts that are obviously different than the users/customers accounts.17:07
KnightHackerShould that be another middleware or should it be the same middleware handling both authentications?17:07
* ttx EODs17:08
*** kashyapc has joined #openstack17:10
zykes-ttx: EOD ?17:13
zykes-creiht: middlewarel ike ?17:14
creihtdepends on what you want to do17:15
creihtIn general ops folks don't interact directly with users accounts, just systems17:15
creihtAt least in our case17:16
creihtwhat is EOD?17:16
*** 16SAARQT8 has quit IRC17:16
* gundlach EOLs, rather tautologically17:17
*** kevnfx has joined #openstack17:18
edaygundlach: damnit, stop breaking my merge req :)17:20
gundlacheday: heh!17:20
gundlachneed me to re-approve?17:20
gundlachor, approve, i guess?17:20
edaygundlach: not now, once your last one lands.. will need to merge trunk again17:20
* gundlach thinks that maybe if eday weren't so slow this wouldn't be a problem <ducks/>17:21
edaygundlach: also, you're supposed to have two nova-core approvals before merging into trunk.. not that your last one did too much though :)17:21
edaygundlach: yeah, that's what we discussed at the summit17:21
gundlachi'm in nova-core, do i get a vote?  i had somehow forgotten that we needed two separate approvals.17:21
edaynot on your own :)17:21
gundlachthanks for the reminder; shame on me17:21
*** kevnfx has quit IRC17:22
openstackhudsonProject nova build #278: SUCCESS in 1 min 13 sec:
*** befreax has quit IRC17:25
*** aliguori has quit IRC17:31
*** lorin has joined #openstack17:34
*** babble has joined #openstack17:35
*** babble has left #openstack17:35
*** babble has joined #openstack17:35
*** al-maisan is now known as almaisan-away17:35
edayvishy: want to try once more? gundlach got another rev in there that broke mine :)17:39
zykes-hmm, RHEL6 seems nice17:46
zykes-contra RHEL517:46
*** dirakx has joined #openstack17:49
*** ibarrera has quit IRC17:50
*** rlucio has joined #openstack17:51
*** eldarnugaev has quit IRC18:00
*** soeren has quit IRC18:01
*** joearnold has joined #openstack18:02
*** deshantm has quit IRC18:02
*** deshantm_cosi has quit IRC18:02
*** deshantm_cosi has joined #openstack18:05
*** deshantm has joined #openstack18:05
*** daleolds has joined #openstack18:08
gundlachsoren: pyflakes in vim has just pointed out a bug in existing code (that i wrote).  rock!18:10
*** deshantm_cosi has quit IRC18:10
*** deshantm has quit IRC18:10
edaygundlach:  if you have a moment :)18:11
Ryan_Lanenow that are services share a common flagfile, how do you set the log file for individual services?18:12
gundlacheday: was just typing you a question asking about when that might appear, as i have more code and didn't want to get in your way again.18:12
gundlachi'll look now.18:12
*** eldarnugaev has joined #openstack18:13
*** Ryan_Lane is now known as Ryan_Lane|away18:16
vishyeday: is this new? it used to be you could just reapprove at the new revision18:21
edayvishy: hudson kicked it back saying there were un-approved revs18:22
vishyweird, never seen that one before18:22
*** deshantm has joined #openstack18:23
edayperhaps there was a race condition, the merge req page hadn't seen the new rev when I marked it18:23
vishyeday: i'm testing.  That seems more likely.  I don't think hudson checks the subreviews at all, just the main status18:24
edayyeah, makes more sense18:25
*** westmaas has quit IRC18:29
gundlacheday: lgtm18:30
*** westmaas has joined #openstack18:30
gundlacheday: i've seen that message before as well -- 'there are additional code changes that need review' -- started within the last week or so.18:30
*** Podilarius has quit IRC18:30
*** GasbaKid has quit IRC18:30
openstackhudsonProject nova build #279: SUCCESS in 1 min 16 sec:
*** irahgel has left #openstack18:35
*** jc_smith has joined #openstack18:38
*** Podilarius has joined #openstack18:45
*** matclayton has left #openstack18:46
uvirtbotNew bug: #684353 in nova "Need a way to determine Nova version after installing" [Undecided,New]
*** aliguori has joined #openstack18:47
*** eldarnugaev has quit IRC18:49
*** deshantm_cosi has joined #openstack18:52
*** westmaas has quit IRC18:54
*** joearnold has quit IRC18:56
*** westmaas has joined #openstack18:57
*** joearnold has joined #openstack19:00
rluciosoren: fyi the packaging branch(es) need to be updated; the nova virt templates have moved from /usr/share/nova to /usr/share/pyshared/nova/virt19:15
*** johnbergoon has quit IRC19:17
*** irahgel has joined #openstack19:19
*** aliguori has quit IRC19:19
*** aliguori has joined #openstack19:22
sorenrlucio: Did that already.19:24
sorenrlucio: 30 seconds after they moved or thereabouts.19:25
rluciosoren: ? i just upgraded to 2011.1~bzr436-0ubuntu0ppa1~lucid1 and all the services are still expecting stuff in /usr/share/nova19:25
*** sophiap has joined #openstack19:25
rlucioby default19:25
rluciook thats not entirely fair, not 'all services' i have ony looked at compute and manage19:26
sorenrlucio: Where are you seeing this?19:27
rluciounless my existing config file clobbered the upgrade.  i just jumped on it b/c it looks like the packaging branch hasnt been updated in a while19:27
sorenrlucio: Oh, you're looking at the wrong packaging branch.19:28
sorenrlucio: I should try to make that more obvious, really.19:28
sorenThe packaging branch that's actually being used to build packages is quite lively.19:28
rluciosoren: oh my bad then...19:29
vishyok the ttylinux freaking rocks19:30
vishysuper helpful for testing19:30
zykes-ttylinux ?19:31
sorenvishy: How's it different from your tiny image?19:31
vishy1/5th of the size and supports scripts passed in via user data19:31
sorenvishy: What /does/ the tiny image do with the meta-data service? Just ping it?19:34
sorenall I could find was a proxy of some sort.19:35
vishysoren: yeah it proxied19:35
vishywith ttylinux i can do this19:35
vishyand then i can verify connectivity to metadata and to internet without sshing in to the instance19:36
vishyeuca-run-instances -t m1.tiny -f ami-tty19:36
zykes-proxy what ?19:37
sorenvishy: Oh, so you just connect to the instance on those ports to check connectivity?19:38
sorenThat makes sense.19:38
*** babble has quit IRC19:38
vishyjust do a curl to verify19:38
sorenI just noticed the error on the console when it couldn't connect, so I thought it actually did something useful with it.19:38
soren...and couldn't find anything in the filesystem that looked remotely interesting. Ok, that clears that up.19:38
vishywe have a burn in script that launches 1000 instances on a new host, verifying connectivity to each one19:40
sorenYeah, that's good thinking.19:40
sorenUsing KVM, right?19:40
vishyscript is at
* soren is happy that /someone/ is doing reasonably large scale, end-to-end testing. :)19:41
vishyer...forget the .git at the end of that19:41
*** eldarnugaev has joined #openstack19:41
*** jordandev has quit IRC19:42
*** KnightHacker has quit IRC19:46
*** KnightHacker has joined #openstack19:46
*** westmaas has quit IRC19:48
*** virgil_ has joined #openstack19:50
*** joearnold has quit IRC19:54
*** Ryan_Lane|away is now known as Ryan_Lane20:04
virgil_part #openstack20:10
*** virgil_ has left #openstack20:15
*** westmaas has joined #openstack20:17
vishysoren: possible to make a uml ttylinux?20:18
vishysoren: seems like it would be really useful for cloudpipes, we could throw a few hundred on a single box20:19
gundlacheday: re-review?20:20
*** gondoi has quit IRC20:24
edaygundlach: so, I know this may be out of the scope of what you were trying to do, but it would be nice to also remove the db.instance get call in the update too. I'm already doing this for delete, so perhaps we should include it as part of another branch20:25
gundlacheday: yeah, i thought about taking care of the other db_driver calls, but didn't want scope creep.20:25
edayit's annoying how ec2 and os do authn differently, ec2 with project beforehand, os after we've looked at the instance record20:26
edayI don't want to start pushing authn checks down into nova.compute.api, but we also need to remove db calls from the api servers20:27
edayideas anyone? :)20:28
gundlacheday: you do mean authentication, right?20:30
gundlachwhich is done before we look at the instance record in OS20:31
edayno, authn (or access control), the instance.user_id == authc'd user_id20:31
edaywe want to be able to perform those checks without looking up instance ids20:32
gundlachheh, i think of authN as authNtication and authZ as authoriZation20:32
edayoh, perhaps i was screwing those up20:33
edayauth* :)20:33
gundlachsooo don't they both do authorization the same way?  aka, in the db layer, with @require_context?20:34
* gundlach goes to look at the code20:34
edayoh, yeah20:35
edayso the os checks are really pointless20:35
gundlachlooks to me like both ec2 and api fetch the instance record and then make the db call that has @require_context20:35
gundlachs/and api/and OS api/20:35
gundlacheday: can you point me to the "os checks" you're talking about?20:35
gundlachfor authorization i'm only aware of @require_[admin_]context and i think it's not pointless20:36
edaylike the one in update we were just talking about:20:37
eday        instance = self.db_driver.instance_get_by_internal_id(ctxt, int(id))20:37
eday        if not instance or instance.user_id != user_id:20:37
eday            return faults.Fault(exc.HTTPNotFound())20:37
edaybut that user_id check has already happened in the db lookup, if I'm reading the db code correctly20:37
gundlachyep, you're right, the 'or instance.user_id != user_id' can go.  of course we need the first part.20:38
edayI think all those can be safely removed20:38
edaywell, the first part can be a NotFound exception from a compute.api method call20:39
*** gondoi has joined #openstack20:39
edayand catch/raise the os-specific faults20:39
gundlachso we'd pass the whole instance in to compute.api.update_instance?20:40
gundlachno, i'm slow, it would just accept an internal_id.20:40
*** joearnold has joined #openstack20:41
gundlachcompute_api.update_instance(context, INTERNAL_id, values)20:41
gundlachthat sound right?20:41
edayI'm doing this for delete right now, one sec20:41
edayif this looks good I'll do the rest like this too20:41
gundlachi was thinking it's bad that we are exposing the sqlalchemy ids outside the DB layer now that we have the db-driver-agnostic internal_ids20:42
edayprobably, we should make sure we use a guid like we were talking about before for all APIs20:44
edayas soon as we have separate compute workers generating the same sqlite id's, things will break pretty quickly :)20:45
eday(or just don't use sqlite id's at all, and use the guid's as primary keys)20:45
gundlachon that topic, can you clarify something for me re: eventual consistency of a scheduler's view of its compute workers?20:47
gundlachlet's say the OS API wants to enforce a constraint: "No two servers owned by a user may have the same name."20:47
gundlachWhen a user requests to update a server's name, who do we check to see if that's allowed?20:48
gundlachCan we check the scheduler's knowledge of its running instances?  I think not, because it's got some lag in its knowledge.20:48
gundlachDo we have to hit every compute node containing VMs for that user?20:48
gundlachOr perhaps we say, "unique server names are not compatible with our architecture" and punt?20:49
edaygundlach: punt, but we can make a best guess using method #120:50
edaygundlach: and on conflict, automatically rename one of them20:50
*** HouseAway is now known as AimanA20:50
gundlacheday: right then, thanks20:51
edayat least that what I was thinking, but yeah, we need to work around having any kind of global consistency :)20:51
spectorclanbrining up nova following Thierry blog post today (; question on creating the network; he says I used my unused address; do I use that or something else based on my existing IP?20:57
sorenvishy: For something like cloudpipe, kvm with ksm should be a good fit, really.20:59
sorenvishy: I doubt you'd get much better density with uml.20:59
sorenKernel Shared Memory (I think).21:00
sorenIt's a thingamajig that goes around looking for identical pages and removes dupes.21:00
*** zaitcev has joined #openstack21:00
sorenSo if two guests have an identical memory page, the kernel throws one away, and lets the guests share the one.21:01
sorenIf either of them writes to it, a copy is created. It's completely transparent to guests.21:01
sorenFor identical guests (apart from a bit of configuration), there should be a lot of potential for high density stuff.21:01
vishysoren: interesting21:02
creihtdendrobates: how do we find out what our echosign transaction number is?21:06
notmynamecreiht: it's supposed to be in an email that we got when we originally signed it21:07
* notmyname can't find his either. hope it's in my home email account21:08
*** DubLo7 has joined #openstack21:09
JordanRinkeanyone here that can manage the openstack mailing list?21:11
spectorclanJordanRinke: which mailing list?21:13
*** daleolds has quit IRC21:13
spectorclanI don;t manage but perhaps I can assist21:13
JordanRinkefor whatever reason, I can not send email to it21:13
JordanRinkeI get stuff from the mailing list, but anything I try to send never goes through21:13
creihtJordanRinke: I saw your emails21:14
creihtthe list doesn't send you posts that you make to the list21:14
JordanRinkeon my email it sends me a copy and they show up in the archive21:14
JordanRinkemy ones weren't getting sent to me (or my other email address) and didn't show up in the archive21:14
spectorclanJordanRinke:  I would check with Thierry21:15
*** brd_from_italy has joined #openstack21:17
JordanRinkecool, just email him.21:17
JordanRinkeit is driving me nuts heh21:17
spectorclansorry I couldn't help more21:18
spectorclan i manage the other email lists21:18
JordanRinkeeh no prob, got me pointed in the right direction... that is more then I had 10 minutes ago21:18
*** ctennis has joined #openstack21:22
sorenJordanRinke: I keep getting moderation requests about your e-mail.21:27
sorenJordanRinke: Does Launchpad know about the address from whence you're sending e-mail?21:27
*** hggdh has quit IRC21:29
*** dendrobates is now known as dendro-afk21:30
*** rlucio has quit IRC21:30
*** westmaas has quit IRC21:37
*** dirakx has quit IRC21:37
*** KyleM1 has joined #openstack21:37
*** AimanA is now known as HouseAway21:37
*** eldarnugaev has quit IRC21:40
*** aliguori has quit IRC21:41
JordanRinkesoren: sorry I went AFK - yes it does, at least I think so -
JordanRinkepicture of my current settings21:41
jk0anyone in here who's used nova with XenServer?21:43
*** kevnfx has joined #openstack21:45
*** hggdh has joined #openstack21:46
vishynotmyname: i couldn't find the transaction id, but when i click to view the signed document, there is an aid= param in the url.  I think that is the id21:50
notmynamefrom an email? or a url? I haven't found the email I got yet21:51
*** jizzo has joined #openstack21:52
vishyI did it from an email, but you might be able to get there by signing in on rackspace.echosign.com21:52
exltif anyone has any interest in python packaging, there is a training session starting in a few minutes in #debian-women on OFTC (not freenode) -
*** sophiap has quit IRC21:56
*** mdomsch has quit IRC21:57
*** westmaas has joined #openstack21:59
*** miclorb_ has joined #openstack21:59
vishywhere is a good place to upload images.tgz with the new ami-tty?21:59
*** sophiap has joined #openstack22:02
notmynamevishy: I think I'm hosting the current one in my cloud files account. but I think someone was talking about putting it in an openstack cloud files account. don't know too many details22:03
*** GasbaKid has joined #openstack22:05
*** gundlach has quit IRC22:08
exltoh, whoops.. my calandar alert was set at the wrong time.. I missed 20:00 UTC..22:09
openstackhudsonProject nova build #280: SUCCESS in 1 min 13 sec:
*** jdarcy has quit IRC22:14
*** aliguori has joined #openstack22:15
JordanRinkeI can up it22:17
JordanRinketo the OpenStack CDN / Images22:18
*** arthurc has quit IRC22:19
*** Cybo has quit IRC22:22
jk0vishy: do you know if it's possible to provide a s3 username/password somehow?22:23
* jk0 lazy irc :)22:23
vishyprovide to ...22:23
jk0well, use the xenapi creds for s3 auth22:24
vishynever used xenapi...22:24
*** spectorclan has quit IRC22:24
* vishy wonders if we are putting in the proper stuff on contributors, since dendros and others start with Q, but everyone elses starts with X22:26
vishys/Q/Q or P/22:27
*** rsampaio has quit IRC22:27
*** Tushar has joined #openstack22:28
*** Tushar has quit IRC22:28
edayvishy: huh?22:30
Ryan_Lanedoes nova require a ramdisk image to create an instance? I used uec-publish-tarball to push a lucid 64 bit image, and it came without a ramdisk22:30
Ryan_Lanebut when I try to run it, I get the following error: NotFound: Image ari-11111 could not be found22:30
*** DubLo7 has quit IRC22:31
JordanRinkeRyan_Lane: currently Yes22:32
JordanRinkeyou can use a ramdisk from something else though, at least I have been able to a few times22:32
JordanRinkedo you have one you could use?22:32
Ryan_Lanethe systems hosting it should work22:32
Ryan_Lanethey are lucid and amd6422:33
JordanRinkeyeah, I used a lucid ramdisk on a maverick image and it booted22:33
Ryan_Lanethough the kernel revision needs to be the same, right?22:33
Ryan_Laneguess not :)22:33
JordanRinkeso, it doesn't seem too picky22:33
JordanRinke(I don't know the inner workings, I may have just gotten lucky)22:34
vishyeday: the contributors page in the wiki where were supposed to show that we've signed the cla22:37
edayvishy: ahh, ok22:41
_0x44vishy: I have two of them, one is X the other starts with Q22:41
vishy_0x44: where did you get these things?22:42
*** brd_from_italy has quit IRC22:42
_0x44vishy: The email from echosign included a pdf, it was at the bottom of every page.22:43
notmynamefrom the CyberDuck dev: "You may also be intersted to know that the upcoming release 3.8 of Cyberduck comes preconfigured with a bookmark option for Swift installations."22:43
_0x44There's a URL, if you click it, it'll take you to check that the number is valid.22:43
vishyah i didn't see the id at the bottom of the pdf22:45
vishylucky me though, it is the same as the aid from the link22:45
*** rlucio has joined #openstack22:46
Ryan_Laneugh. I'm trying to register a bundled and uploaded ramdisk, and I'm getting this error on register: Warning: failed to parse error message from AWS: <unknown>:132:75: undefined entity22:48
JordanRinkeIs there a PPA or something similar that just has the stable release? For instance we want a script that downloads the stable release, installs, configurs (puppet/bash/chef whatever) we cant use the current PPA since it gets updated with new changes, and we dont want to have to BZR and pull Austin22:48
JordanRinkemaybe it is there already and I just didn't see it when I was looking around22:49
*** Tushar464 has joined #openstack22:49
*** allsystemsarego has quit IRC22:49
*** GasbaKid has quit IRC22:50
*** GasbaKid has joined #openstack22:50
JordanRinkeI see the nova-core PPA and Trunk, but neither have a package from the Austin release from what I can see22:50
JordanRinkeohhh maybe I found it22:51
JordanRinkeyeah, think I found it in instead of nova-core :/22:52
Ryan_Lanenevermind. I was trying to register the thing on the filesystem, not in the bucket (duh)22:52
JordanRinkeeh, not sure that is it either, the package details show build errors for the lucid package, and it says nova is 0.9.1 instead of what I would expect to be 1.0? Anyone have the info to set me straight here?22:55
*** matiu has joined #openstack22:56
*** sophiap has quit IRC22:56
*** sophiap has joined #openstack22:56
*** jbaker has quit IRC22:58
*** jizzo_ has joined #openstack22:59
*** jizzo has quit IRC23:01
*** jizzo_ is now known as jizzo23:01
*** gondoi has quit IRC23:02
*** hazmat has quit IRC23:02
*** westmaas has quit IRC23:04
KyleM1Aright, I'm trying to run nova on ubuntu lucid.  I've got the script and run the branch, install, and run commands.  The branch and install appeared to happen correctly.23:14
KyleM1When I did ./ run it complained that nova/nova.sqlite didn't exist so it couldn't rm it.23:15
KyleM1Then spat out some INFO lines and exported the keys.23:15
KyleM1When screen came up screen 6 (volume) wasn't was back to a prompt.23:15
KyleM1I was able to create the test.pem key pair and launch an instance following the instructions in the quickstart guide.23:16
KyleM1But when I tried to ssh into it I got a no route to host error.23:16
KyleM1Any suggestions?23:16
Ryan_LaneKyleM1: you don't actually need nova-volume23:16
Ryan_Laneyou also may need to add ssh to the default security group23:17
Ryan_Laneeuca-authorize default -P tcp -p 22 -s
KyleM1I'll try that.23:17
*** ppetraki has quit IRC23:18
KyleM1Hmm...last time I terminated nova by detaching from the screen as suggested in the quickstart guide.  When I ran it again the euca tools are giving me connection failed errors:23:20
KyleM1EC2_ACCESS_KEY environment variable must be set.23:20
KyleM1That didn't happen last I need to do more than detach the screen and wait for my prompt to come back to clean it up?23:21
KyleM1It also appears to be cluttering the directory I run it from with files called something like "hostname.1234" which seem to be lock files of some kind.23:22
dubsquared1Getting what seems to be a pretty obvious error here, but the files do indeed not exist…are they missing from the packages in /ppa and/or /trunk:
*** dendro-afk is now known as dendrobates23:27
*** aliguori has quit IRC23:28
*** rlucio has quit IRC23:28
Ryan_Lanedubsquared1: you need to set the path23:29
Ryan_Lanedubsquared1: --state_path=/var/lib/nova23:29
Ryan_Laneif you are using a newer version (like trunk)23:29
dubsquared1i see23:30
dubsquared1thank you!23:30
Ryan_Laneor --ca_path and --key_path otherwise23:30
Ryan_Laneor is it keys_path?23:30
dubsquared1can i just append that to the nova-manage, like the —flagpole ?23:30
dubsquared1damn autocorrect23:31
dendrobatesJordanRinke: your messages are getting held in moderation.  Have you added the email address you are using to your lp account?23:32
*** jizzo has quit IRC23:32
KyleM1Does anyone have any ideas why nova is complaining that the EC2_ACCESS_KEY env. variable isn't set the second time I use './ run'?23:38
*** Ryan_Lane is now known as Ryan_Lane|away23:41
*** rlucio has joined #openstack23:45
*** Podilarius has left #openstack23:49
*** kashyapc has joined #openstack23:55
*** dendrobates is now known as dendro-afk23:57
*** trbs has quit IRC23:59

Generated by 2.14.0 by Marius Gedminas - find it at!