| *** MotoMilind has joined #openstack | 00:00 | |
| pvo | Ryan_Lane: very cool | 00:01 |
|---|---|---|
| *** st-17217 has joined #openstack | 00:01 | |
| Ryan_Lane | just created a new instance: http://nova-controller.tesla.usability.wikimedia.org/t-1/index.php?title=Resource_query_examples&action=purge | 00:01 |
| Ryan_Lane | gah. bad url :) | 00:02 |
| Ryan_Lane | http://nova-controller.tesla.usability.wikimedia.org/trunk.1/Resource_query_examples | 00:02 |
| termie | justinsb: responded to your volumes bits | 00:02 |
| justinsb | termie: thanks | 00:03 |
| *** santhoshtw has joined #openstack | 00:08 | |
| termie | justinsb: ping when you're ready and i'll look again | 00:08 |
| *** reldan has quit IRC | 00:09 | |
| justinsb | termie: Thanks - also found a leftover from the EC2 code - the attachment format in the volume is different from the attachment format if you query it directly, so just fixing that. Even fewer lines of non-supporting code :-) | 00:09 |
| aixenv | im trying vishy's anso packages openstack build.. scrapping what i have currently | 00:09 |
| *** icarus901 has joined #openstack | 00:11 | |
| *** zaccone has quit IRC | 00:14 | |
| *** adiantum has quit IRC | 00:19 | |
| *** dragondm has quit IRC | 00:19 | |
| justinsb | termie: Delousification complete: https://code.launchpad.net/~justin-fathomdb/nova/volumes-api/+merge/54464 | 00:23 |
| *** santhoshtw has quit IRC | 00:23 | |
| termie | justinsb: that docstirng on get_console_pool_info is still whacky | 00:24 |
| termie | what syntax is it? why don't you know what the values mean? | 00:24 |
| justinsb | termie: I don't know what the values mean because I didn't write the function | 00:25 |
| justinsb | termie: I can just delete it... | 00:25 |
| justinsb | termie: No docstring better than poor docstring? | 00:26 |
| *** notmyname has quit IRC | 00:28 | |
| *** notmyname has joined #openstack | 00:28 | |
| *** ChanServ sets mode: +v notmyname | 00:28 | |
| *** benb__ has joined #openstack | 00:30 | |
| justinsb | termie: Docstring will be gone once LP finishes running the diff | 00:30 |
| justinsb | I think we should see if Canonical want to use our test cloud to run their diffs :-) | 00:31 |
| *** adiantum has joined #openstack | 00:32 | |
| winston-d | creiht : around? | 00:33 |
| termie | justinsb: okies | 00:37 |
| termie | justinsb: approved | 00:39 |
| justinsb | termie: Thanks! | 00:40 |
| justinsb | vishy: You around for a review of the volumes-api? | 00:40 |
| vishy | justinsb: yeah i'll do it tonight | 00:41 |
| vishy | I'm tracking an annoying bug or three atm | 00:41 |
| justinsb | vishy: OK! | 00:42 |
| justinsb | Anyone else from nova-core around for a review of the volumes-api? My FFE expires in a few hours... | 00:42 |
| vishy | note to self: don't merge and push a changed bzr branch that builds off of version numbers | 00:42 |
| *** dirakx has joined #openstack | 00:45 | |
| *** jeffjapan has quit IRC | 00:46 | |
| justinsb | devcamcar: Any chance I could ask you to have a look at the volumes-api branch? | 00:47 |
| devcamcar | justinsb: hah, weird timing, i had just pulled it up in fact | 00:48 |
| *** st-17217 has quit IRC | 00:48 | |
| justinsb | devcamcar: Awesome! Just saw you review another branch, which is why I asked! | 00:48 |
| *** jeffjapan has joined #openstack | 00:52 | |
| devcamcar | justinsb: class ExtensionDescriptor(object) | 00:54 |
| devcamcar | i do agree with jay, that does seem above and beyond the proposal | 00:54 |
| devcamcar | its useful but mixing multiple features in a single branch makes it hard for reviewers | 00:54 |
| justinsb | devcamcar: It's just a base class, to simplify creating an extension. It makes my extension shorter. | 00:55 |
| justinsb | devcamcar: I thought everyone's objection was to the _use_ of the base class in discovering the extension | 00:56 |
| justinsb | devcamcar: Not to the use of the base class to simplify creating extensions ? | 00:56 |
| devcamcar | k may have been biased by prev reviews, i'll keep reading | 00:56 |
| devcamcar | justinsb: still looks like there was a lot of foundational work done in nova/api/openstack/extensions.py | 00:59 |
| devcamcar | that looks like it a should be a prereq branch | 00:59 |
| devcamcar | how about contrib vs incubator? | 00:59 |
| justinsb | devcamcar: This was the first 'real' extension | 00:59 |
| justinsb | devcamcar: nova/api/openstack/contrib or contrib/ ? | 01:00 |
| *** pmatulis has quit IRC | 01:00 | |
| devcamcar | nova/api/openstack/contrib instead of nova/api/openstack/incubator | 01:00 |
| justinsb | If it'll get you to approve, I'll call it nova/api/openstack/devcamcarisawesome/ | 01:01 |
| benb__ | Hi, I'm a complete newbie. Does Nova run on 64 bit Intel processors? | 01:01 |
| dabo | benb__: yes | 01:01 |
| *** pmatulis has joined #openstack | 01:01 | |
| *** pmatulis has left #openstack | 01:02 | |
| creiht | winston-d: kinda? :) | 01:03 |
| winston-d | creiht : sorry to bug you again. | 01:03 |
| creiht | hehe | 01:03 |
| creiht | what's up? | 01:03 |
| benb__ | It looks like Bleeding edge OpenStack Nova on Maverick by Thierry Carrez is the easiest way to get started. I have a new (blank) hard disk on my server. Where do I get (download) the software? | 01:04 |
| *** icarus901 has quit IRC | 01:04 | |
| justinsb | devcamcar: There's almost no new genuine code in extensions.py. There's add_extension but I can remove that if you'd prefer. | 01:04 |
| *** adiantum has quit IRC | 01:04 | |
| devcamcar | justinsb: yea i'm starting to agree | 01:05 |
| devcamcar | justinsb: approve with a nit: | 01:06 |
| devcamcar | 833+class MockImageService(service.BaseImageService): | 01:06 |
| devcamcar | A nit - can we call this a FakeImageService? A mock is different. | 01:06 |
| winston-d | creiht : 1) can i alter container ACL via API bindings? i'd obtain the URL for public accessable object (to be used in website), but with PHP bindings, the only option is to get URL via $container->public_uri(), which requires CDN support. | 01:06 |
| justinsb | devcamcar: Sure - will change Mock -> Fake ... good point! | 01:06 |
| justinsb | devcamcar: Thanks | 01:07 |
| creiht | winston-d: since the feature for public containers isn't in cloudfiles yet, the current bindings do not have that functionality, yet | 01:07 |
| devcamcar | justinsb: glad to see this patch going in! | 01:07 |
| devcamcar | if you can knock that out real quick i'll approve it | 01:07 |
| devcamcar | back in a few | 01:07 |
| justinsb | devcamcar: Me too! | 01:07 |
| winston-d | creiht : that's bad news for me | 01:07 |
| creiht | winston-d: if you want binding compatibility, then what you could do is either | 01:08 |
| dabo | benb__: best place is from the nova trunk, available on Launchpad. If you have bzr installed, just do: bzr branch lp:nova trunk | 01:08 |
| creiht | a: have auth return the cdn-management-url which is an endpoint that you would write that would make the container changes | 01:08 |
| dabo | benb__: with the caveat that trunk is changing often, and may break at times. | 01:08 |
| creiht | b: write a piece of middleware that turns the cdn-management functions into public container functions | 01:09 |
| *** rchavik has joined #openstack | 01:09 | |
| *** rchavik has joined #openstack | 01:09 | |
| *** adiantum has joined #openstack | 01:10 | |
| winston-d | creiht : yeah, i'd perfer binding compatibility. option a look more straightforward to me. | 01:10 |
| uvirtbot | New bug: #745388 in nova "images fail to register with --use_project_ca=True" [Medium,In progress] https://launchpad.net/bugs/745388 | 01:12 |
| winston-d | creiht : thanks for the hints. btw, any plan to support 1) cdn in swift 2) public container in CloudFile API? | 01:13 |
| benb__ | Thanks dabo. I see lp:nova/bexar, probably more stable than Cactus. Does this download contain Ubuntu, everything needed to install on bare metal (a freshly formatted hard disk)? | 01:13 |
| *** mray has joined #openstack | 01:13 | |
| dabo | benb__: You'll need some sort of hypervisor; nova supports several of the most common | 01:14 |
| *** MotoMilind has quit IRC | 01:15 | |
| dabo | benb__: e.g., I have a box with Xenserver installed; I created a domU instance on that, running ubuntu 10.10, and then installed nova on that. | 01:15 |
| *** zul has quit IRC | 01:17 | |
| *** zul has joined #openstack | 01:18 | |
| creiht | winston-d: 1) is a little complicated, as our current solution is kinda tailored for our cdn provider. That functionality is fairly thin though, so it wouldn't be hard to add a thin layer on top of swift to act as an origin for a cdn provider. At some point we *may* have a more general solution for this, but I can't gurantee it | 01:18 |
| vishy | benb__: fresh install of maverick + this might be your easiest install: http://images.ansolabs.com/single_node_install.sh | 01:19 |
| creiht | 2.) There is some discussion of making openstack specific bindings, with the cloudfiles specific bindings an extension to those. Once we do that, then it should be easier to track openstack features in the bindings. | 01:19 |
| creiht | and to all of the above, patches are always welcome :) | 01:20 |
| winston-d | creiht : sounds good. :) | 01:20 |
| *** ewindisch has quit IRC | 01:23 | |
| *** miclorb has quit IRC | 01:26 | |
| *** EdwinGrubbs has joined #openstack | 01:29 | |
| *** EdwinGrubbs has joined #openstack | 01:29 | |
| *** nelson has quit IRC | 01:30 | |
| *** nelson has joined #openstack | 01:30 | |
| benb__ | Hi vishy. First I install Ubuntu Maverick 64 bit, then I follow http://images.ansolabs.com/single_node_install.sh? I don't need to install a hypervisor? | 01:34 |
| *** dprince has joined #openstack | 01:37 | |
| *** rchavik has quit IRC | 01:39 | |
| *** mray has quit IRC | 01:40 | |
| *** gregp76 has joined #openstack | 01:42 | |
| *** kashyap has joined #openstack | 01:43 | |
| *** dprince has quit IRC | 01:45 | |
| *** Ryan_Lane has quit IRC | 01:54 | |
| *** RobertLaptop has quit IRC | 01:56 | |
| *** adiantum has quit IRC | 01:56 | |
| *** ewindisch has joined #openstack | 01:58 | |
| *** RobertLaptop has joined #openstack | 01:59 | |
| *** joearnold has quit IRC | 02:03 | |
| *** adiantum has joined #openstack | 02:09 | |
| *** pharkmillups has joined #openstack | 02:14 | |
| *** littleidea has quit IRC | 02:15 | |
| justinsb | Any nova core still around to repush volumes-api? Unit tests failed due to new requirements from the image service. All fixed now... | 02:16 |
| justinsb | https://code.launchpad.net/~justin-fathomdb/nova/volumes-api/+merge/54464 | 02:16 |
| *** littleidea has joined #openstack | 02:20 | |
| *** miclorb has joined #openstack | 02:22 | |
| *** ccustine has quit IRC | 02:26 | |
| winston-d | creiht : now that i'm using swauth, should i put 'cdn-management-url' part into proxy-server? | 02:28 |
| creiht | well I would add it to the headers returned from the swauth middleware when a user authenticates | 02:30 |
| *** pandemicsyn has joined #openstack | 02:31 | |
| *** ChanServ sets mode: +v pandemicsyn | 02:31 | |
| creiht | winston-d: handle_get_token in swift/common/middleware/swauth.py | 02:33 |
| *** adiantum has quit IRC | 02:33 | |
| winston-d | creiht : got it. | 02:33 |
| * winston-d looking to swift source | 02:33 | |
| creiht | winston-d: looks like you would add the header at the very end of the function | 02:34 |
| *** adiantum has joined #openstack | 02:39 | |
| *** yamahata has quit IRC | 02:45 | |
| *** aimon has joined #openstack | 02:51 | |
| *** aimon has left #openstack | 02:52 | |
| openstackjenkins | Project nova build #744: SUCCESS in 2 min 29 sec: http://hudson.openstack.org/job/nova/744/ | 02:54 |
| openstackjenkins | Tarmac: Support for volumes in the OpenStack API | 02:54 |
| openstackjenkins | Two new endpoints: /volumes and /servers/<id>/volume_attachments | 02:54 |
| *** aliguori has quit IRC | 02:54 | |
| *** miclorb has quit IRC | 03:03 | |
| *** santhoshtw has joined #openstack | 03:08 | |
| *** maplebed has quit IRC | 03:13 | |
| *** mdomsch has joined #openstack | 03:18 | |
| aixenv | when im logging into the openstack dashboard , where is that username/password generated from? | 03:22 |
| aixenv | is this user supposed to be my nova_user in the novarc ? | 03:22 |
| *** adjohn has joined #openstack | 03:22 | |
| aixenv | and is the password the key associated in nova-manage? | 03:22 |
| aixenv | user = my nova-manage user and pass = my EC2_ACCESS_KEY (or SECRET_KEY)? | 03:24 |
| *** santhoshtw has quit IRC | 03:26 | |
| *** miclorb has joined #openstack | 03:37 | |
| *** littleidea has quit IRC | 03:44 | |
| *** yamahata has joined #openstack | 03:44 | |
| *** MotoMilind has joined #openstack | 03:46 | |
| *** AimanA is now known as HouseAway | 03:53 | |
| *** jeffjapan has quit IRC | 03:56 | |
| *** guynaor has joined #openstack | 03:56 | |
| *** guynaor has left #openstack | 03:57 | |
| *** MotoMilind has quit IRC | 04:02 | |
| *** jeffjapan has joined #openstack | 04:05 | |
| *** santhoshtw has joined #openstack | 04:10 | |
| sleepsonthefloor | aixenv - the username/password is generated by and for the django app (probably when you initialized the db). separate credentials, specified in settings.py, are used to communicate with nova | 04:12 |
| sleepsonthefloor | but... usernames in the dash need to map to nova users to behave as expected. dashboard will try to create nova users upon registration if they do not exist | 04:14 |
| *** MotoMilind has joined #openstack | 04:15 | |
| *** MotoMilind has joined #openstack | 04:16 | |
| *** joearnold has joined #openstack | 04:16 | |
| *** littleidea has joined #openstack | 04:16 | |
| *** burris has quit IRC | 04:21 | |
| *** littleidea has quit IRC | 04:23 | |
| *** hadrian has quit IRC | 04:24 | |
| *** burris has joined #openstack | 04:26 | |
| *** MotoMilind has quit IRC | 04:26 | |
| *** littleidea has joined #openstack | 04:29 | |
| *** mdomsch has quit IRC | 04:33 | |
| *** t_yas1 has joined #openstack | 04:34 | |
| *** nijaba has quit IRC | 04:41 | |
| aixenv | sleepsonthefloor: ty, you can sleep on the sofa | 04:42 |
| *** nijaba has joined #openstack | 04:42 | |
| *** f4m8_ is now known as f4m8 | 04:49 | |
| *** pvo has quit IRC | 04:54 | |
| *** pvoccio has joined #openstack | 04:55 | |
| *** pvoccio is now known as pvo | 04:55 | |
| *** pvo has joined #openstack | 04:56 | |
| *** ChanServ sets mode: +v pvo | 04:56 | |
| *** pharkmillups has quit IRC | 04:56 | |
| *** kashyap has quit IRC | 04:58 | |
| *** _morfeas has quit IRC | 05:10 | |
| *** mastermind has quit IRC | 05:10 | |
| *** kashyap has joined #openstack | 05:17 | |
| *** dendro-afk is now known as dendrobates | 05:17 | |
| *** joearnold has quit IRC | 05:20 | |
| *** mastermind has joined #openstack | 05:24 | |
| *** ttt has joined #openstack | 05:25 | |
| *** adiantum has quit IRC | 05:32 | |
| *** gregp76 has quit IRC | 05:34 | |
| *** jeffjapan has quit IRC | 05:35 | |
| *** dendrobates is now known as dendro-afk | 05:37 | |
| *** dirakx has quit IRC | 05:41 | |
| *** dendro-afk is now known as dendrobates | 05:42 | |
| *** adiantum has joined #openstack | 05:45 | |
| *** gregp76 has joined #openstack | 05:48 | |
| *** gregp76 has quit IRC | 05:53 | |
| *** freeflying has quit IRC | 05:53 | |
| *** jeffjapan has joined #openstack | 05:53 | |
| *** benb__ has quit IRC | 05:54 | |
| *** freeflying has joined #openstack | 05:55 | |
| *** jeremyb has quit IRC | 05:56 | |
| *** jeremyb has joined #openstack | 05:56 | |
| *** ewindisch has quit IRC | 05:56 | |
| *** ivan has quit IRC | 05:58 | |
| *** jbryce has quit IRC | 06:00 | |
| openstackjenkins | Project nova build #745: SUCCESS in 2 min 30 sec: http://hudson.openstack.org/job/nova/745/ | 06:04 |
| openstackjenkins | Tarmac: The VNC Proxy is an OpenStack component that allows users of Nova to access | 06:04 |
| openstackjenkins | their instances through a websocket enabled browser (like Google Chrome). | 06:04 |
| openstackjenkins | A VNC Connection works like so: | 06:04 |
| openstackjenkins | * User connects over an api and gets a url like http://ip:port/?token=xyz | 06:04 |
| openstackjenkins | * User pastes url in browser | 06:04 |
| openstackjenkins | * Browser connects to VNC Proxy though a websocket enabled client like noVNC | 06:04 |
| openstackjenkins | * VNC Proxy authorizes users token, maps the token to a host and port of an | 06:04 |
| openstackjenkins | instance's VNC server | 06:04 |
| openstackjenkins | * VNC Proxy initiates connection to VNC server, and continues proxying until | 06:04 |
| openstackjenkins | the session ends | 06:04 |
| openstackjenkins | For more info see vncconsole.rst | 06:04 |
| *** matiu has quit IRC | 06:07 | |
| *** gregp76 has joined #openstack | 06:07 | |
| *** jeremyb has quit IRC | 06:08 | |
| *** ivan has joined #openstack | 06:09 | |
| *** ivan has joined #openstack | 06:09 | |
| *** jeremyb has joined #openstack | 06:14 | |
| *** dendrobates is now known as dendro-afk | 06:15 | |
| uvirtbot | New bug: #745481 in nova "FLAGS.vncproxy_topic doesn't exist in compute.api" [Low,In progress] https://launchpad.net/bugs/745481 | 06:16 |
| *** adiantum has quit IRC | 06:16 | |
| *** MotoMilind has joined #openstack | 06:26 | |
| *** nerens has joined #openstack | 06:29 | |
| *** adiantum has joined #openstack | 06:29 | |
| HugoKuo_ | does anyone use deployment tool before ? | 06:30 |
| *** gregp76 has quit IRC | 06:32 | |
| *** MotoMilind has quit IRC | 06:36 | |
| *** littleidea has quit IRC | 06:42 | |
| *** mgoldmann has joined #openstack | 06:45 | |
| *** zenmatt has quit IRC | 06:45 | |
| openstackjenkins | Project nova build #746: SUCCESS in 2 min 28 sec: http://hudson.openstack.org/job/nova/746/ | 06:48 |
| openstackjenkins | Tarmac: Uses the proc filesystem to check the volume size in volume smoketests so that it works with a very limited busybox image. | 06:49 |
| *** lionel has quit IRC | 06:52 | |
| *** lionel has joined #openstack | 06:52 | |
| *** burris has quit IRC | 06:59 | |
| *** flopflip_ has joined #openstack | 07:04 | |
| *** allsystemsarego has joined #openstack | 07:05 | |
| *** allsystemsarego has joined #openstack | 07:05 | |
| *** flopflip has quit IRC | 07:08 | |
| *** CloudChris has joined #openstack | 07:15 | |
| *** miclorb has quit IRC | 07:17 | |
| *** miclorb has joined #openstack | 07:22 | |
| *** miclorb has quit IRC | 07:25 | |
| *** bkkrw has joined #openstack | 07:27 | |
| *** koolhead11|afk is now known as koolhead11 | 07:33 | |
| *** miclorb_ has joined #openstack | 07:36 | |
| *** daveiw has joined #openstack | 07:37 | |
| *** santhoshtw has quit IRC | 07:45 | |
| *** Nacx has joined #openstack | 07:47 | |
| *** reldan has joined #openstack | 07:50 | |
| *** drico has quit IRC | 07:54 | |
| *** drico has joined #openstack | 07:55 | |
| *** uksysadmin has joined #openstack | 07:59 | |
| *** ramkrsna has joined #openstack | 08:14 | |
| *** ramkrsna has joined #openstack | 08:14 | |
| *** nijaba has quit IRC | 08:20 | |
| *** nijaba has joined #openstack | 08:22 | |
| *** nijaba has joined #openstack | 08:22 | |
| *** Daviey has quit IRC | 08:32 | |
| *** Daviey has joined #openstack | 08:34 | |
| *** Daviey has quit IRC | 08:38 | |
| *** t_yas1 has quit IRC | 08:38 | |
| *** Daviey has joined #openstack | 08:39 | |
| uvirtbot | New bug: #745541 in nova "nova-compute failed on import_class error" [Undecided,New] https://launchpad.net/bugs/745541 | 08:46 |
| *** gasbakid has joined #openstack | 08:50 | |
| *** kashyap has quit IRC | 08:54 | |
| *** jeffjapan has quit IRC | 08:56 | |
| *** winston-d has quit IRC | 08:58 | |
| *** kashyap has joined #openstack | 09:02 | |
| soren | W000t! The green is back: https://jenkins.linux2go.dk/ | 09:03 |
| *** jfluhmann has quit IRC | 09:03 | |
| *** miclorb_ has quit IRC | 09:04 | |
| *** miclorb__ has joined #openstack | 09:04 | |
| * soren treats himself to an orange to celebrate | 09:04 | |
| zigo-_- | soren: Hi there! | 09:05 |
| soren | zigo-_-: Hi. | 09:09 |
| zigo-_- | I'm having issues to start some of the daemons. | 09:10 |
| soren | zigo-_-: Which ones? | 09:10 |
| zigo-_- | Let me check. | 09:10 |
| zigo-_- | nova-network, I believe, but not sure. | 09:10 |
| ttx | soren: did we fix bug 739609 already ? rings a bell but cannot find reference. | 09:11 |
| uvirtbot | Launchpad bug 739609 in nova "Cannot set body to unicode without a charset errors" [Undecided,New] https://launchpad.net/bugs/739609 | 09:11 |
| zigo-_- | You guys don't use debian/*.init at all in Ubuntu, right? | 09:11 |
| zigo-_- | I mean, I can change things, it wont influence? | 09:11 |
| soren | zigo-_-: Right. | 09:11 |
| zigo-_- | Cool. | 09:11 |
| soren | Yes. | 09:11 |
| soren | ttx: Ah, it's the webob version that's off. | 09:12 |
| * soren fixes | 09:12 | |
| zigo-_- | Do we HAVE to use libvirt? My understanding is that we can do without, right? | 09:14 |
| *** reldan has quit IRC | 09:14 | |
| zigo-_- | Setting up nova-compute (2011.2~bzr828-1) ... | 09:14 |
| zigo-_- | adduser: The group `libvirtd' does not exist. | 09:14 |
| soren | zigo-_-: Well, sure, if you don't want to run any instances. | 09:14 |
| zigo-_- | :) | 09:14 |
| soren | zigo-_-: Yeah, you guys call it libvirt. | 09:14 |
| zigo-_- | Oh, shit, yet another difference? | 09:15 |
| zigo-_- | That's crap. | 09:15 |
| soren | Hysterical raisins. | 09:15 |
| *** reldan has joined #openstack | 09:15 | |
| zigo-_- | :) | 09:15 |
| zigo-_- | Are you for dpkg-vendor --derives-from ubuntu or for getent? | 09:15 |
| zigo-_- | I'd be for a getent check ... | 09:15 |
| *** jfluhmann has joined #openstack | 09:16 | |
| soren | Why? | 09:16 |
| zigo-_- | Because there's more chance it's gona work. | 09:18 |
| soren | I'm not convinced that's true. | 09:19 |
| zigo-_- | if getent group libvirtd | 09:19 |
| zigo-_- | AND | 09:19 |
| zigo-_- | if getent group libvirt | 09:19 |
| zigo-_- | Then add to both groups if we see both ... | 09:19 |
| zigo-_- | (which wont happen, but still ...) | 09:20 |
| soren | I don't like that. | 09:20 |
| HugoKuo_ | there r several point should be describe more detail or correct in deplomentTool? | 09:20 |
| HugoKuo_ | where should I report it ? | 09:20 |
| HugoKuo_ | on launchpad ? | 09:20 |
| soren | Ubuntu systems have this group named libvirtd. The admin may have created another group called libvirt and I'm sure he wouldn't enjoy us fiddling around with it. | 09:20 |
| zigo-_- | So what do you suggest? | 09:20 |
| soren | dpkg-vendor. | 09:20 |
| zigo-_- | dpkg-vendor --derives-from ubuntu? | 09:20 |
| zigo-_- | I don't mind ... | 09:21 |
| alekibango | zigo-_-: libvirt, yes thats the small patch i wanted to give you :) | 09:21 |
| zigo-_- | Ok, I do that then. | 09:21 |
| alekibango | libvirtd | 09:21 |
| soren | I don't see the downside to using dpkg-vendor? | 09:21 |
| alekibango | zigo-_-: where is your code? | 09:21 |
| soren | What am I missing? | 09:21 |
| alekibango | zigo-_-: i can send you patch | 09:21 |
| zigo-_- | soren: As I said, I'm ok either ways, as long as we have a fix. | 09:22 |
| soren | ok | 09:22 |
| alekibango | btw i used getent group like you zigo-_- :) | 09:23 |
| soren | zigo-_-: With bzr, you shouldn't just remove a file and add it again with a different name. Use bzr rename to track renames. | 09:23 |
| zigo-_- | I saw that. | 09:23 |
| zigo-_- | Git wouldn't mind, but I read bzr does. | 09:23 |
| zigo-_- | I should use "bzr mv" ? | 09:24 |
| zigo-_- | What is the way? | 09:24 |
| soren | It's not about "minding". it's about tracking these things as two different actions. | 09:24 |
| alekibango | soren: git is smarter a bit there :) its code management, not file management | 09:24 |
| soren | bzr's data model is somewhat richer than git's (for better or worse). | 09:24 |
| alekibango | lol | 09:24 |
| soren | alekibango: that's a pointless generalisation. | 09:25 |
| zigo-_- | Could we stop digressing and progress in what we do? :) | 09:25 |
| zigo-_- | "bzr mv", right? | 09:25 |
| alekibango | soren: i could say that too :) hehe | 09:25 |
| soren | zigo-_-: Yes, or "bzr rename". | 09:25 |
| zigo-_- | Cheers. | 09:25 |
| soren | alekibango: About what? | 09:25 |
| alekibango | soren: dont worry, we should rather work on something good | 09:26 |
| alekibango | flamewars about editor or code management systems will not help :) | 09:26 |
| soren | I don't see the flamewar, to be honest. | 09:27 |
| soren | Not yet, at least. | 09:27 |
| alekibango | soren: it always start like that :) | 09:27 |
| soren | It's a simple conversation, exchanging information about version control systems. | 09:27 |
| alekibango | zigo-_-: where is your code repo? | 09:27 |
| alekibango | soren: OS will not move from bzr, so its moot to discuss this now :) | 09:28 |
| soren | At least that's what I was aiming for. | 09:28 |
| alekibango | but i would love to use git, tahts sure | 09:28 |
| * soren gives up | 09:28 | |
| zigo-_- | alekibango: My lp ID is thomas-goirand | 09:28 |
| *** miclorb_ has joined #openstack | 09:28 | |
| *** miclorb__ has quit IRC | 09:29 | |
| zigo-_- | Starting nova compute: nova-computelibvir: error : Unknown failure | 09:29 |
| zigo-_- | :/ | 09:29 |
| *** miclorb_ has quit IRC | 09:30 | |
| alekibango | zigo-_-: thats almost as goot as 'error: operation finished successfully' | 09:30 |
| alekibango | good* | 09:30 |
| soren | zigo-_-: What do you need lsb-base for? something about the init scripts? | 09:32 |
| *** reldan has quit IRC | 09:33 | |
| zigo-_- | It's for when you use log_daemon_msg things. | 09:33 |
| soren | Ah, right. | 09:34 |
| alekibango | zigo-_-: imho it might be nice to use runit... what do you think? | 09:34 |
| soren | zigo-_-: Isn't lsb-base required in Debian? | 09:34 |
| alekibango | soren: not required | 09:34 |
| alekibango | not always | 09:34 |
| soren | Well, either it is, or it isn't. | 09:35 |
| soren | Which is it? | 09:35 |
| alekibango | you can isnstall base system without lsb | 09:35 |
| zigo-_- | runit? What's that? | 09:35 |
| alekibango | zigo-_-: something like smarters daemontools | 09:35 |
| zigo-_- | soren: Required doesn't mean essential, you still need to depend on it. | 09:35 |
| zigo-_- | At least, to my knowledge. | 09:35 |
| zigo-_- | Or, I should say... | 09:36 |
| alekibango | yes its not essential... | 09:36 |
| soren | Package: lsb-base | 09:36 |
| soren | Priority: required | 09:36 |
| zigo-_- | If it didn't change since last time I did use lsb-base. | 09:36 |
| zigo-_- | Again, required != essential | 09:36 |
| soren | Yeah. | 09:36 |
| alekibango | zigo-_-: i would not mind this dependency | 09:36 |
| zigo-_- | required means that it's going to be installed by default by the debian installer. | 09:36 |
| zigo-_- | While essential means you don't need to depend on it, it's going to be always there. | 09:37 |
| alekibango | ... i do not use installer much :) | 09:37 |
| zigo-_- | alekibango: I'm not fan of stuff like daemontools and such. | 09:37 |
| zigo-_- | It hides crashes... | 09:37 |
| alekibango | ... http://smarden.org/runit/benefits.html | 09:38 |
| zigo-_- | A daemon should always run, and never crash, that's it. | 09:38 |
| zigo-_- | Also, that's not very convential to use these. | 09:38 |
| zigo-_- | This doesn't mean that an admin can't use it... | 09:38 |
| alekibango | maybe as alternative? | 09:38 |
| zigo-_- | Just that the official package shouddn't by default. | 09:39 |
| alekibango | ok zigo-_-... | 09:39 |
| alekibango | i just was tempted to do so... as i like how it treats restarts, logs etc | 09:39 |
| alekibango | old initscripts are not that good | 09:39 |
| *** sebastianstadil has quit IRC | 09:40 | |
| alekibango | and it is really small, clean, easy to understand | 09:40 |
| zigo-_- | alekibango: What I like is monit! | 09:41 |
| zigo-_- | Ever used it? | 09:41 |
| zigo-_- | It's great, modular, etc. | 09:41 |
| alekibango | zigo-_-: i like zabbix :) | 09:41 |
| zigo-_- | Monit is *not* a monitoring software. | 09:41 |
| alekibango | ah, no never used it :) | 09:41 |
| zigo-_- | It checks if a service is up and running, and restarts it if needed. | 09:41 |
| zigo-_- | What's good is that it doesn't only check for the process, but for the functionality of the daemon. | 09:42 |
| alekibango | <zigo-_-> A daemon should always run, and never crash, that's it. | 09:42 |
| alekibango | :) | 09:42 |
| zigo-_- | Right! :) | 09:42 |
| zigo-_- | BRB | 09:42 |
| alekibango | zigo-_-: zabbix can do some actions too... :) its not only monitoring... | 09:43 |
| soren | zigo-_-: This will turn your add/removes into proper renames: for f in debian/nova-*.upstart.in; do bzr revert -r 123 ${f%.in}; bzr rm ${f} ; bzr rename ${f%.in} ${f}; done | 09:46 |
| soren | zigo-_-: For each file it reverts it to r123 (you deleted it in r124), removes the new .in, and does a proper bzr rename instead. | 09:47 |
| *** adiantum has quit IRC | 09:47 | |
| soren | zigo-_-: We don't really need the lsb-base dependency under Ubuntu. You could put it in substvars, perhaps? | 09:49 |
| zigo-_- | soren: How do you do that? | 09:51 |
| soren | zigo-_-: Those ${misc:Depends} things you often see in debian/control... You can add your own. | 09:52 |
| zigo-_- | Yes, but then? How do I substitute? | 09:52 |
| soren | dpkg-gencontrol does it. | 09:52 |
| soren | zigo-_-: You know... Never mind. It just adds complexity and I don't really care if we needlessly depend on a required package. | 09:53 |
| zigo-_- | Ok, found it. | 09:54 |
| soren | zigo-_-: But see deb-substvars(5), if you're interested. | 09:54 |
| zigo-_- | dh_gencontrol --dpkg-gencontrol-params=params | 09:54 |
| *** miclorb has joined #openstack | 09:55 | |
| soren | Uh.. | 09:55 |
| soren | well, yeah, or you just just put it in debian/substvars | 09:55 |
| *** benbenhappy has joined #openstack | 09:58 | |
| *** adiantum has joined #openstack | 10:01 | |
| *** joloughlin has joined #openstack | 10:04 | |
| joloughlin | nova-compute service not enabled after latest update from trunk | 10:05 |
| soren | joloughlin: https://bugs.launchpad.net/nova/+bug/744833 by any chance? | 10:06 |
| uvirtbot | Launchpad bug 744833 in nova "python-suds shouldn't be a hard dependency" [Medium,In progress] | 10:06 |
| soren | joloughlin: Put differently: does it help if you install python-suds? | 10:07 |
| joloughlin | soren: let me check | 10:07 |
| joloughlin | this is the error logs | 10:08 |
| joloughlin | http://pastebin.com/b5ZMhwhg | 10:08 |
| soren | yup, that's it | 10:08 |
| joloughlin | yeah, thanks | 10:09 |
| joloughlin | that enabled now | 10:09 |
| *** reldan has joined #openstack | 10:14 | |
| *** adiantum has quit IRC | 10:15 | |
| HugoKuo_ | https://answers.launchpad.net/nova-deployment-tool/+question/151020 | 10:23 |
| HugoKuo_ | package never been installed on nova-network host with deployment tool | 10:23 |
| *** yamahata has quit IRC | 10:25 | |
| soren | HugoKuo_: Is that "service dnsmasq stop" thing something from the deployment tool, or is it something you're trying to do yourself? | 10:26 |
| *** benbenhappy has left #openstack | 10:27 | |
| *** adiantum has joined #openstack | 10:27 | |
| soren | zigo-_-: You could use the ${f%.in} substitution thing in your override_dh_installinit thing, by the way. That would be easier on the eyes :) | 10:30 |
| soren | zigo-_-: ...and you should remove them up again in the clean target. | 10:30 |
| soren | s/ up / / | 10:30 |
| soren | zigo-_-: Any particular reason why you put the man pages in a separate directory? | 10:31 |
| soren | zigo-_-: ..and you seem to be missing a ; before done in override_dh_installinit. | 10:32 |
| soren | zigo-_-: I think. I haven't actually run the build, I'm just looking at the diff so far. | 10:32 |
| *** omidhdl has joined #openstack | 10:36 | |
| *** adiantum has quit IRC | 10:43 | |
| *** drico has quit IRC | 10:44 | |
| *** drico has joined #openstack | 10:45 | |
| *** adiantum has joined #openstack | 10:48 | |
| zigo-_- | soren: Why do you think everything should be in the same folder? :) | 10:49 |
| *** adjohn has quit IRC | 10:49 | |
| zigo-_- | Anyway, these mans have to go in the upstream code. | 10:49 |
| zigo-_- | Not in our Debian folder. | 10:49 |
| soren | I wouldn't spend a lot of time on that. | 10:51 |
| soren | I believe the arg parsing system is being revamped. | 10:52 |
| soren | You might be interested to learn about gflags2man, by the way. | 10:52 |
| zykes- | are anyone running rpm builds for openstack ? | 10:53 |
| *** rcc has joined #openstack | 10:59 | |
| rcc | ttx: hi thierry, lp702636 apparently is no longer valid, Unknown exception is no longer shown . as there's no branch, i can't ask for a review | 11:00 |
| zigo-_- | soren: Fixed accordingly. | 11:00 |
| rcc | should i put it as INVALID ? | 11:00 |
| zigo-_- | Funny that the missing ; didn't bother me when building! | 11:00 |
| zigo-_- | now pushing... | 11:00 |
| zigo-_- | pushed | 11:01 |
| *** dlafferty has quit IRC | 11:06 | |
| *** timbo_ has joined #openstack | 11:06 | |
| *** drico has quit IRC | 11:06 | |
| *** drico has joined #openstack | 11:07 | |
| timbo_ | can anybody please explain what i need to install or configure in my custom images so that iscsi volumes are injected? it works flawlessly with an uec-image, but nothing is logged inside my custom instance, based on a normal ubuntu server installation. | 11:08 |
| *** adiantum has quit IRC | 11:09 | |
| *** miclorb has quit IRC | 11:11 | |
| HugoKuo_ | soren : as you said | 11:16 |
| HugoKuo_ | soren : radvd and dnsmasq and nova-nework would not be installed | 11:17 |
| HugoKuo_ | I'm going to try different component on different host to make sure the problem | 11:18 |
| soren | HugoKuo_: I don't understand. | 11:18 |
| soren | HugoKuo_: What did I say? | 11:19 |
| *** ctennis has quit IRC | 11:19 | |
| HugoKuo_ | hmm... | 11:19 |
| HugoKuo_ | ok make it simpler | 11:19 |
| HugoKuo_ | deploymenttool process stop at 'service dnsmasq stop' | 11:20 |
| HugoKuo_ | so I checked if dnsmasq has been installed on nova-network host .. the answer is not | 11:21 |
| HugoKuo_ | after that I check radvd & nova-netowork packages on nova-netowork component, same answer , they were not been installed | 11:22 |
| *** benbenhappy has joined #openstack | 11:27 | |
| HugoKuo_ | oh~wait.....I found those package is installed while I change a host computer !!! | 11:27 |
| HugoKuo_ | soren : just forget about it , it must be something wrong with my laptop | 11:28 |
| soren | Ok. | 11:28 |
| *** adjohn has joined #openstack | 11:28 | |
| HugoKuo_ | thanks | 11:29 |
| *** lionel has quit IRC | 11:29 | |
| *** photron has joined #openstack | 11:30 | |
| HugoKuo_ | is there any problem if I'm using different OS version ? for example 3 ubuntu 10.10/ 64bits and 1 ubuntu10.10/ 32bits | 11:30 |
| photron | what's the recommended ubuntu version to run openstack? | 11:31 |
| HugoKuo_ | photron : 10.10 & 10.04 | 11:31 |
| *** ctennis has joined #openstack | 11:31 | |
| *** ctennis has joined #openstack | 11:31 | |
| benbenhappy | is ubuntu 32bit ok? | 11:31 |
| HugoKuo_ | I think yes | 11:32 |
| *** timbo_ has quit IRC | 11:32 | |
| *** benbenhappy has left #openstack | 11:33 | |
| *** lionel has joined #openstack | 11:34 | |
| *** omidhdl has quit IRC | 11:39 | |
| *** morfeas has joined #openstack | 11:41 | |
| joloughlin | I can add images to nova via uec-publish-image or nova-manage image and glance details show they are there | 11:44 |
| joloughlin | but euca-describe-images returns 'an Unknown error has occured' | 11:45 |
| joloughlin | is nova with glance currently working ? | 11:54 |
| zigo-_- | http://paste.debian.net/112410/ <-- What happens when I start nova-network | 12:01 |
| zigo-_- | Any idea what's happening here? | 12:01 |
| zigo-_- | Could it be because /var/lock/nova doesn't exist? | 12:01 |
| * zigo-_- tries... | 12:01 | |
| *** adjohn has quit IRC | 12:02 | |
| sandywalsh | ttx, taking 718869 | 12:03 |
| ttx | sandywalsh: yay! | 12:03 |
| *** aloga has joined #openstack | 12:04 | |
| alekibango | zigo-_-: was it missing dir? | 12:07 |
| zigo-_- | It seems so yes. | 12:09 |
| alekibango | zigo-_-: i think i had to do the same... | 12:09 |
| zigo-_- | But now, nova-network doesn't create the pid file. | 12:09 |
| alekibango | zigo-_-: why? dont you see something in logs? | 12:10 |
| *** dmshelton has quit IRC | 12:10 | |
| * zigo-_- is buzy with server migration, will resume after done | 12:11 | |
| alekibango | zigo-_-: imho you should be done soon with nova... | 12:12 |
| *** h0cin has joined #openstack | 12:12 | |
| *** koolhead11 has quit IRC | 12:15 | |
| *** romain_lenglet_ has joined #openstack | 12:17 | |
| *** guigui1 has joined #openstack | 12:19 | |
| *** dendro-afk is now known as dendrobates | 12:19 | |
| *** stewart has quit IRC | 12:27 | |
| *** stewart has joined #openstack | 12:27 | |
| joloughlin | is nova with glance currently working ? | 12:28 |
| HugoKuo_ | see you guys ~ | 12:28 |
| openstackjenkins | Project nova build #747: SUCCESS in 2 min 28 sec: http://hudson.openstack.org/job/nova/747/ | 12:28 |
| openstackjenkins | Tarmac: TopicAdapterConsumer uses a different callback model than TopicConsumer. This patch updates the console proxy to use this pattern. | 12:28 |
| openstackjenkins | Without this patch, nova-ajax-console proxy will generate an error every time an auth token is cast to it, and hence it won't work. | 12:29 |
| *** santhoshtw has joined #openstack | 12:32 | |
| *** shentonfreude has quit IRC | 12:36 | |
| n1md4 | do I install ALL nova- packages for a compute node? | 12:36 |
| *** gasbakid has quit IRC | 12:41 | |
| *** kirshil has joined #openstack | 12:43 | |
| *** spring has joined #openstack | 12:43 | |
| spring | hi | 12:43 |
| spring | could u help me please, how can i reset the instance ids? | 12:43 |
| *** hadrian has joined #openstack | 12:43 | |
| spring | i had a VM with i-0000001 id, after i have terminated it | 12:43 |
| spring | but new vm get an incremented id | 12:44 |
| *** adiantum has joined #openstack | 12:44 | |
| spring | is there any way to reset this id? | 12:44 |
| zul | morning | 12:44 |
| joloughlin | are there currently any problems with nova and glance ? | 12:48 |
| n1md4 | hi. failing to spawn instances on compute node http://paste.openstack.org/show/1032/ any ideas? | 12:49 |
| *** shentonfreude has joined #openstack | 12:52 | |
| *** aliguori has joined #openstack | 12:56 | |
| *** zenmatt has joined #openstack | 12:57 | |
| *** santhoshtw has quit IRC | 13:08 | |
| zul | n1md4: " Instance '13' failed to spawn. Is virtualization enabled in the BIOS? " | 13:14 |
| *** dendrobates is now known as dendro-afk | 13:15 | |
| *** dendro-afk is now known as dendrobates | 13:15 | |
| *** MarcMorata has joined #openstack | 13:21 | |
| *** dprince has joined #openstack | 13:24 | |
| *** joloughlin has quit IRC | 13:24 | |
| *** dendrobates is now known as dendro-afk | 13:26 | |
| blamar | spring: I think the instance id is based on an auto-increment field in the database so you'd have to reset that value I think? | 13:28 |
| n1md4 | zul: yeah, i'd seen that too. I'm sure I'd had instances running on that machine before though. | 13:31 |
| n1md4 | zul: it's not ideal hardware, gigabyte boards, so flashing them with an update is a window$ affair, that I'm having difficulty doing. | 13:36 |
| *** burris has joined #openstack | 13:41 | |
| openstackjenkins | Project nova-tarmac build #73,800: FAILURE in 7 min 26 sec: http://hudson.openstack.org/job/nova-tarmac/73800/ | 13:43 |
| cjreyn | hey guys, I'm having the following error for glance: ERROR: Unable to load glance-registry from configuration file /etc/glance/glance.conf. | 13:46 |
| cjreyn | Got: ImportError('No module named MySQLdb',) | 13:46 |
| cjreyn | I'm guessing there is a dependency here? | 13:46 |
| *** EdwinGrubbs has quit IRC | 13:50 | |
| dubs | cjreyn: yes, you'll need to install the python mysql adapter | 13:52 |
| *** f4m8 is now known as f4m8_ | 13:52 | |
| cjreyn | and the --log-config switch is broken? Gives me the error: | 13:54 |
| pvo | soren: are you also doing the packaging in Ubuntu for glance? | 13:54 |
| cjreyn | ConfigParser.NoSectionError: No section: 'formatters' | 13:54 |
| openstackjenkins | Project nova-tarmac build #73,801: STILL FAILING in 11 min: http://hudson.openstack.org/job/nova-tarmac/73801/ | 13:55 |
| pvo | soren: I found it. | 13:56 |
| *** theOG has joined #openstack | 13:56 | |
| *** santhoshtw has joined #openstack | 13:57 | |
| *** theOG is now known as OliG | 13:57 | |
| OliG | Hi All# | 13:57 |
| openstackjenkins | Project nova build #748: SUCCESS in 2 min 29 sec: http://hudson.openstack.org/job/nova/748/ | 13:59 |
| openstackjenkins | Tarmac: Makes the image decryption code use the per-project private key to decrpyt uploaded images if use_project_ca is set. This allows the decryption code to work properly when we are using a different ca per project. | 13:59 |
| *** ppetraki has joined #openstack | 13:59 | |
| OliG | I'm looking to do a compute install on Ubuntu in our blade environment, does anyone know if I need TV on the board/chip in order to run 64-bit OSes | 14:00 |
| OliG | *VT even :s | 14:01 |
| openstackjenkins | Yippie, build fixed! | 14:02 |
| openstackjenkins | Project nova-tarmac build #73,802: FIXED in 7 min 15 sec: http://hudson.openstack.org/job/nova-tarmac/73802/ | 14:02 |
| dprince | So it looks like the glance PPA is broken: | 14:10 |
| dprince | http://hudson.openstack.org/job/glance-ppa/71/ | 14:10 |
| dprince | Any idea why? | 14:10 |
| *** adjohn has joined #openstack | 14:10 | |
| *** adjohn has quit IRC | 14:10 | |
| *** Zangetsue has joined #openstack | 14:11 | |
| *** nRy has joined #openstack | 14:17 | |
| n1md4 | "Instance '17' failed to spawn. Is virtualization enabled in the BIOS?" this is a separate machine, and yes, it is. Any ideas? | 14:17 |
| spring | OliG u need VT for nodes | 14:18 |
| *** omidhdl has joined #openstack | 14:18 | |
| *** zul has quit IRC | 14:21 | |
| *** zul has joined #openstack | 14:23 | |
| *** dirakx has joined #openstack | 14:24 | |
| OliG | but not the controller right spring :? | 14:29 |
| *** jaypipes-afk is now known as jaypipes | 14:30 | |
| *** jeremyb has quit IRC | 14:31 | |
| *** jeremyb has joined #openstack | 14:31 | |
| * zigo-_- just finished migration from one DC to another. :) | 14:32 | |
| *** shentonfreude has quit IRC | 14:33 | |
| *** rds__ has joined #openstack | 14:33 | |
| *** jaypipes has joined #openstack | 14:34 | |
| *** ttt has joined #openstack | 14:36 | |
| *** ttt has quit IRC | 14:37 | |
| *** t_yas1 has joined #openstack | 14:38 | |
| n1md4 | where does openstack look for api_host? | 14:41 |
| *** gondoi has joined #openstack | 14:41 | |
| *** allsystemsarego_ has joined #openstack | 14:42 | |
| *** CloudChris has quit IRC | 14:42 | |
| *** MarcMorata has quit IRC | 14:46 | |
| *** sebastianstadil has joined #openstack | 14:47 | |
| *** MarcMorata has joined #openstack | 14:49 | |
| *** t_yas1 has quit IRC | 14:49 | |
| zigo-_- | nova-network fails with the following log: http://paste.debian.net/112435/ | 14:50 |
| zigo-_- | Then never creates the pid file ... | 14:50 |
| *** MarcMorata has quit IRC | 14:51 | |
| *** littleidea has joined #openstack | 14:52 | |
| zigo-_- | Anyone knows why? | 14:52 |
| *** mray has joined #openstack | 14:57 | |
| *** omidhdl has left #openstack | 14:58 | |
| ttx | zigo-_-: that's https://bugs.launchpad.net/nova/+bug/735974 | 14:58 |
| uvirtbot | Launchpad bug 735974 in nova "nova-network logs exception when starting up" [Low,Confirmed] | 14:58 |
| ttx | zigo-_-: the trace is actually harmless | 14:58 |
| ttx | your pidfile issue is probably unrelated | 14:59 |
| zigo-_- | Thanks. | 14:59 |
| *** dragondm has joined #openstack | 14:59 | |
| *** nelson has quit IRC | 14:59 | |
| zigo-_- | Trying last bzr. | 14:59 |
| ttx | zigo-_-: we may not be using pidfiles anymore -- ask soren to be sure | 15:01 |
| n1md4 | i'm having trouble with every node that is not the controller. saying "Is virtualization enabled in the BIOS" even though it is. This must be the way I'm configuring the nodes, but I can't figure out what. I'm part following these guidelines http://wiki.openstack.org/NovaInstall/MultipleServer and http://docs.openstack.org/openstack-compute/admin/content/ch03s02.html | 15:02 |
| zigo-_- | I don't mind, start-stop-daemon can create them for me. | 15:02 |
| zigo-_- | But I think that the option can stay... | 15:02 |
| *** reldan has quit IRC | 15:05 | |
| *** rcc has quit IRC | 15:06 | |
| *** dmshelton has joined #openstack | 15:06 | |
| *** daveiw has quit IRC | 15:06 | |
| *** guigui1 has quit IRC | 15:07 | |
| *** ramkrsna has quit IRC | 15:07 | |
| openstackjenkins | Project nova build #749: SUCCESS in 2 min 33 sec: http://hudson.openstack.org/job/nova/749/ | 15:09 |
| openstackjenkins | Tarmac: Fixes metadata for ec2_api to specify owner_id so that it filters properly. | 15:09 |
| *** reldan has joined #openstack | 15:09 | |
| markwash | termie: if you're in, can you see if I have addressed your concerns on my merge prop? https://code.launchpad.net/~rackspace-titan/nova/change-password-v1-1/+merge/54917 | 15:09 |
| *** mgoldmann has quit IRC | 15:13 | |
| openstackjenkins | Project nova build #750: SUCCESS in 2 min 30 sec: http://hudson.openstack.org/job/nova/750/ | 15:16 |
| openstackjenkins | * Tarmac: Moving server update adminPass support to be v1.0-specific | 15:16 |
| openstackjenkins | OS API servers update tests actually assert and pass now | 15:16 |
| openstackjenkins | Enforcing server name being a string of length > 0 | 15:16 |
| openstackjenkins | * Tarmac: Adding Content-Type code to openstack.api.versions.Versions wsgi.Application | 15:16 |
| *** pharkmillups has joined #openstack | 15:16 | |
| *** OliG has quit IRC | 15:16 | |
| *** uksysadmin has quit IRC | 15:22 | |
| *** dendro-afk is now known as dendrobates | 15:28 | |
| n1md4 | on the node I only have nova-common | 15:32 |
| n1md4 | ...and nova-compute. Is this enough to build the node successfully? | 15:32 |
| *** syn__ has joined #openstack | 15:34 | |
| *** ccustine has joined #openstack | 15:34 | |
| *** syn__ has quit IRC | 15:34 | |
| *** dendrobates is now known as dendro-afk | 15:34 | |
| *** dendro-afk is now known as dendrobates | 15:34 | |
| *** kapil has quit IRC | 15:34 | |
| *** adiantum has quit IRC | 15:35 | |
| *** kapil has joined #openstack | 15:37 | |
| *** maplebed has joined #openstack | 15:38 | |
| *** gaveen has joined #openstack | 15:42 | |
| openstackjenkins | Project nova build #751: SUCCESS in 2 min 28 sec: http://hudson.openstack.org/job/nova/751/ | 15:43 |
| openstackjenkins | Tarmac: Key was converted through str() even if None, resulting in "None" being | 15:43 |
| openstackjenkins | added to authorized_keys when no key was specified. | 15:44 |
| *** aloga has quit IRC | 15:44 | |
| *** aloga has joined #openstack | 15:44 | |
| *** aloga has quit IRC | 15:45 | |
| *** aloga has joined #openstack | 15:46 | |
| *** adiantum has joined #openstack | 15:51 | |
| *** pharkmillups has quit IRC | 15:52 | |
| *** niksnut has joined #openstack | 15:53 | |
| mtaylor | devcamcar: fixed it! | 15:57 |
| mtaylor | devcamcar: lp:~mordred/nova-adminclient/fix-dist | 15:57 |
| n1md4 | can someone paste the output of 'dpkg -l | grep nova' on a working compute node please? | 15:59 |
| blamar | n1md4: http://paste.openstack.org/show/1034/ | 16:01 |
| n1md4 | blamar: cheers | 16:02 |
| *** gregp76 has joined #openstack | 16:04 | |
| *** spring has quit IRC | 16:05 | |
| *** ppetraki has quit IRC | 16:05 | |
| n1md4 | blamar: having now installed the other nova services (I only had nova-common and compute), nova-scheduler and network and now entries in the services table, is that right? | 16:05 |
| n1md4 | I thought only nova-compute was supposed to be present | 16:06 |
| blamar | n1md4: sorry, that might not have been what you wanted...that node I pasted has multiple jobs, not just compute :\ | 16:07 |
| n1md4 | blamar: don't worry, I'm still breaking things here. Trying to get it to work | 16:08 |
| n1md4 | http://paste.openstack.org/show/1035/ | 16:09 |
| n1md4 | Still can't get it right! :( | 16:09 |
| *** photron has quit IRC | 16:14 | |
| *** MotoMilind has joined #openstack | 16:17 | |
| n1md4 | blamar: have you got an output of one that is only being a node? :) | 16:19 |
| blamar | n1md4: I don't at the moment, no.. sorry I can't be of more help but I'm just learning a lot of the internals myself. If you need some help with the OS API I'd love to be more assistance heh | 16:20 |
| *** dubsquared has joined #openstack | 16:20 | |
| n1md4 | i'm having enough trouble getting the nodes to work with the controllers, I'm sure there'll just be a couple of lines of config that I'm messing up, but I can't yet find the needle ;) | 16:21 |
| *** pharkmillups has joined #openstack | 16:24 | |
| *** MotoMilind has quit IRC | 16:24 | |
| *** lionel has quit IRC | 16:25 | |
| justinsb | Hi guys - Python style question. I need to create a base class for _every_ OpenStack API Controller. I wanted to call it openstack.Controller instead of wsgi.Controller, but I'm not sure if I then fall foul of Python style. Should I put it in __init__.py? In a new file called nova/api/openstack/openstack.py? | 16:25 |
| justinsb | (This is to support XML namespacing, not for type-checking, before anyone asks :-) ) | 16:26 |
| *** bkkrw has quit IRC | 16:26 | |
| *** johnpur has joined #openstack | 16:27 | |
| *** ChanServ sets mode: +v johnpur | 16:27 | |
| markwash | justinsb: that sounds like a lot of work--I think we are coming up against the limitations of the template method pattern | 16:27 |
| *** lionel has joined #openstack | 16:28 | |
| justinsb | markwash: What's the alternative you're suggesting? | 16:28 |
| markwash | justinsb: one sec | 16:29 |
| markwash | justinsb: so its not 100% clear to me, but I think basically we should factor the deserialization (and probably the wsgification) out of the controllers | 16:30 |
| markwash | create a strategy that is a wsgified callable class | 16:30 |
| markwash | which upon instantiation is given a deserializer and controller | 16:30 |
| n1md4 | "Error: operation failed: failed to retrieve chardev info in qemu with 'info chardev'" and "qemu: could not load kernel '/var/lib/nova/instances/instance-00000014/kernel': Inappropriate ioctl for device" on different nodes, any ideas? I'm I commenting in the right place too, I do get ignored a lot ;) | 16:30 |
| markwash | the latter _not_ inherited from wsgi.Controller | 16:30 |
| n1md4 | s/I'm/Am/ | 16:31 |
| markwash | in any case, this gives us an opportunity to inject an os api specific deserializer | 16:31 |
| *** MotoMilind has joined #openstack | 16:31 | |
| markwash | justinsb: this is probably as much if not more work | 16:31 |
| justinsb | markwash: Good suggestion... I think though that this is already done in wsgi.Controller | 16:31 |
| justinsb | markwash: It creates a Serializer object | 16:31 |
| justinsb | markwash: Or do you mean going even further? | 16:32 |
| markwash | justinsb: I agree--however the problem is that serializer is given to the controllers through inheritance | 16:32 |
| justinsb | markwash: Ah, I think I get it now | 16:32 |
| *** s1cz has quit IRC | 16:32 | |
| markwash | thus in order to give a new one, you have to modify the inheritance hierarchy | 16:32 |
| *** maplebed has quit IRC | 16:33 | |
| *** s1cz has joined #openstack | 16:33 | |
| markwash | justinsb: honestly, I'm about to start working on a blueprint to refactor how the serialization works in openstack api | 16:33 |
| justinsb | markwash: I think we've chosen inheritance for this now | 16:33 |
| justinsb | markwash: No objection to a blueprint | 16:33 |
| markwash | so we might want to just do something expedient here | 16:33 |
| *** gregp76 has quit IRC | 16:33 | |
| justinsb | markwash: But I just need XML namespaces in there... | 16:33 |
| justinsb | I think we likely have a lot of hidden XML problems, and I think we need to get the namespace in asap | 16:34 |
| markwash | I don't think there is going to be any "right" way to do this short of a larger refactoring, unfortunately | 16:34 |
| markwash | will the namespaces help flush out the xml problems? | 16:35 |
| justinsb | markwash: Thanks. I think I'll probably just do it "my way" and wait for someone to shoot me down in the MP :-) | 16:35 |
| markwash | justinsb: lol :-) good hunting | 16:35 |
| justinsb | markwash: Yes... without a NS, we can't parse the XML against the schema | 16:35 |
| justinsb | markwash: Well, I could inject it, but I _thought_ it would be easier just to fix the OS API :-) | 16:35 |
| justinsb | markwash: Inject it client-side, that is | 16:36 |
| markwash | lol :-) | 16:36 |
| *** adjohn has joined #openstack | 16:37 | |
| justinsb | I'm putting it into common - seems the right place for it for now | 16:42 |
| *** joearnold has joined #openstack | 16:42 | |
| *** joearnold has quit IRC | 16:43 | |
| *** maplebed has joined #openstack | 16:43 | |
| *** adjohn has quit IRC | 16:45 | |
| *** Ryan_Lane has joined #openstack | 16:47 | |
| *** adjohn has joined #openstack | 16:49 | |
| *** kapil has quit IRC | 16:52 | |
| *** zenmatt has quit IRC | 16:53 | |
| *** adjohn has quit IRC | 16:54 | |
| *** zenmatt has joined #openstack | 16:54 | |
| *** Nacx has quit IRC | 16:59 | |
| *** gregp76 has joined #openstack | 16:59 | |
| *** nelson has joined #openstack | 16:59 | |
| *** bcwaldon has joined #openstack | 17:00 | |
| n1md4 | I'm currently using a localuser image, I assume from launchpad Q+A this will not work for a multi-node install, what image should I use? | 17:01 |
| *** Ryan_Lane1 has joined #openstack | 17:01 | |
| *** maple_bed has joined #openstack | 17:04 | |
| *** maplebed has quit IRC | 17:04 | |
| *** Ryan_Lane has quit IRC | 17:05 | |
| *** sparkycollier has joined #openstack | 17:09 | |
| *** keds has joined #openstack | 17:10 | |
| cjreyn | hey guys, | 17:11 |
| cjreyn | Im having trouble getting Glance and Euca-describe-images to work | 17:11 |
| cjreyn | I've uploaded an image to Glance, and it shows up in Glance details | 17:11 |
| cjreyn | but when queried via euca-describe-images, I get "An unknown error has occurred." | 17:12 |
| cjreyn | The database lookup happens on the Glance server (I can see the log print) | 17:12 |
| cjreyn | but nova-api just dumps "File "/usr/lib/pymodules/python2.6/nova/api/ec2/cloud.py", line 863, in _image_ec2_id" | 17:13 |
| cjreyn | in the trace | 17:13 |
| cjreyn | this is the latest ppa. I know jologhlin was having the same error | 17:14 |
| justinsb | cjreyn: Could you please pastie the full trace from nova-api. It sounds like a known issue, albeit one we thought was fixed. | 17:16 |
| cjreyn | justinsb: http://pastebin.com/k1GZDBcV | 17:18 |
| *** maple_bed has quit IRC | 17:19 | |
| justinsb | cjreyn: Looks like the same one... I'm finding the links now | 17:19 |
| *** maplebed has joined #openstack | 17:19 | |
| *** MotoMilind has quit IRC | 17:19 | |
| cjreyn | justinsb: cheers. I treid both disk images, kernels, and initrd's | 17:20 |
| justinsb | cjreyn: Here's the fix I proposed: https://code.launchpad.net/~justin-fathomdb/nova/ec2-api-with-glance/+merge/54288 | 17:21 |
| justinsb | cjreyn: And here's a related bug: https://bugs.launchpad.net/nova/+bug/742113 | 17:21 |
| uvirtbot | Launchpad bug 742113 in nova "EC2 API doesn't work well with arbitrary glance images" [Medium,In progress] | 17:21 |
| justinsb | cjreyn: I guess the bug is still there :-( | 17:22 |
| *** maplebed has quit IRC | 17:22 | |
| *** maplebed has joined #openstack | 17:22 | |
| justinsb | cjreyn: The problem is that (some versions of) glance disagrees with (some versions of) nova on what the image format should be, and (some versions of) image upload tools upload with unexpected metadata | 17:23 |
| justinsb | cjreyn: How did you upload your image? | 17:23 |
| *** lionel has quit IRC | 17:23 | |
| cjreyn | justinsb: using the cmd: glance add name="PBS_NODE_debian_v1.4.img" is_public=True container_format=bare disk_format=raw < PBS_NODE_debian_v1.4.img -H 172.16.10.4 | 17:23 |
| *** lionel has joined #openstack | 17:23 | |
| justinsb | cjreyn: Ah - there's some magic that you need there... let me dig it up | 17:24 |
| cjreyn | justinsb: ok cool. Does the magic relate to uploading kernels and initrd's aswell? | 17:24 |
| justinsb | cjreyn: Well the bug right now is that even one malformed image will block euca-describe-images (unless it's been fixed) | 17:25 |
| cjreyn | justinsb, the images are all rqaw format, i.e. unmodified kernels initrds and diska images with no metadat | 17:26 |
| cjreyn | justinsb, srry "raw" | 17:26 |
| justinsb | cjreyn: I just use raw disk images (no kernels / initrds) so that's all I know... sorry | 17:27 |
| justinsb | cjreyn: I think you need to pass a "type=ami" argument to glance so that the type is populated in the properties | 17:27 |
| *** MotoMilind has joined #openstack | 17:27 | |
| justinsb | cjreyn: You'll also have to delete your 'bad' image from glance | 17:27 |
| justinsb | cjreyn: I've also seen a recommendation to upload using the 'proper' tool rather than glance directly | 17:28 |
| cjreyn | justinsb, ah ok makes sense. Maybe passing type="ari" will sort out the ramdisk image. | 17:28 |
| cjreyn | justinsb, whats the 'proper' tool? | 17:28 |
| justinsb | cjreyn: " If you want images to work with ec2, you | 17:28 |
| justinsb | should probably upload them with nova-manage image commands." | 17:28 |
| justinsb | cjreyn: That's what vishy said on my merge proposal | 17:28 |
| justinsb | cjreyn: The values should be type=machine, type=kernel and type=ramdisk | 17:29 |
| justinsb | cjreyn: Based on looking at the code | 17:29 |
| justinsb | cjreyn: They get mapped to ami, aki, ari | 17:29 |
| justinsb | cjreyn: (It's that mapping that's fragile, BTW) | 17:29 |
| *** joearnold has joined #openstack | 17:30 | |
| cjreyn | justinsb, ok cool. I'm not sure using nova-manage solves the problem. I was speaking with a guy who had the same problem and he pasted me his upload. It included the 'type': 'machine' entry | 17:31 |
| termie | markwash: checking it out | 17:31 |
| cjreyn | justinsb, ok, ill try the explict setting of the properties, and if its broke, report a bug. | 17:32 |
| justinsb | cjreyn: Yes, great, thanks. That's a bit worrying if specifying type=machine still fails | 17:32 |
| cjreyn | justinsb, cheers for the help btw | 17:32 |
| justinsb | cjreyn: np | 17:32 |
| *** kashyap has quit IRC | 17:32 | |
| termie | markwash: commented | 17:33 |
| *** reldan has quit IRC | 17:35 | |
| *** MarcMorata has joined #openstack | 17:37 | |
| *** dragondm has quit IRC | 17:40 | |
| *** joearnold has quit IRC | 17:40 | |
| *** joearnold has joined #openstack | 17:40 | |
| *** reldan has joined #openstack | 17:42 | |
| cjreyn | justinsb, well adding type=machine didn't work. Same error | 17:45 |
| justinsb | cjreyn: Bugger | 17:45 |
| cjreyn | justinsb, bug I guess? | 17:46 |
| justinsb | cjreyn: Can you please pastie the output of your glance images details... | 17:46 |
| justinsb | cjreyn: Just getting the URL. Likely a bug! | 17:46 |
| justinsb | cjreyn: Well, I think it's a bug if it's possible to break it just by uploading bad metadata, but that's a different issue... | 17:46 |
| justinsb | cjreyn: URL is /images/details Probably http://127.0.0.1:9292/images/details | 17:46 |
| cjreyn | justinsb, URI: http://172.16.10.4/images/7 | 17:47 |
| cjreyn | Id: 7 | 17:47 |
| cjreyn | Public: Yes | 17:47 |
| cjreyn | Name: PBS_NODE_debian_v1.4.img | 17:47 |
| cjreyn | Size: 1049624576 | 17:47 |
| cjreyn | Location: file:///srv/glance/images/7 | 17:47 |
| cjreyn | Disk format: raw | 17:47 |
| cjreyn | Container format: bare | 17:47 |
| cjreyn | justinsb, it only shows up in the properties field if I dry-run the glance add cmd | 17:47 |
| cjreyn | justinsb, should details show the properties fields? | 17:48 |
| justinsb | cjreyn: I think it only shows up in details. Can you pastie that (it's likely to be big) | 17:48 |
| cjreyn | justinsb, that was it, after running the cmd glance details --verbose -H 172.16.10.4 | 17:48 |
| justinsb | cjreyn: Can you do a curl http://172.16.10.4:9292/images/details | 17:49 |
| justinsb | cjreyn: It gets the raw JSON data | 17:49 |
| *** drico has joined #openstack | 17:51 | |
| vishy | disk format conatainer format should both be ami for ec2 | 17:53 |
| justinsb | vishy: Do you think this should work? I don't see how it can... | 17:56 |
| justinsb | I just commented on here: https://code.launchpad.net/~justin-fathomdb/nova/ec2-api-with-glance/+merge/54288 | 17:56 |
| cjreyn | vishy, I'm uploading a raw disk image. Tried both ami types, and base/raw, and euca-describe-instances fails | 17:56 |
| justinsb | cjreyn: Any luck on that curl output? It would help us here I think... | 17:57 |
| *** kapil has joined #openstack | 17:57 | |
| vishy | cjreyn: and you have type set in properties? | 17:57 |
| cjreyn | vishy, type=machine | 17:58 |
| *** kapil has quit IRC | 17:58 | |
| *** kapil has joined #openstack | 17:58 | |
| openstackjenkins | Project nova build #752: SUCCESS in 2 min 31 sec: http://hudson.openstack.org/job/nova/752/ | 17:59 |
| openstackjenkins | Tarmac: OpenStack API faults have been changed to now return the appropriated Content-Type header. | 17:59 |
| markwash | termie: thanks for the comments | 17:59 |
| markwash | termie: I think that for now I would like to keep the mock class defined in the function since it does not need to be used anywhere else | 17:59 |
| termie | markwash: most other tests define their mocks outside of the test itself | 18:00 |
| *** kapil has quit IRC | 18:00 | |
| termie | markwash: for things that are just classes | 18:00 |
| termie | markwash: i can dig up a couple examples | 18:00 |
| *** dendrobates is now known as dendro-afk | 18:00 | |
| *** dendro-afk is now known as dendrobates | 18:00 | |
| markwash | termie: cool, I'll look too | 18:00 |
| cjreyn | justinsb, just sorting the network. Glance is running in a non routed VM at the mo... 2 secs | 18:00 |
| *** kapil has joined #openstack | 18:01 | |
| justinsb | cjreyn: No worries.. If you're able to run glance details though, you should be able to run curl from the same machine | 18:01 |
| *** kapil has quit IRC | 18:01 | |
| termie | markwash: test_api, test_compute, test_direct, test_rpc, test_service | 18:01 |
| vishy | cjreyn: it is properties: {'type': 'machine'} | 18:02 |
| termie | markwash: test_scheduler | 18:02 |
| markwash | termie: I think perhaps I was copying test_server_rebuild, test_delete-server_instance, test_resize_* all in the test_servers.py file | 18:02 |
| termie | markwash: none of those appear to have any classes defined in them? | 18:03 |
| markwash | termie: oh, my class is actually a function | 18:03 |
| cjreyn | vishy, yes that shows on the dry run, but not on a print via glance details | 18:03 |
| vishy | cjreyn: regardless this needs to be fixed | 18:03 |
| markwash | termie: that's right, that's how it happened, it started as a function, and then I made it a callable | 18:04 |
| termie | markwash: it is a callable, yes | 18:04 |
| markwash | termie: but what is the difference? | 18:04 |
| vishy | justinsb, cjreyn: probably we should key type off of container format and default bare to ami | 18:04 |
| termie | markwash: nothing much, just that if it is the kind of thing that could be useful outside that method it is effectively hidden by being inside it | 18:04 |
| markwash | termie: ah, I don't think it is useful outside of that | 18:05 |
| termie | markwash: it's not a big issue obvs, just looked unusual to me | 18:05 |
| *** zaitcev has joined #openstack | 18:05 | |
| markwash | termie: at least not with the current tests | 18:05 |
| termie | markwash: aight, well i guess let's leave it there | 18:05 |
| justinsb | vishy: Do the other image stores all define container_format always? | 18:05 |
| justinsb | vishy: Could we just default to ami always instead of axi? | 18:05 |
| justinsb | vishy: (I proposed axi as the minimal fix) | 18:06 |
| vishy | justinsb: seems ok | 18:06 |
| markwash | termie: thanks again for the comments | 18:06 |
| vishy | justinsb; we don't really need properties type now that container format is in | 18:06 |
| *** dmshelton has quit IRC | 18:07 | |
| vishy | justinsb: we should key off of container_format ari -> prefix ari type ramdisk | container_format aki -> prefix aki type kernel | all others -> prefix ami type machine | 18:08 |
| *** BinaryBlob has joined #openstack | 18:09 | |
| *** dmshelton has joined #openstack | 18:09 | |
| justinsb | vishy: I'm a bit bored of this bug... I've already proposed fixes for it twice :-( | 18:09 |
| justinsb | vishy: I'm just checking if the other image services return container_format | 18:10 |
| vishy | justinsb: the only other service is local | 18:10 |
| cjreyn | justinsb, srry, curl wasn;t installed and my ethernet adapter went down! Ok,so it returns a 404 | 18:10 |
| vishy | and it returns whatever is set (doesn't check) | 18:10 |
| justinsb | vishy: And S3 service, right? | 18:11 |
| justinsb | vishy: S3 seems to populate it | 18:11 |
| vishy | s3 just wraps the other services | 18:11 |
| vishy | it populates the metadata on image create | 18:11 |
| vishy | justinsb: if you don't want to mess with it, I can propose a fix | 18:12 |
| *** gaveen_ has joined #openstack | 18:12 | |
| justinsb | vishy: That would be great. I've tried twice already now... | 18:12 |
| justinsb | vishy: And I have to go to the post office :-) | 18:13 |
| vishy | ok i'm on it | 18:13 |
| justinsb | Thanks! | 18:13 |
| *** sparkycollier has quit IRC | 18:13 | |
| cjreyn | vishy, is there a workaround so that I can add a kernel, ramdisk and image? All raw/no meta-data | 18:14 |
| jk0 | would two cores mind looking at https://code.launchpad.net/~jk0/nova/lp744833/+merge/55604 for me please? | 18:14 |
| cjreyn | vishy, without crashing euca-describe-images | 18:15 |
| vishy | cjreyn: it should work if type is in properties, not sure why it isn't adding type properly | 18:15 |
| *** _0x44 has quit IRC | 18:15 | |
| *** gaveen has quit IRC | 18:16 | |
| cjreyn | vishy, and for a raw disk image the container and disk format should be ami | 18:16 |
| cjreyn | vishy, ? | 18:16 |
| vishy | in theory yes | 18:17 |
| dendrobates | jk0: done | 18:17 |
| jk0 | thanks dendrobates | 18:17 |
| *** _0x44 has joined #openstack | 18:18 | |
| *** ChanServ sets mode: +v _0x44 | 18:18 | |
| *** nelson has quit IRC | 18:20 | |
| *** nelson has joined #openstack | 18:20 | |
| dprince | jaypipes: What are your thoughts on quotas in Glance? | 18:21 |
| dprince | jaypipes: specifically for properties. | 18:21 |
| dprince | I'm looking at doing https://bugs.launchpad.net/nova/+bug/742809 | 18:21 |
| uvirtbot | Launchpad bug 742809 in nova "OSAPI: enforce quota for number of image metadata items" [Medium,In progress] | 18:21 |
| dprince | Which can certainly be done in nova but that may not be the right place for it. | 18:21 |
| *** metoikos has quit IRC | 18:22 | |
| *** localhost has quit IRC | 18:23 | |
| *** localhost has joined #openstack | 18:23 | |
| *** kashyap has joined #openstack | 18:24 | |
| *** sparkycollier has joined #openstack | 18:28 | |
| btorch | is there a problem with nova after bzr 866 ? If I roll back to that version everything works fine on my CN but if I upgrade to the latest 900+ the same instance (maverick-server-uec-amd64) that I can run fine on 866 won't run anymore | 18:30 |
| btorch | either some network or iptables rules inserted are screwing up with the snat rules to reach http://169.254.169.254 and causing that url error | 18:38 |
| *** joearnol_ has joined #openstack | 18:40 | |
| *** joearnold has quit IRC | 18:40 | |
| dubs | dprince: what is the motivation for implementing those quotas? | 18:40 |
| *** clauden has joined #openstack | 18:41 | |
| dprince | dubs: Well. I guess you could consider it a security risk. | 18:43 |
| dubs | dprince: how so? | 18:43 |
| dprince | dubs: Anyone with an API key could fill up the glance image_properties table. | 18:43 |
| dubs | can't that same person upload arbitrarily large images as well? | 18:44 |
| dprince | dubs: I thought we had a ticket on that too. :) | 18:44 |
| dubs | heh | 18:44 |
| *** ppetraki has joined #openstack | 18:44 | |
| dprince | dubs: Hmm. Well. The images aren't store in the DB anyway. | 18:45 |
| pvo | dprince: but they could fill whatever backend glance is using | 18:45 |
| dubs | not trying to be confrontational, just seems that Glance is a service that already allows people to upload large amounts of untrusted data. i'm not sure limiting the number of properties on an image solves any security issue | 18:45 |
| dprince | dubs: So if someone wants to upload massive images into S3 or Swift I suppose they'd pay for it somehow. | 18:46 |
| dprince | Sure. | 18:46 |
| *** dragondm has joined #openstack | 18:46 | |
| dubs | dprince: that's true, there is a monetary cost in most cases (though not all — public images). | 18:46 |
| *** sparkycollier has quit IRC | 18:46 | |
| dprince | I'm happy to punt on this for now if we need to talk out a longer term solution. | 18:47 |
| *** gregp76 has quit IRC | 18:47 | |
| dprince | In the short term I was just trying to prevent our image metadata controller in the OSAPI from exposing the data. | 18:48 |
| dubs | dprince: jaypipes and sirp may have more thoughts on this. I just spoke up to try to understand what problem it would be solving. I'll keep quiet now :) | 18:48 |
| *** arun_ has quit IRC | 18:48 | |
| dprince | It kind of seems like Glance may be the right place for a quota mechanism however. | 18:48 |
| cjreyn | vishy, ok, so I tried adding a kernel image through both the glance cli, and through nova image image-create. I checked the mysql glance database and type is set to kernel. | 18:48 |
| *** joearnol_ has quit IRC | 18:48 | |
| jaypipes | dprince: sorry, was on phone.... | 18:48 |
| dprince | dubs: Sure. Thanks for the input. Much appreciated. | 18:48 |
| dubs | dprince: there may be good reasons for it. better security is probably not one of them | 18:48 |
| jaypipes | dprince: I would prefer to make any quotas a middleware in Glance that can easily be excluded. | 18:49 |
| cjreyn | vishy, running euca-describe-instances returns the original error I pastebin dumped earlier (line 863 in cloud.py) | 18:49 |
| deepy | Can I try out nova in virtualbox if I lack VT-x and such? | 18:49 |
| dprince | jaypipes: Okay. I'd go for that. So is this something to punt on for now then? I mean we are in feature freeze for GLance too right. | 18:50 |
| jaypipes | dprince: no, it should be a trivial middleware after https://blueprints.launchpad.net/glance/+spec/api-limited-results is done in early Diablo. | 18:50 |
| jaypipes | dprince: yes, FF in Glance too. | 18:50 |
| dprince | jaypipes: gotcha. When I said punt I just meant for Cactus. So it shoulds like maybe we should make this ticket a Glance ticket then and put it in for diablo? | 18:51 |
| *** gregp76 has joined #openstack | 18:52 | |
| jaypipes | dprince: well, Glance work would just be making it more efficient (i.e. passing the quote limit all the way down to the registry DB API as a LIMIT X, X. but OpenStack doesn't depend on Glance, and therefore quotas in the OS API need to be enforced on the Nova side for image services other than Glance, no? | 18:54 |
| jaypipes | s/OpenStack doesn't depend/Nova doesn't depend/ ^^ | 18:54 |
| dprince | jaypipes: The options are really local image service or glance right? The S3 image service (used by the EC2 API now wraps either local or glance too). | 18:56 |
| *** daveiw has joined #openstack | 18:56 | |
| btorch | anyone has an image that I could use that does not pull anything from " http://169.254.169.254/2009-04-04/meta-data/instance-id" | 18:57 |
| justinsb | jaypipes: Isn't glance going to be our images API in OpenStack? | 18:57 |
| dprince | jaypipes: And how likely is it that someone would deploy with a local image service instead of using Glance? | 18:57 |
| jaypipes | justinsb: ? totally separate. | 18:57 |
| dprince | jaypipes: I mean I can put this in the API sure. Initially I filed the ticket with that in mind. | 18:57 |
| justinsb | jaypipes: I thought it was the first devolved service? Like volumes is going to be its own API and service | 18:58 |
| vishy | dreyn: describe instances or describe images? | 18:58 |
| dprince | jaypipes: After talking about it a bit more that seems like it may not be the best spot. | 18:58 |
| vishy | dprince, jaypipes almost zero for deployments | 18:58 |
| justinsb | jaypipes: So there will be no /images in the openstack API long term | 18:58 |
| justinsb | jaypipes: We'll just hit glance directly | 18:58 |
| jaypipes | dprince, justinsb: the point of pulling glance out was to create a stand-alone image delivery/registry service. making the OpenStack /images and quote system somehow intrinsic to Glance is counter-productive to making decoupled components. | 18:59 |
| dprince | jaypipes: I could certianly bang out a patch to make this work for cactus (as a bug fix) and then we can revert it for Glance. If we aren't concerned about the security aspects of someone deploying with it then why not just wait and do it in Glance Diablo. | 18:59 |
| justinsb | jaypipes: You've lost me. Isn't the point of making a standalone service that clients would hit it directly rather than proxying through the OpenStack API? | 19:00 |
| *** maplebed has quit IRC | 19:00 | |
| justinsb | jaypipes: That's what I thought everyone's been saying on the ML | 19:00 |
| jaypipes | justinsb: we're talking about the OS API /images endpoint, which has limitations (quotas) on it that have nothing to do with Glance. | 19:00 |
| justinsb | jaypipes: But there will be no /images endpoint; it will be Glance, right? | 19:01 |
| jaypipes | justinsb: Glance is the image service. It has it's own REST API. That REST API does not correspond to the RS CloudServers or OpenStack 1.1 API. | 19:01 |
| justinsb | jaypipes: Agreed. But that REST API will be the only API that is used for images at OS API v1.X where X>=2 | 19:02 |
| *** kashyap has quit IRC | 19:02 | |
| jaypipes | justinsb: that's not up to me. I have no idea about that. | 19:02 |
| *** pharkmillups has quit IRC | 19:03 | |
| jaypipes | justinsb: the point is that someone can stand up Glance as an image service without knowing a thing about Nova or the CloudServers/OpenStack Compute API. | 19:03 |
| justinsb | jaypipes: Well, that's what all those long API conversations on the ML were saying | 19:03 |
| justinsb | jaypipes: Sure, they can stand up glance standalone if they want to | 19:03 |
| jaypipes | justinsb: link? | 19:03 |
| *** brd_from_italy has joined #openstack | 19:04 | |
| justinsb | jaypipes: Oh great, now I have to re-read that thread :-) | 19:04 |
| jaypipes | justinsb: the point is that Quotas have nothing to do with /images. They are a Rackspace-specific part of the Rackspace API. | 19:04 |
| justinsb | jaypipes: Which, unfortunately for you, is the Glance API | 19:04 |
| justinsb | jaypipes: Working on the link! | 19:05 |
| jaypipes | justinsb: I don't know what you're talking about. the Glance API is not the Rackspace API. | 19:05 |
| *** Ryan_Lane1 is now known as Ryan_Lane | 19:05 | |
| justinsb | jaypipes: Don't shoot the messenger! | 19:05 |
| jaypipes | wtf are people talking about here... I thought it was a perfectly clear concept that quotas have ZERO to do with images. | 19:06 |
| dprince | The quotas I initially brought up where image property quotas. | 19:06 |
| uvirtbot | New bug: #745965 in swift "Lookback Command Line Options For Log Processor Daemon" [Undecided,New] https://launchpad.net/bugs/745965 | 19:07 |
| *** gregp76 has quit IRC | 19:07 | |
| eday | jaypipes: so, just trying to understand... if I wanted to use glance as a stand-alone image service, who would do the quota checking? | 19:08 |
| jk0 | termie: mind having another quick looksee at that merge? | 19:09 |
| justinsb | https://lists.launchpad.net/openstack/msg00665.html https://lists.launchpad.net/openstack/msg00786.html | 19:09 |
| justinsb | jaypipes: Those are the links | 19:10 |
| justinsb | jaypipes: But it was one of those meandering threads | 19:10 |
| jaypipes | dprince: that has nothing to do with Glance, though. That is the Cloud Servers API... the *compute* API. Glance will support limiting results using, say, ?offset=X&limit=Y, but in no way is the Glance API supposed to be a mirror of the Cloud Servers /images API, along with concepts that only apply to Cloud Servers API, like quotas. | 19:10 |
| justinsb | jaypipes: That's not what I interpreted from Jorge and Pvo's vision | 19:11 |
| *** BinaryBlob has quit IRC | 19:11 | |
| justinsb | jaypipes: Sounds like a design summit topic | 19:11 |
| justinsb | jaypipes: But I think you're going to end up with quotas unless you're very lucky ;-) | 19:12 |
| cjreyn | btorch, how did you roll back? | 19:13 |
| jaypipes | justinsb: you wrote "A better approach is to simply provide a service catalog with list of endpoints, you can easily detect whether a volume service is available etc". while that is true, that's pointless unless each endpoint can describe itself and its API. Furthermore, if the /images endpoint is always supposed to be the same API, what is the point of having separate services publishing different APIs to begin with? | 19:14 |
| justinsb | jaypipes: Sadly I think I lost the argument, so there's not much point quoting me back to me :-) | 19:14 |
| justinsb | jaypipes: I agree that the "why" was very much missing from the thread | 19:15 |
| *** sparkycollier has joined #openstack | 19:15 | |
| jaypipes | justinsb: I don't see how pvo's end vision in http://wiki.openstack.org/api_transition is at all contradictory to what I am saying above... that quotas (which are part of the Public API) have nothing to do with API of the image service. | 19:17 |
| vishy | sirp_: are you here? | 19:17 |
| dprince | jaypipes: You gave me the info I need for now. I'll work up a quota patch for the OSAPI for this specific issue (which is in the SPEC). If Glance middleware eventually evolves into something cleaner maybe we can make use of that at some point in the future. | 19:17 |
| dprince | Sorry for the confusion. Just wanted to ask about Glance quota plans before I started in on things. | 19:18 |
| justinsb | jaypipes: The way I read pvo's "proxy" was as a dumb proxy. I think you'If you read that as a "smart proxy" then | 19:18 |
| justinsb | jaypipes: Whoops... early enter key | 19:18 |
| jaypipes | justinsb: in other words, the public API controller is the one that a) knows what the quotas are and b) updates those quotas. The image service shouldn't do either. The public API controller will send a message (REST or otherwise) to the image service containing the value of the quota (a limit), but the image service itself doesn't know what the quotas for the API are set to... | 19:19 |
| vishy | jaypipes, justinsb: general question about images. What container format should a whole-disk-image have? Bare? | 19:19 |
| jaypipes | vishy: yes | 19:19 |
| justinsb | jaypipes: I read pvo's "proxy" as dumb. You're reading it as "smart". Pistols at the design summit :-) | 19:19 |
| jaypipes | justinsb: :) ok. | 19:19 |
| jaypipes | justinsb: what? you're saying pvo's dumb? pvo! | 19:20 |
| * jaypipes runs and hides. | 19:20 | |
| justinsb | jaypipes: Well, if pvo's around, I suppose he can clarify his thoughts for our benefit :-) | 19:20 |
| justinsb | jaypipes: But we really do need a design summit discussion to figure out where we're going here and make sure everyone's on the same page | 19:20 |
| justinsb | jaypipes: Because if nova assumes glance is doing it, and glance assumes nova is doing it, that's not really going to work! | 19:21 |
| jaypipes | justinsb: my general theory is this: if it doesn't have to do with images, it doesn't belong in glance. | 19:21 |
| justinsb | jaypipes,vishy: Oh, the container_format for a whole disk image, are we using compression? | 19:22 |
| justinsb | jaypipes, vishy: Because I tend to use an 8G image which compresses down to say 250MB | 19:22 |
| jaypipes | justinsb: no, vishy was just referring to what to set the container_format attribute of the image to... | 19:22 |
| sirp | vishy: hey | 19:22 |
| vishy | justinsb: you can use a disk format like qcow2 | 19:22 |
| justinsb | jaypipes: I know, I'm asking a related but different question :-) | 19:22 |
| vishy | sirp: hey i was just modifying a bit of your code | 19:22 |
| jaypipes | oh, sorry | 19:22 |
| justinsb | vishy: Any way to specify 'just the bits, maam' with compression? | 19:23 |
| justinsb | vishy: I'm asking for my notsovm code | 19:23 |
| cjreyn | justinsb, so shall I go ahead and post a bug report? I checked the database and the fields seem top be set correctly | 19:23 |
| justinsb | cjreyn: I think vishy is looking at it right now, hence his questions about container_format | 19:23 |
| *** reldan has quit IRC | 19:24 | |
| justinsb | cjreyn: But no harm in posting a bug report | 19:24 |
| justinsb | cjreyn: It'll presumably give vishy a bit more karma when he closes it | 19:24 |
| justinsb | cjreyn: And you may have some extra information to provide! | 19:24 |
| cjreyn | justinsb, ok cool. I had a working revision, pre-glance from ppa. Is there any way of installing from an old revision? | 19:24 |
| cjreyn | justinsb, karma eh!? | 19:24 |
| justinsb | cjreyn: No idea on the ppa stuff.. I tend to use the source. Sorry! | 19:25 |
| btorch | cjreyn: sorry got dragged to a meeting ... I installed the debs that are cached | 19:26 |
| cjreyn | btorch, now worries. ok, where are they on ubuntu? | 19:26 |
| btorch | cjreyn: under /var/cache/apt/archives/ just installed the old nova pkgs from there | 19:26 |
| justinsb | jaypipes, vishy: So no compression? :-( I know qcow2 is sparse, but it isn't compressed either, is it? | 19:27 |
| btorch | cjreyn: i got figure out what's up with this cause the only pkg I don't have cached is nova-volume and we need that now :( | 19:27 |
| vishy | justinsb: i think we could support compression, and have logic to decompress on the other end | 19:27 |
| justinsb | vishy: But not yet... maybe a feature request for Diablo | 19:28 |
| vishy | sirp: the lines are in _do_get_kernel_and_ramdisk | 19:28 |
| sirp | vishy: have a paste of the diff? | 19:28 |
| cjreyn | btorch, yeah i don't have that one listed. so it was as simple as purging the current pkg, then dpkg -i the cached versions? | 19:29 |
| btorch | cjreyn: yep | 19:29 |
| btorch | cjreyn: didn't purge anything | 19:29 |
| vishy | sirp: if image_meta['properties']['disk_format'] != 'ami': | 19:30 |
| vishy | sirp: changed it to if image_meta['container_format'] != 'ami': | 19:30 |
| btorch | cjreyn: just did a dpkg -i nova-doc_2011.2~bzr866-0ubuntu0ppa1~maverick1_all.deb ... and dpkg will take care of downgrading them | 19:30 |
| vishy | because the formats aren't in properties, and i think it is the container format that is important there no? | 19:30 |
| cjreyn | btorch, ok, nice | 19:30 |
| justinsb | vishy: how about image_meta.get('container_format', 'ami') != 'ami': for safety? | 19:30 |
| cjreyn | btorch, I had a pre-glance revison working. Then updated a compute node and it ruined the cloud controller | 19:31 |
| *** adiantum has quit IRC | 19:31 | |
| *** sebastianstadil has quit IRC | 19:31 | |
| vishy | justinsb: get seems good, i don't want to default it to ami there though, because it will try to grab the kernel and ramdisk for non ami images there | 19:31 |
| sirp | vishy, that's fine, we just need to update the ImageService contract so 'disk_format' and 'container_format' are required first-class attributes across the board, and then tweak the GlanceImageService to populate the field accordling..... assuming that's done, your patch should be good-to-go | 19:32 |
| justinsb | vishy: Sounds great! Thanks for taking this on! | 19:32 |
| *** gaveen_ has quit IRC | 19:35 | |
| vishy | sirp: yes there is a bug to update the docstrings on ImageService | 19:35 |
| vishy | sirp: my basic plan is to essentially have image service use glances default format | 19:35 |
| vishy | and just treat local image service as sort of a "fake" glance client | 19:35 |
| cjreyn | btorch, did you have to set up a new database? The problem I had was that the schema changed | 19:35 |
| cjreyn | vishy, justinsb its late here in the ukso I have to go. Is there anything you need from me re this problem? | 19:37 |
| btorch | cjreyn: no I didn't, but from version 866 to the 913 the db version stayed on 13. I don't know or have checked the kind of db schema changes take place but if it's only adding new tables or fields to a table having it at a higher version may not be an issue since the old code would not use the new schemas ( that just an aggressive guess) | 19:39 |
| btorch | cjreyn: to be safe though I would create a new DB and start from scratch anyway | 19:39 |
| cjreyn | btorch, ok cool. cheers. | 19:40 |
| vishy | justinsb, sirp, cjreyn: https://code.launchpad.net/~vishvananda/nova/fix-describe-images/+merge/55617 | 19:40 |
| uvirtbot | New bug: #745985 in nova "Support compressed raw images" [Undecided,New] https://launchpad.net/bugs/745985 | 19:41 |
| *** HouseAway is now known as AimanA | 19:43 | |
| *** jstinson has joined #openstack | 19:43 | |
| cjreyn | vishy, I've not looked at the source, but will defaulting to ami cause problems when uploading raw say kernel or ramdisks | 19:44 |
| *** Zangetsue has quit IRC | 19:49 | |
| *** Zangetsue has joined #openstack | 19:53 | |
| *** sparkycollier has quit IRC | 19:53 | |
| *** vernhart has quit IRC | 19:55 | |
| *** joearnold has joined #openstack | 19:58 | |
| *** dprince has quit IRC | 19:58 | |
| *** sparkycollier has joined #openstack | 20:04 | |
| *** Ryan_Lane is now known as Ryan_Lane|food | 20:05 | |
| *** brd_from_italy has quit IRC | 20:08 | |
| *** nerens has quit IRC | 20:09 | |
| *** nerens has joined #openstack | 20:09 | |
| *** sparkycollier has quit IRC | 20:10 | |
| *** westmaas has quit IRC | 20:14 | |
| *** reldan has joined #openstack | 20:15 | |
| *** nelson has quit IRC | 20:17 | |
| *** nelson has joined #openstack | 20:17 | |
| nelson | This is somewhat amusing: Example: | 20:18 |
| nelson | st -A https://auth.api.rackspacecloud.com/v1.0 -U user -K key stat | 20:18 |
| nelson | st: error: no such option: --A | 20:18 |
| nelson | Oh, --A. DOH! | 20:18 |
| creiht | haha | 20:19 |
| btorch | hmm it should just be -A unless the code has changed ... I'm using 1.0 | 20:19 |
| creiht | wait... nelson are you saying it thinks -A is --A, or did you accidently use --A? | 20:20 |
| *** vernhart has joined #openstack | 20:20 | |
| *** westmaas has joined #openstack | 20:21 | |
| *** westmaas has quit IRC | 20:21 | |
| *** westmaas has joined #openstack | 20:22 | |
| *** westmaas has quit IRC | 20:22 | |
| *** westmaas has joined #openstack | 20:23 | |
| *** gregp76 has joined #openstack | 20:24 | |
| *** pharkmillups has joined #openstack | 20:25 | |
| justinsb | Question: Is there anything in OSAPI v1.1 that is not also in v1.0 | 20:25 |
| uvirtbot | New bug: #746016 in nova "Limits response is not correct in OSAPI v1.1" [Undecided,New] https://launchpad.net/bugs/746016 | 20:26 |
| justinsb | That's the motivating bug for the question, BTW :-) | 20:26 |
| bcwaldon | justinsb: Yes, there is. Mostly changes | 20:26 |
| bcwaldon | justinsb: extensions comes to mind | 20:27 |
| justinsb | bcwaldon: Yes, I love extensions :-) | 20:27 |
| justinsb | bcwaldon: But anything else? | 20:27 |
| bcwaldon | do you have the specs? | 20:27 |
| justinsb | bcwaldon: I mean there's some element rearranging... | 20:27 |
| bcwaldon | justinsb: changePassword action | 20:27 |
| justinsb | bcwaldon: I do have the specs | 20:27 |
| vishy | cjreyn, if you are uploading kernels and ramdisks you should be setting container_format properly | 20:28 |
| justinsb | bcwaldon: Could we retrofit that into the v1.0 API if we had to? | 20:28 |
| justinsb | bcwaldon: I presume you can guess where I'm going with this... | 20:28 |
| *** vernhart has quit IRC | 20:28 | |
| bcwaldon | justinsb: how about you just tell me | 20:29 |
| justinsb | bcwaldon: I'm wondering if we're going to be able to support both v1.0 and v1.1 | 20:29 |
| justinsb | bcwaldon: By Cactus | 20:29 |
| bcwaldon | justinsb: not fully, but 1.0 isnt even supported yet | 20:29 |
| bcwaldon | justinsb: we already have the versioning code in | 20:30 |
| *** littleidea has quit IRC | 20:30 | |
| justinsb | bcwaldon: I know we have a lot of v1.1 code in there. I think it'll be a death march to get both v1.0 and v1.1 working in Cactus | 20:30 |
| justinsb | bcwaldon: because of issues like bug 746016 | 20:30 |
| uvirtbot | Launchpad bug 746016 in nova "Limits response is not correct in OSAPI v1.1" [Undecided,New] https://launchpad.net/bugs/746016 | 20:31 |
| bcwaldon | justinsb: I agree there is a lot of work to be done to separate the versions where they need to be | 20:31 |
| bcwaldon | justinsb: but it is just a matter of getting it done, it isn't necessarily difficult | 20:31 |
| *** littleidea has joined #openstack | 20:32 | |
| justinsb | bcwaldon: OK! It's not me that's death marching :-) | 20:32 |
| bcwaldon | justinsb: the difficult part was establishing the conventions for versioning | 20:32 |
| markwash | justinsb:is any part of the death march related to XML issues? | 20:32 |
| jk0 | termie: you go night night? | 20:33 |
| justinsb | markwash: Not really... the XML schemas let us check more, but the real issue is that the data just isn't to spec | 20:33 |
| termie | jk0: lunch, just got back | 20:33 |
| jk0 | :) | 20:33 |
| *** hggdh has quit IRC | 20:33 | |
| justinsb | markwash: I did think about dumping XML. | 20:33 |
| justinsb | markwash: If we dump XML though, it's much harder to check the JSON | 20:33 |
| *** littleidea has quit IRC | 20:33 | |
| *** joearnol_ has joined #openstack | 20:34 | |
| *** RobertLaptop has quit IRC | 20:34 | |
| markwash | justinsb: :-) | 20:34 |
| *** joearnold has quit IRC | 20:34 | |
| bcwaldon | markwash: ;-) | 20:34 |
| justinsb | markwash, bcwaldon: I just see a lot of work for v1.1 and comparatively little payback. But I'm hoping to understand the payback better! | 20:34 |
| *** RobertLaptop has joined #openstack | 20:34 | |
| justinsb | markwash: You want to dump the XML? | 20:34 |
| markwash | justinsb: I don't see how the xml is going to work in Cactus | 20:35 |
| *** hggdh has joined #openstack | 20:35 | |
| justinsb | markwash: I think the XML will work if the JSON is to spec | 20:35 |
| *** littleidea has joined #openstack | 20:35 | |
| *** littleidea has quit IRC | 20:36 | |
| justinsb | markwash: But I agree that I don't see how we're going to get the JSON to spec in time :-) | 20:36 |
| markwash | justinsb: interesting. I guess the limits bug is an example of this problem? | 20:36 |
| jk0 | termie: thanks, you can see now why I did it the first way :) | 20:36 |
| justinsb | markwash: Yes, I think it's a fairly good example | 20:36 |
| markwash | justinsb: cool I'll check it out | 20:36 |
| justinsb | markwash: It came up because the XML is much stricter about all this | 20:36 |
| justinsb | markwash: And the format changed between v1.0 and v1.1 | 20:37 |
| justinsb | markwash: Though I can't see why it changed for limits | 20:37 |
| justinsb | markwash: Which then led me to ponder what did motivate v1.1? | 20:37 |
| justinsb | markwash: And all that extra work! | 20:37 |
| justinsb | Brian Lamar pointed this out to me on this MP: https://code.launchpad.net/~justin-fathomdb/nova/bug740576/+merge/54447 | 20:38 |
| markwash | justinsb: I'm not really sure I can answer the question "why v1.1" | 20:42 |
| markwash | justinsb: but I think the general idea is extensions and tweaks for extensibility | 20:42 |
| justinsb | markwash: I mean I think changePassword is a good answer | 20:42 |
| justinsb | markwash: But I don't understand the random tweaks | 20:42 |
| markwash | justinsb: I'm not sure I understand them any better, so I'll have to refrain from comment | 20:43 |
| justinsb | markwash: Because I really want XML, and I was still debating whether we should dump it | 20:43 |
| justinsb | markwash: I think something's got to give. | 20:43 |
| justinsb | markwash: And I don't think dumping XML is the answer, because validating JSON is probably harder | 20:43 |
| justinsb | markwash, bcwaldon: I was thinking about proposing a super-limited v1.1 API, that would just have the bare minimums we need over v1.0 to do what made it into Cactus | 20:44 |
| markwash | justinsb: so I'm all about validating stuff up front to make our parsing easier | 20:44 |
| markwash | justinsb: but do we *need* validation | 20:44 |
| justinsb | markwash: Not sure what you mean by "up front". I totally agree that we need validation though | 20:45 |
| justinsb | markwash: We have an awesome test suite in the form of our XML schema | 20:45 |
| justinsb | markwash: We just need to use it ;-) | 20:45 |
| justinsb | markwash, bcwaldon: Any reason not to just do a minimal v1.1 and revert out the "unnecessary" tweaks? | 20:46 |
| justinsb | markwash, bcwaldon: Where "unnecessary" will no doubt be the subject of intense flamewars :-) | 20:46 |
| *** bkkrw has joined #openstack | 20:46 | |
| *** sebastianstadil has joined #openstack | 20:47 | |
| justinsb | markwash, bcwaldon: Or rather, can you think of any reasons not to propose that on the ML? | 20:47 |
| openstackjenkins | Project swift build #231: SUCCESS in 27 sec: http://hudson.openstack.org/job/swift/231/ | 20:48 |
| openstackjenkins | * Tarmac: Remove boto dependency by implementing request canonicalization logic in swift3 middleware. | 20:48 |
| openstackjenkins | * Tarmac: fixed lockups and lost messages in swift-init and capture_stdio | 20:48 |
| bcwaldon | justinsb: what would you deem "minimal" at this stage? | 20:48 |
| markwash | justinsb: I think proposing it along with its motivations would be a good idea | 20:48 |
| justinsb | bcwaldon: Whatever we need to support the actions that v1.0 can't. So, your example of changePassword is a great one... | 20:48 |
| justinsb | bcwaldon: It returns a new element "changePassword" I presume, which isn't in v1.0 | 20:49 |
| justinsb | bcwaldon: Sorry, didn't mean to just restrict to actions | 20:49 |
| bcwaldon | justinsb: it doesn't return anything different, it supports POSTing a new entity | 20:49 |
| justinsb | bcwaldon: Whatever we need to support the functionality that v1.0 can't | 20:49 |
| bcwaldon | justinsb: I understand, and I do agree with the idea | 20:50 |
| bcwaldon | justinsb: but I don't think we can take this approach at this point | 20:50 |
| justinsb | bcwaldon: I dpm | 20:50 |
| justinsb | bcwaldon: Whoops | 20:50 |
| openstackjenkins | Project nova build #753: SUCCESS in 2 min 36 sec: http://hudson.openstack.org/job/nova/753/ | 20:50 |
| openstackjenkins | Tarmac: Handles situation where Connection._instances doesn't exist (ie. production) | 20:50 |
| justinsb | bcwaldon: I don't think it's a great option, and ideally you guys can get everything working in both v1.0 and v1.1 formats | 20:51 |
| bcwaldon | justinsb: it is our main goal to get v1.0 and v1.1 as well supported as possible for cactus | 20:52 |
| bcwaldon | justinsb: more of a focus on 1.1 | 20:52 |
| justinsb | bcwaldon: BTW, just checked, you're right, I misread the spec. It's a new element that must be POSTed, not returned. | 20:52 |
| *** Ryan_Lane|food is now known as Ryan_Lane | 20:52 | |
| *** johnpur has quit IRC | 20:53 | |
| justinsb | bcwaldon: That's great - hopefully there's not as much work as I thought. | 20:53 |
| bcwaldon | justinsb: most of our team's branches recently have been version-related issues | 20:53 |
| justinsb | bcwaldon: Another problem is that without a big v1.1 carrot, are clients going to bother porting to the v1.1 API | 20:53 |
| justinsb | bcwaldon: I know - you're all doing great stuff | 20:53 |
| justinsb | bcwaldon: If you think you can make the deadline, then I can drop it. | 20:54 |
| bcwaldon | justinsb: we recognize there is still a lot of testing to get done to find all of these version-specific inconsistencies. And I agree, clients may not see the value in supporting v1.1 in cactus, and wait till diablo | 20:54 |
| bcwaldon | justinsb: I can guarantee we won't meet the full v1.0 AND v1.1 specs by the cactus deadline | 20:55 |
| justinsb | bcwaldon: Would it be OK to propose it a fallback to the ML then for general discussion? | 20:55 |
| bcwaldon | justinsb: what exactly would you want to propose? | 20:56 |
| justinsb | bcwaldon: That we consider making v1.1 experimental, possibly renaming it to v2... | 20:56 |
| justinsb | bcwaldon: And we redefine v1.1 to be a tiny set of extensions to v1.0 | 20:56 |
| justinsb | bcwaldon: Adding _only_ what is actually needed | 20:56 |
| justinsb | bcwaldon: And no 'nice to haves' | 20:57 |
| justinsb | bcwaldon: e.g. I see the list of atom URLs as a nice to have | 20:57 |
| bcwaldon | justinsb: I do agree that it is a major version change. But you have to understand that the spec wasn't developed by the community. I do agree most of the changes in v1.1 shouldnt be there | 20:57 |
| *** adiantum has joined #openstack | 20:57 | |
| justinsb | bcwaldon: Do you mean "was developed by the community"? | 20:57 |
| markwash | labeling it experimental for Cactus could be good | 20:57 |
| openstackjenkins | Project nova build #754: SUCCESS in 2 min 30 sec: http://hudson.openstack.org/job/nova/754/ | 20:57 |
| openstackjenkins | Tarmac: Avoid any hard dependencies in nova.virt.vmwareapi.vim. | 20:57 |
| bcwaldon | justinsb: the spec came from a small group in rackspace, i.e. not the OS community | 20:58 |
| justinsb | bcwaldon: Oh great, I've just unearthed another land mine :-) | 20:59 |
| termie | justinsb: ding ding ding | 20:59 |
| bcwaldon | justinsb: didn't you wonder where that magical spec came from? And why it matches the Cloud Servers spec so closely? | 20:59 |
| justinsb | bcwaldon: I thought everyone else was on board | 21:00 |
| bcwaldon | justinsb: yeah, it was accepted as a blueprint, so it was | 21:00 |
| termie | what is the acceptance process? | 21:00 |
| justinsb | bcwaldon: Oh no, this involves blueprints as well ? :-( | 21:00 |
| bcwaldon | justinsb: yeah, stand by for link | 21:01 |
| *** Glace has quit IRC | 21:01 | |
| termie | justinsb: anyway, i agree with bcwaldon here, i'm not a fan of the api nor how it came into being | 21:01 |
| bcwaldon | https://blueprints.launchpad.net/nova/+spec/openstack-api-1-1 | 21:02 |
| justinsb | I'm going to draft an email to the ML. I want to make sure everyone's aware of the XML issues anyway. If anyone would like to see the email before I send it to avoid stepping on anyones toes (bcwaldon?), just let me know. | 21:02 |
| *** ctennis has quit IRC | 21:02 | |
| bcwaldon | so just to make it clear, the spec was proposed and accepted through the proper channels | 21:02 |
| *** Glace has joined #openstack | 21:02 | |
| termie | bcwaldon: it was automatically accepted | 21:03 |
| termie | bcwaldon: at the very least i certainly had no input on it nor anybody on my team | 21:03 |
| justinsb | I think this could be a scheduling/release issue. i.e. we could defer the blueprint | 21:03 |
| *** Ep5iloN_ has quit IRC | 21:03 | |
| justinsb | I want at least one API that works 100% | 21:03 |
| blamar | termie: It was discussed on the ML for a while? I think the biggest hurt was this coming in between summits... | 21:04 |
| justinsb | And I'd like that to be the OS API | 21:04 |
| termie | blamar: it was discussed that we needed an api and various arguments were brought up, the blueprint itself was just marked accepted by somebody randomly | 21:04 |
| bcwaldon | termie: it was approved by Rick Clark | 21:05 |
| termie | blamar: long before the giant pdf that is the spec was even in existence | 21:05 |
| justinsb | Can we leave aside the blueprint acceptance for the moment... | 21:05 |
| *** troytoma` is now known as troytoman-away | 21:05 | |
| justinsb | ...if the Titan team can do magic (as they have been doing) and get working APIs that's great | 21:05 |
| *** pharkmillups has quit IRC | 21:05 | |
| justinsb | But there's a lot of work to be done, and I'm going to propose an option that could free up resources | 21:05 |
| *** Ep5iloN_ has joined #openstack | 21:05 | |
| justinsb | If the Titan team says they've got it under control, they can smack me down on the ML :-) | 21:06 |
| bcwaldon | I dont think we can really accomplish much here in IRC at the moment. Send out an email to the ML describing your concerns and we will see what everybody thinks | 21:06 |
| blamar | bcwaldon: Agreed | 21:06 |
| justinsb | Will do... wanted to run it by the IRC crowd first to get a better understanding before the ML email. I think I've got that, so I will email the ML. Thanks! | 21:07 |
| bcwaldon | justinsb: Good idea. Glad we could help | 21:08 |
| openstackjenkins | Project nova build #755: SUCCESS in 2 min 30 sec: http://hudson.openstack.org/job/nova/755/ | 21:09 |
| openstackjenkins | Tarmac: Add obviously-missing method that prevents an Hyper-V compute node from even | 21:09 |
| openstackjenkins | starting up | 21:09 |
| *** adiantum has quit IRC | 21:13 | |
| *** ctennis has joined #openstack | 21:14 | |
| *** ctennis has joined #openstack | 21:14 | |
| openstackjenkins | Project swift build #232: SUCCESS in 27 sec: http://hudson.openstack.org/job/swift/232/ | 21:16 |
| openstackjenkins | Tarmac: Added two command line options to bin/swift-log-stats-collector. | 21:16 |
| *** YellowMenace has joined #openstack | 21:24 | |
| *** h0cin has quit IRC | 21:25 | |
| *** YellowMenace has quit IRC | 21:25 | |
| *** allsystemsarego has quit IRC | 21:26 | |
| *** jaypipes is now known as jaypipes-afk | 21:31 | |
| *** joearnol_ has quit IRC | 21:32 | |
| *** hadrian has quit IRC | 21:33 | |
| *** joearnold has joined #openstack | 21:33 | |
| *** littleidea has joined #openstack | 21:36 | |
| *** joearnol_ has joined #openstack | 21:43 | |
| *** joearnold has quit IRC | 21:46 | |
| *** spectorclan_ has joined #openstack | 21:47 | |
| *** _vinay has joined #openstack | 21:50 | |
| _vinay | Hi, | 21:50 |
| *** bcwaldon has quit IRC | 21:50 | |
| *** jfluhmann has quit IRC | 21:51 | |
| _vinay | the nova.sh script for installation, uses kvm as virtualization software right? | 21:51 |
| *** miclorb has joined #openstack | 21:51 | |
| uvirtbot | New bug: #746068 in swift "stats system needs to better track differences between clients and load balancers" [Low,Confirmed] https://launchpad.net/bugs/746068 | 21:51 |
| spectorclan_ | OpenStack Design Summit - Registration at 287; PLEASE REGISTER TODAY if you plan to come; only have 350 slots (http://openstack.org/register) | 21:51 |
| *** jfluhmann has joined #openstack | 21:55 | |
| *** dragondm has quit IRC | 21:56 | |
| vishy | _vinay: kvm or qemu yes | 21:57 |
| *** mray has quit IRC | 22:04 | |
| *** spectorclan_ has quit IRC | 22:05 | |
| *** ppetraki has quit IRC | 22:06 | |
| *** lionel has quit IRC | 22:06 | |
| btorch | vishy: u got a sec ? I'm having some problems trying to get a uec image to boot using the nova client ... I have uploaded the kernel and img with the nova-mange but no ramdisk. | 22:08 |
| *** gondoi has quit IRC | 22:10 | |
| *** reldan has quit IRC | 22:14 | |
| *** reldan has joined #openstack | 22:17 | |
| vishy | hmm | 22:18 |
| vishy | does that work? | 22:18 |
| vishy | :) | 22:18 |
| vishy | I just put out a patch to make the kernel and ramdisk thing easier | 22:19 |
| *** lionel has joined #openstack | 22:19 | |
| vishy | I would just put a blank ramdisk in if it isn't needed | 22:19 |
| btorch | vishy: I still using an older nova version on this node that I have glance .. since I'm still troubleshooting the other node that I have with the latest trunk . I can't reach the VMs on the latest trunk | 22:20 |
| btorch | vishy: I have registered the image and kernel and then used glance to set the ramdisk=None since it wasn't set | 22:20 |
| *** dmshelton has quit IRC | 22:21 | |
| *** aixenv2 has joined #openstack | 22:21 | |
| btorch | vishy: but whenever I try to nova boot the image is complains that there is no ramdisk and quits | 22:21 |
| btorch | anyway got take the dog out .. brb | 22:21 |
| *** dubsquared has quit IRC | 22:21 | |
| *** gregp76 has quit IRC | 22:22 | |
| *** aixenv has quit IRC | 22:22 | |
| *** reldan has quit IRC | 22:23 | |
| vishy | btorch: yes i don't think nova supports that | 22:27 |
| vishy | you need a null ramdisk | 22:27 |
| *** santhoshtw has quit IRC | 22:32 | |
| *** Zangetsue has quit IRC | 22:32 | |
| *** anotherjesse has joined #openstack | 22:38 | |
| *** kapil has joined #openstack | 22:39 | |
| *** kapil has joined #openstack | 22:41 | |
| uvirtbot | New bug: #746088 in nova "nova with glance image problem" [Undecided,New] https://launchpad.net/bugs/746088 | 22:41 |
| anotherjesse | Any experts at ubuntu/debian preseed automation? | 22:42 |
| anotherjesse | I'm having aperiodic issues with needing to hit return on "Partition Disks" - "Write the changes to disks?" - screenshot - http://twitpic.com/4f3liy | 22:43 |
| _vinay | vishy, so which one? kvm or qemu?? | 22:44 |
| _vinay | I see this line in nova.sh | 22:44 |
| _vinay | LIBVIRT_TYPE=${LIBVIRT_TYPE:-qemu} | 22:44 |
| _vinay | I installed using nova.sh install | 22:44 |
| _vinay | is there a way from cmd line to find out what was installed? | 22:44 |
| vishy | it defaults using to qemu | 22:45 |
| vishy | it is a run param though | 22:45 |
| vishy | qemu == software kvm | 22:45 |
| vishy | so you can change it to kvm when you run | 22:45 |
| vishy | if your hardware supports it | 22:46 |
| _vinay | right, ok so run using ./nova.sh run | 22:46 |
| _vinay | so I assume default its using qemu | 22:46 |
| _vinay | cool.. makes sense .. thx.. I will figure out how to change to kvm from cmd line :) | 22:47 |
| _vinay | btw: do you know if xen is supported as well ? vishy | 22:49 |
| *** jstinson has quit IRC | 22:50 | |
| vishy | _vinay: xen is supported through libvirt (largely untested) | 22:51 |
| *** reldan has joined #openstack | 22:51 | |
| vishy | and XenServer is also supported natively | 22:51 |
| vishy | install instructions for XenServer here: http://wiki.openstack.org/XenServerDevelopment | 22:53 |
| *** joearnol_ has quit IRC | 22:54 | |
| *** gregp76 has joined #openstack | 23:00 | |
| _vinay | ok great.. thanks for pointing out the link | 23:00 |
| vishy | _vinay: swich to kvm by changing the line above from :-qemu to :-kvm | 23:04 |
| vishy | or | 23:04 |
| vishy | sudo LIBVIRT_TYPE=kvm ./nova.sh run | 23:05 |
| *** bkkrw has quit IRC | 23:05 | |
| *** MotoMilind has quit IRC | 23:23 | |
| *** adjohn has joined #openstack | 23:26 | |
| *** MotoMilind has joined #openstack | 23:27 | |
| *** MotoMilind has quit IRC | 23:32 | |
| *** gregp76 has quit IRC | 23:36 | |
| *** vernhart has joined #openstack | 23:36 | |
| _vinay | great thanks vishy | 23:37 |
| *** joearnold has joined #openstack | 23:41 | |
| *** littleidea has quit IRC | 23:44 | |
| *** jeffjapan has joined #openstack | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!