Friday, 2010-09-03

*** tobym has quit IRC00:03
*** zooko has joined #openstack00:04
*** abecc has joined #openstack00:05
*** littleidea has quit IRC00:06
*** adjohn has joined #openstack00:15
*** gazumps has quit IRC00:18
*** abecc has quit IRC00:26
*** abecc has joined #openstack00:26
*** jc_smith has quit IRC00:27
*** littleidea has joined #openstack00:28
*** aliguori has quit IRC00:30
*** littleidea has quit IRC00:30
*** littleidea has joined #openstack00:41
*** gazumps has joined #openstack00:45
*** joearnold has quit IRC00:47
*** RobertLJ has joined #openstack00:47
*** maplebed has quit IRC00:56
*** rlucio has quit IRC01:01
*** zooko has quit IRC01:10
*** smithpg1002 has quit IRC01:26
*** tr3buchet has quit IRC01:29
*** Cybodog has quit IRC01:32
*** alekibango has joined #openstack01:37
*** justinsheehy has quit IRC01:41
*** gazumps has quit IRC01:46
*** rossij has quit IRC01:47
*** tobym has joined #openstack01:47
*** rossij has joined #openstack01:48
*** justinsheehy has joined #openstack02:05
*** littleidea has quit IRC02:08
*** gazumps has joined #openstack02:09
*** silassewell has joined #openstack02:12
adjohnHeya, what's the current status of the openstack api?  Is it functional enough for the web control panel to work?02:18
*** Cybodog has joined #openstack02:24
*** rajijoom has joined #openstack02:27
*** RobertLJ has quit IRC02:30
*** rajijoom has quit IRC02:32
*** ArdRigh has quit IRC03:06
*** gazumps has quit IRC03:17
*** miclorb has quit IRC03:29
*** blamar has quit IRC03:29
*** ChrisAM has quit IRC03:29
*** blamar has joined #openstack03:37
*** miclorb has joined #openstack03:39
*** ChrisAM has joined #openstack03:39
pjzanyone awake?03:39
*** rsampaio has joined #openstack03:41
*** joearnold has joined #openstack03:44
*** veszig` has quit IRC03:51
*** tobym has quit IRC03:55
*** gazumps has joined #openstack03:57
*** veszig has joined #openstack04:01
*** adjohn has quit IRC04:08
*** tr3buchet has joined #openstack04:13
*** adjohn has joined #openstack04:25
*** joearnold has quit IRC04:25
*** zaitcev has quit IRC04:32
*** littleidea has joined #openstack04:40
*** rsampaio has quit IRC04:42
*** rajijoom has joined #openstack04:46
*** f4m8_ is now known as f4m804:50
*** abecc has quit IRC04:57
*** gazumps has quit IRC05:17
*** tr3buchet has quit IRC05:30
*** kashyapc has joined #openstack05:33
*** sirp1 has quit IRC05:37
*** justinsheehy has quit IRC05:46
*** justinsheehy has joined #openstack05:51
*** ibarrera has joined #openstack05:58
*** guigui1 has joined #openstack06:18
*** zheng_li has joined #openstack06:38
*** f4m8 has quit IRC06:40
littleideagholt: Is this feature done?
*** f4m8 has joined #openstack06:41
*** ArdRigh has joined #openstack06:43
*** brd_from_italy has joined #openstack07:00
*** calavera has joined #openstack07:03
*** allsystemsarego has joined #openstack07:14
*** brd_from_italy has quit IRC07:15
gholtlittleidea: Nope. I've coded a couple of prototypes already but we're still deciding on the final direction. Probably be finished up next week.07:29
*** jc_smith has joined #openstack07:29
littleideagholt: is it hard from the blueprints/bug in launchpad to know what is going to be in Austin or is it just me?07:29
gholtWell, it's supposed to be anything that we have to have in Austin is prefixed with austin-. Everything else is hopeful, but secondary. ;)07:31
gholtSo, even though I might finish up the code for publiccontainers in time for Austin, it might not pass full scale tests and get pulled to a future branch.07:32
*** brd_from_italy has joined #openstack07:32
*** alekibango has quit IRC07:35
littleideagholt: ok, so some issues are prefixed, austin, then some are tagged with austin in the 'series', and that's the blue prints, then there are the bugs list07:35
littleideagholt: we're trying to figure out what we need/want/can work on for austin07:38
littleideagholt: public containers is one we have some interest in07:39
*** rnewson has joined #openstack07:39
gholtYeah, I have no idea why that is. I got from the meeting that I was supposed to prefix with austin, so I've been following that. The Swift blueprints seem fairly consistent, but there aren't that many. You're right that the Nova stuff is a bit all over the place right now.07:40
littleideawe can bring it up in the meeting tomorrow.07:40
gholt:) And the openstack-common has a 1.0 series.07:41
gholtI think the only thing that would make publiccontainers not make austin is the scalability tests. And I think that's pretty unlikely.07:43
gholtWhat do you plan to use for auth?07:44
*** miclorb has quit IRC07:45
littleideagholt: for public containers? or generally?07:47
gholtGenerally, the devauth isn't really a production thing.07:47
littleideayeah, we are going to integrate with an existing auth infrastructure07:49
*** zheng_li has quit IRC07:49
gholtlittleidea: Ah, okay. Can I ask what? Have you started in on that code yet? The changes going in with public containers and ACLs is going to affect that integration a bit.07:52
littleideaahh, ok, I haven't been working on that part. I'm mostly working on installation and orchestration with chef, and I've also been task with trying to make sure we know what is happening in the project so we don't get caught by surprise if possible.07:55
littleideaI can find out what we are doing in the morning and get back to you.07:56
littleideathe person doing it is didn't think it would be too bad07:56
littleideabut any guidance you can give us on the chance would be appreciated07:56
littleideagholt: I think the SAIO doc is really confusing now.07:59
gholtHmm. Which part? I think the only recent change has been the pastedeploy stuff.08:00
gholtAnd creiht should be done messing with that now. :)08:02
littleideagholt: the current instructions also changes stuff about mounting a drive and the directories are all different, the old version had a demo device mounting a loopback08:05
*** jc_smith has quit IRC08:05
littleideagholt: sdb may not exist for example08:06
gholtAh, that wasn't SAIO. I'm not sure where the loopback thing was. That was an outside party and wasn't submitted into the project. SAIO has always assumed "you’re using VMware Fusion 3 on Mac OS X Snow Leopard," and has you make an extra disk, which ends up sdb.08:07
*** zheng_li has joined #openstack08:13
littleideaI think the loopback thing used to be in docs, but now it is gone. Seems like it would be better if SAIO instructions would work on any install of lucid. no?08:16
gholtYes, but that's a bit of work to upkeep and the team I'm on that works on Swift is pretty busy. Kinda have tradeoffs on that stuff. :)08:22
gholtBasically that doc started as how I set up SAIO on my Mac way back when and we've been keeping it up to date with that one type of install all along. And that itself has been a decent amount of work, hehe.08:22
gholtAh man, almost 3:30am here. I definitely have to go to bed before I have to get up. See ya later.08:23
*** littleidea has quit IRC08:32
*** alekibango has joined #openstack08:48
*** gcc has quit IRC08:57
*** jerryk has quit IRC09:00
chmouelfor account and containers database replication when is it replicated as whole database file and when just only by record, when the other replica does not have any data in it or not ? (I can't find it anywhere in the code)09:03
*** nefo has joined #openstack09:07
*** seats has quit IRC09:10
*** seats has joined #openstack09:11
*** GasbaKid has joined #openstack09:11
*** adjohn_ has joined #openstack09:11
*** adjohn has quit IRC09:12
*** adjohn_ is now known as adjohn09:12
chmouelnevermind i got it from db_replication._repl_to_node()09:21
*** justinsheehy has quit IRC09:21
*** justinsheehy has joined #openstack09:26
*** metoikos has joined #openstack09:35
*** f4m8 has quit IRC10:06
*** f4m8 has joined #openstack10:06
*** kashyapc has quit IRC10:10
*** kashyapc has joined #openstack10:27
*** jonesy_ has quit IRC11:13
*** kashyapc has quit IRC11:20
*** GasbaKid has quit IRC11:29
*** kashyapc has joined #openstack11:32
*** guigui1 has quit IRC11:34
*** guigui3 has joined #openstack11:36
*** justinsheehy has quit IRC11:42
*** gustavomzw has joined #openstack11:47
*** kashyapc has quit IRC11:48
*** ctennis has quit IRC11:52
*** justinsheehy has joined #openstack12:09
*** ctennis has joined #openstack12:09
*** ctennis has joined #openstack12:09
*** tobym has joined #openstack12:13
*** xfaf has joined #openstack12:14
*** zul has quit IRC12:15
*** xfaf is now known as zul12:16
*** zul has joined #openstack12:16
*** justinsheehy has quit IRC12:18
*** sophiap has quit IRC12:25
*** justinsheehy has joined #openstack12:37
*** metoikos has quit IRC12:49
*** aliguori has joined #openstack12:51
*** Podilarius has joined #openstack12:55
*** DubLo7 has joined #openstack13:00
*** ppetraki has joined #openstack13:02
*** RobertLJ has joined #openstack13:07
*** pcrews has joined #openstack13:08
*** ppetraki_ has joined #openstack13:10
*** ppetraki_ has quit IRC13:16
*** tobym has quit IRC13:16
*** coredump777 has joined #openstack13:17
*** mrayzenoss has joined #openstack13:17
*** mrayzenoss is now known as mray13:17
*** mray has joined #openstack13:17
*** dendro-afk is now known as dendrobates13:20
*** f4m8 has quit IRC13:26
*** f4m8 has joined #openstack13:28
*** f4m8 has quit IRC13:30
*** f4m8 has joined #openstack13:31
*** calavera has quit IRC13:34
*** gondoi has joined #openstack13:39
*** _abien has joined #openstack13:43
*** jonesy_ has joined #openstack13:43
*** _abien has quit IRC13:44
*** _abien has joined #openstack13:46
*** matclayton has joined #openstack13:50
*** f4m8 is now known as f4m8_13:52
*** abien has quit IRC13:52
*** adjohn has quit IRC13:54
*** tobym has joined #openstack13:55
*** nefo has quit IRC14:01
*** pvo has quit IRC14:07
*** vvuksan has joined #openstack14:11
*** metoikos has joined #openstack14:13
*** vvuksan has quit IRC14:15
*** vvuksan has joined #openstack14:15
*** ibarrera has quit IRC14:18
*** abecc has joined #openstack14:21
*** pvo has joined #openstack14:25
*** ChanServ sets mode: +v pvo14:25
*** matclayton has left #openstack14:26
*** ArdRigh has quit IRC14:38
burrisso I'm running swift-solo in a guest os and using cyberduck on my host os to access it but it wasn't working, then I realized that the auth server was returning for the storage url which doesn't work on the host, ssh port forwarding to the rescue!14:40
vvuksani noticed that too14:41
vvuksanactually what you can do is this14:41
vvuksanin chef/cookbooks/swift/attributes/swift.rb14:42
vvuksanchange  default[:swift][:proxy_server][:hostname] = ""14:42
vvuksanto say14:42
vvuksan  default[:swift][:proxy_server][:hostname] = node[:ipaddress]14:42
*** gustavomzw has quit IRC14:43
burrisah cool thanks14:44
vvuksanburris: you will have to redo the /etc/swift/auth.db14:44
vvuksanperhaps even other pieces14:44
vvuksanit hardcodes the URL location14:44
creihtburris: yeah you have to set the default_cluster_url in the auth config14:47
creihtwhen an account is created, that value is stored in the authdb, so you would either need to update the sqlite auth.db manually, or just delete it and re-create the users14:48
creihtre-create the accounts14:48
vvuksani did update sqlite auth.db manually and it didn't cut it14:49
vvuksancreating new accounts did14:49
*** tr3buchet has joined #openstack14:49
creihtvvuksan: it should work if you update it manually14:51
*** littleidea has joined #openstack14:52
*** sirp1 has joined #openstack14:52
creihtlittleidea: good morning14:53
littleideacreiht: good morning14:53
creihtI'm wondering if you had been looking at a different doc, as the SAIO never had docs for a loopback device14:53
creihtthough I have been meaning to add that14:54
creihtI know the initial chef scripts that holoway created used a loopback device14:54
vvuksancreiht: for some reason it didn't :-/14:58
vvuksani tried14:58
creihtvvuksan: well if you end up needing to do it again, let me know so that we can check it out15:00
*** Phacops has joined #openstack15:04
*** Phacops_ has quit IRC15:05
*** pharkmillups has joined #openstack15:14
dendrobateseveryone should be getting an email saying that we are extending the blueprint deadline to Tuesday15:16
*** tr3buchet has quit IRC15:28
*** zaitcev has joined #openstack15:38
*** tr3buchet has joined #openstack15:57
*** JimCurry has joined #openstack16:00
*** whitenhiemer has joined #openstack16:03
*** Cybodog has quit IRC16:06
*** maplebed has joined #openstack16:07
*** brd_from_italy has quit IRC16:08
*** burris has quit IRC16:25
*** rnewson has quit IRC16:25
*** zheng_li has quit IRC16:31
*** guigui3 has quit IRC16:33
*** joearnold has joined #openstack16:34
jaypipesgholt: heya, quick question for you....16:39
jaypipesgholt: in Swift, what is the difference between an account and a user?16:39
jaypipesgholt: is account like a project in Nova?  i.e. a user can belong to multiple projects?16:41
*** localhost2 has joined #openstack16:44
*** localhost2 has quit IRC16:46
*** localhost has quit IRC16:46
*** localhost2 has joined #openstack16:47
*** pcrews is now known as pcrews_bbl16:49
*** DubLo7 has quit IRC16:49
*** gustavomzw has joined #openstack16:53
*** brd_from_italy has joined #openstack16:54
*** tobym has quit IRC16:56
*** zheng_li has joined #openstack16:58
*** DubLo7 has joined #openstack17:02
*** burris has joined #openstack17:02
*** RobertLJ has quit IRC17:06
*** ali``` has joined #openstack17:08
*** DubLo7 has quit IRC17:11
*** joearnold has quit IRC17:19
*** joearnold has joined #openstack17:21
*** rnewson has joined #openstack17:29
*** pharkmillups has quit IRC17:30
*** ali``` has quit IRC17:35
*** tobym has joined #openstack17:45
*** rajijoom has quit IRC17:50
*** annegentle has joined #openstack17:55
*** jbartels has joined #openstack17:57
*** jbartels has quit IRC17:58
*** pcrews_bbl is now known as pcrews18:00
*** kashyapc has joined #openstack18:02
dendrobatesAnyone with questions or wanting with blueprints should head over to #openstack-meeting in 48 min.18:12
dendrobatesthat should be wanting help with blueprints18:12
*** kashyapc has quit IRC18:13
sorendendrobates: Wait, what? You said 2000 UTC in your e-mail?18:15
*** kashyapc has joined #openstack18:15
*** DubLo7 has joined #openstack18:17
*** brd_from_italy has quit IRC18:17
*** rnewson has quit IRC18:18
*** rnewson has joined #openstack18:18
*** rnewson has joined #openstack18:18
*** kashyapc has quit IRC18:18
zul200 utc on the wiki as well18:19
zul2000 i mean18:19
dendrobatesok, I must have made a tz error18:22
dendrobatesha I did sorry add an hour to that18:23
gholtjaypipes: I don't know Nova (yet) :( But in Swift there's the storage account /v1/<storageaccount>/blah and all else is up to the auth.18:23
*** mray has quit IRC18:23
dendrobatessoren: you know I 'm timezone deficient :)18:23
jaypipesgholt: so what is the difference b/w a user and an account?18:23
gholtjaypipes: The devauth has a concept of an account == storageaccount and that authaccount can have multiple users.18:24
jaypipesgholt: devauth?18:24
gholtThe "sample" auth system within the Swift repo.18:24
gholtFor Rackspace CloudFiles, their auth only has the concept of authaccount == storageaccount.18:24
gholtSo the answer is: It depends. :)18:25
jaypipesgholt: ah, ok. so I guess I'm still blanking on the difference between a storage account and an auth account...not sure why >1 account is needed?  Is this a RS-specific thing?18:25
jaypipesgholt: sorry if I sound dumb about this...just don't understand :(18:25
*** dendrobates is now known as dendro-afk18:26
gholtNo, not dumb at all. The separation is there for flexibility in bolting on any auth system.18:26
jaypipesgholt: are you saying that there is an LDAP user that is called the "auth account" and a separate identifier for the Swift/Cloud Files "account"?18:26
gholtAlso, you could have one auth and fifty different Swift clusters (in different regions or w/e)18:26
jbrycejaypipes: auth accounts are separate from the "account" in swift which is the top level aggregator (account -> container)18:27
*** pharkmillups has joined #openstack18:27
gholtjaypipes: yep18:27
*** amscanne has quit IRC18:27
jaypipesgholt, jbryce: aha. got it now :)18:27
gholtWe love many levels of indirection. :)18:27
jaypipesgholt: and you're saying that in the functional test system, the auth account is identical to the Swift account, correct?18:27
gholtWell, they're 1:1 yes.18:28
jaypipesgholt: right, sorry18:28
*** dendro-afk is now known as dendrobates18:28
*** Tybstar has joined #openstack18:28
*** mrayzenoss has joined #openstack18:29
sorenok, so your email eas right? it's an hour and a half from now?18:29
jaypipesgholt:  in Nova there is a user (by default, an LDAP user DN, but that is becoming adaptable) and there are "projects" that a user can belong to.  The permissions a user has depends on the roles that the user has in a specific project.18:29
jaypipesgholt: since I've been working on blueprints and prototyping for the openstack-common project, I've been reading the code for swift and nova, trying to identify some common areas that could be consolidated (thus, my comments on your branch proposals) :)18:30
*** kr8tr has joined #openstack18:31
jaypipesgholt: the idea is to take the best code from each project and push it up to an openstack-common Python lib for all OS projects to import.18:31
gholtjaypipes: Gotcha. I'm not certain how that would map to storage-accounts being unfamiliar with the territory, but I'm certain it could map. :) If that sorts out we should probably change the Swift devauth to use the same sort of structure.18:32
gholtKeeping the terminology the same would be a big bonus. So, Swift's devauth would have users and projects etc.18:33
sorendendrobates: ok, so your email was right? it's an hour and a half from now?18:36
*** kapten has joined #openstack18:39
*** kapten has left #openstack18:40
*** amscanne has joined #openstack18:41
*** mrayzenoss has quit IRC18:42
*** dendrobates is now known as dendro-afk18:49
jaypipesgholt: possibly.  or we change project -> account in Nova ;)18:50
jaypipessoren: yep18:51
*** mrayzenoss has joined #openstack18:52
*** dendro-afk is now known as dendrobates18:52
*** mrayzenoss is now known as mray218:53
*** mray2 is now known as mrayzenoss18:53
*** lmcdowell has joined #openstack18:53
jbrycejaypipes, gholt: in swift, can a single user account belong to multiple swift accounts?18:56
creihtjbryce: no18:57
creihton the current auth18:57
*** greenisus has joined #openstack18:57
creihtbut there isn't anything that prevents someone from creating an auth that does that18:57
jaypipescreiht: what do you mean "the current auth"?  You mean "not gholt's new stuff"? :)18:57
creihtI mean both :)18:57
jaypipescreiht: k18:57
jbryceso that's a difference from the nova project concept18:57
gholtHehe, the new stuff will allow users from other accounts access to your account if you want.18:58
creihtThe dev auth is patterened after RS auth18:58
*** ctennis has quit IRC18:58
jaypipesjbryce: yeah, but it sounds like it could be pretty easy to adapt.18:58
creihtyeah you can give other accounts/users access in gholt's branch, but they aren't "members"18:58
creihtthere may not be much of a distinction though18:58
*** klord has joined #openstack19:04
*** joearnold has quit IRC19:08
*** joearnold has joined #openstack19:10
* jaypipes notes that paste.deploy's docs are obtuse...I hate when coders write documentation as if users already understand how their software works.19:10
creihtyeah it took me a little bit to grok19:10
*** joearnold has quit IRC19:10
creihtdo you have a specific question?19:10
*** Tybstar has left #openstack19:12
jaypipescreiht: just trying to wrap my head around Ian's terminology.  Plus, the examples don't really explain what the heck paste.deploy actually does...19:13
jaypipesshit like "Typically you have one or two sections, named “main”: an application section ([app:main]) and a server section ([server:main])."19:13
jaypipesand then there is no server:main section in the example directlyfollowing...19:13
jaypipes :(19:13
jaypipescreiht: I'm totally clueless as to where the heck you would put the loadapp() call... :( He provides no clear indication of where that would go...a deployment script of some sort?19:15
creihtso for swift, we call loadapp() in run_wsgi()19:16
creihtinstead of manually instantiating the wsgi app, loadapp reads the config, and instantiates the wsgi app based on the config19:17
* jaypipes ack-greps for it...19:17
*** jasonb365 has joined #openstack19:18
jaypipescreiht: can you pass CLI args to these servers (like bin/swift-account-server)?19:19
jaypipescreiht: or does everything need to be in a configuration file?19:19
*** greenisus has quit IRC19:19
creihtjaypipes: we do everything with config files19:19
*** greenisus has joined #openstack19:19
jaypipescreiht: so in order to test a different option for a program, I need to edit a configuration file?  I can't just pass an option on the CLI?19:20
creihtor have a couple of configs19:20
gholtNormally you'd change the conf and init reload the server.19:20
jaypipesgholt, creiht: ok.19:21
jaypipescreiht: so, you guys do *not* use paster server, correct?  you do everything in swift.common.wsgi.run_wsgi() if I'm not mistaken?19:24
jaypipess/paster server/paster serve/19:24
jaypipesI guess I'm still confused as to the point of paste deploy (and loadapp()).  Seems to be just a way of loading configuration stuff from a file...does it really *load* some application into, say, a web server?19:25
gholtIt's a way to load confs and to load wsgi app chains, also seems to provide a cool way to "publish" your wsgi middleware.19:27
*** justinsheehy has quit IRC19:27
jaypipesgholt: publish to *what*?19:27
gholtWell, in you can indicate your package has a wsgi filter or app...19:27
jaypipesgholt: and what does that do exactly?19:28
* jaypipes very confused at Python web deployment...seems very complicated for something that should be very simple.19:28
gholtWell, take a peek at the proxy-server.conf on
jaypipesgholt: that page is just scary. :)19:30
gholtImagine I write a cool new wsgi filter. I install it with that pastedeploy thing. To use it with Swfit, I'd just add it to the pipeline as whatever name I want, then make a [filter:name] use = egg:mycoolproj#mywsgifilter and it's installed in Swift's proxy server.19:30
jaypipesI like this: "It assumes you’re using VMware Fusion 3 on Mac OS X Snow Leopard"19:30
gholtWell, yeah, but ignore that for now.... I just wanted a sample proxy-server.conf for you to read. :)19:31
jaypipesgholt: no, I understand...19:31
dendrobates30 min until blueprint q&a in #openstack-meeting19:31
jaypipesgholt: I guess I'm just used to PHP web development where I just put the darn PHP files at some directory and set an Apache conf. ;)19:32
jaypipesgholt: much simpler than all this. :)19:32
gholtHehe, well it is a bit more complicated when you think that Swift alone has like what, 5 different wsgi servers that could be running?19:33
gholt[In the case of SAIO, even more since it's emulating 4 boxes of "stuff"]19:34
jaypipesgholt: ya. Nova has about the same...19:34
creihtjaypipes: The biggest reason for using paste.deploy is that to support easily pluggable middleware, I was either going to have to write my own version of paste.deploy, or just use it19:38
creihtthe advanatage is that you can easily change middleware, or wsgi flow without changing the app code19:39
creihtit also makes it super simple to use any other middleware that is currently available that supports paste.deploy (which is most)19:39
jaypipescreiht: I guess I don't see how paste.deploy does that. seems that run_wsgi does a heck of a lot more than paste.deploy does...looks like loadapp is simply loading some configuration data19:40
*** Cybodog has joined #openstack19:40
creihtso before, we had something like:19:40
jaypipescreiht: you still have to do all the eventlet stuff, logging, etc19:41
creihtapp = MyApp()19:41
creihtapp = MyAuthMiddleware(app)19:41
creihtapp = OtherMiddleware(app)19:41
*** greenisus_ has joined #openstack19:41
creihtwith paste.deploy, it is all config19:41
gholtYeah, I'm with you on that jaypipes, paste.deploy pretty much just does conf and app chain loading. The wsgi environment is separate19:41
creihtall loadapp does is instantiate the wsgi app19:41
jaypipescreiht: understood.19:42
creihtall the other stuff is just server stuff that we would have to do either way19:42
jaypipescreiht: by "instantiate" do you mean just calling the wsgi-fied object/callable and creating environ?19:42
creihtthe environ isn't created until a request is made19:42
* jaypipes trying to get a handle on what "magic" is done or not done so I can better understand how these applications are deployed19:42
creihtit just initializes the objects19:43
jaypipescreiht: ok, so "instantiate" just calls the wsgi-field callable then?19:43
creihtwhihc is determined by the python entry point in your setup.py19:43
jaypipescreiht: or rather, looks up the config, determines the chain, and creates the chain.19:43
jaypipescreiht: gotcha. ok, understanding better now.19:43
creihtso in theory, you could even run swift with paster server if you really wanted to19:44
jaypipescreiht: sorry for all these dumb questions...I've had a steep learning curve as I haven't had much Python web dev experience. Only Python systems programming and PHP web dev from years ago..19:45
jaypipescreiht: you do C++ database development for a time and things get icky to re-learn ;)19:45
*** greenisus has quit IRC19:45
*** greenisus_ is now known as greenisus19:45
creihtIt all boils down to allowing the addition of middleware through config rather than having to edit source to do that19:45
jaypipescreiht: yep, understanding it now.19:46
creihtjaypipes: I already graduated from C++ systems 8 years ago :)19:46
gholtI never graduated; I just dropped out I think.19:46
jaypipescreiht: :)19:47
jaypipescreiht: believe me, I'm not bashing Python at all...just a lot to learn from my side.19:48
creihtno worries19:48
creihtIt took me a while to grok the paste.deploy stuff as well19:48
creihtjaypipes: at least you aren't trying to mangling xml for java web services :)19:50
jaypipescreiht: excellent point ;)19:50
*** xtoddx has joined #openstack19:51
*** greenisus has quit IRC19:52
*** greenisus has joined #openstack19:52
*** greenisus_ has joined #openstack19:55
*** greenisus has quit IRC19:58
*** greenisus_ is now known as greenisus19:58
dendrobatesblueprint discussion in #openstack-meeting in 1 minute!19:58
zulRUN! :)19:59
*** gustavomzw has quit IRC19:59
*** Cybodog has quit IRC20:04
littleidealet the games begin?20:04
* creiht wonders what color the blueprints should be20:05
*** tobym has quit IRC20:08
*** rnewson has quit IRC20:11
vishycreight: I vote for copen, because azure is overused20:12
*** ewanmellor has joined #openstack20:13
sorenewanmellor: So, what the Ubuntu guys have done, is to detect if the kernel given is a multiboot compliant kernel image...20:13
sorenewanmellor: ...if it is, they generate a boot floppy (for a number of reasons (to work around shortcomings in other places))...20:14
sorenewanmellor: ...boot from that, which loads the given kernel image..20:14
vishysoren: is there any reason we can't just boot whole machines by not specifying a kernel and ramdisk?20:14
sorenewanmellor: ...and let it rip.20:14
sorenSo, the cool part is this:20:14
creihtvishy: I like ultramarine, because well it has the work ultra in it :)20:14
sorenYou can use Grub as this multiboot compliant kernel image.20:15
*** rnewson has joined #openstack20:15
soren...and grub can be configured to go look for more grub on the filesystem and boot from that..20:15
soren...and failing that, it can chainload.20:15
soren...and either boot from (hd0) or (hd0,0)-.20:15
*** allsystemsarego has quit IRC20:15
*** kr8tr has left #openstack20:16
sorenewanmellor: Of course wë́'ll provide these grub kernel images.20:16
*** coredump777 is now known as coredump-br20:17
ewanmellorThere's no need for any of that on XenServer (cool as it sounds).  You can just specify a boot order (CD, network, disk in any order) and it will boot a QEMU BIOS and away you go.20:17
*** coredump-br is now known as coredump|br20:17
sorenewanmellor: ..which means people can use that aki and shove a boot loader in the image they upload, and voila.20:17
sorenewanmellor: Same for kvm.20:17
sorenewanmellor: The problem is that that requires changes to the way disks are stitched together.20:17
vishysoren: yes it seems cool, but i don't understand the need to make it that complicated.20:18
vishysoren: you mean the partition code?20:18
sorenewanmellor: As it it, we upload the first partition as the image. This gets stitched together with a partition table and whatever remaining space there may be for ephemeral storage.20:18
sorenewanmellor: no mbr or boot loader.20:18
sorenvishy: Yes.20:18
ewanmellorsoren: But for it to be a supported Windows installation, it needs to use the Windows bootloader.20:19
sorenewanmellor: It will.20:19
vishysoren: but we can have two types of images, it can check the image, if it is a whole disk image, skip partitioning and just boot as a whole disk20:19
sorenvishy: Yes. There a several reasons why it was done this way. One was that it was very close to what Amazon had done.20:20
vishyin fact, the partition code in general is just due to legacy concerns of having a disk with an mbr,20:20
sorenvishy: They support pvgrub booting now.20:20
ewanmellorSo what does the boot-whole-images branch do, if not what vishy is describing?20:20
vishythat is what it does20:22
vishyi think justin used a special kernel id for whole disk aki-000000 or some such20:22
*** tobym has joined #openstack20:24
vishysoren: i noticed you put in something for ec2-security-groups.  We have that on our roadmap as well, but I don't think we were planning to use libvirt for it20:24
sorenvishy: Any particular reason?20:24
vishysoren: have you started work on it?20:24
sorenvishy: No.20:24
sorenvishy: I've started working on nwfilter support for uml, though, which is a prerequisite (for testing).20:24
vishysoren: no reason, didn't know that libvirt supported it, can you change the fw rules on the fly?20:24
*** joearnold has joined #openstack20:26
sorenvishy: Yes.20:26
vishysoren: the linux_net driver for network already is creating iptables rules, so it seemed like an obvious place to add them20:26
sorenvishy: libvirt does iptables, ip6tables and ebtabales.20:27
vishyhow does libvirt handle creating rules for instances on different machines?20:27
sorenvishy: google for: libvirt nwfilter20:27
*** lew has joined #openstack20:27
sorenThere's a good primer on the libvirt website. It seems very cool, and I really like the idea of keeping it on the host.20:27
vishydo you specify for each instance individually?20:27
sorenvishy: You can.20:27
sorenvishy: ...but I don't belive you have to.20:28
sorenvishy: I could be wrong, though.20:28
vishysoren: ok, I'll pass it on to the guy on our team who it is assigned to20:28
sorenvishy: I mean... You specify them globally, and apply them to virtual interfaces.20:28
vishysoren: internally we'll probably do whatever is fastest20:28
vishysoren: but that may not be the best solution for openstack as a whole20:29
sorenvishy: Perhaps not. I think it will, but perhaps not :)20:30
vishysoren: I'm a fan of using libvirt wherever possible, but it may be faster for us to get something working just doing iptables rules on the network node20:31
vishysoren: we're in a bit of a time crunch here :)20:31
sorenvishy: I understand.20:32
sorenvishy: If you don't work on the libvirt approach, I will.20:32
sorenvishy: Just so you know.20:32
vishysoren: cool20:33
*** jc_smith has joined #openstack20:34
*** whitenhiemer has quit IRC20:35
*** joearnold has quit IRC20:37
vishysoren: I suggest for the HVM that we start with justin's simple version, and we can add chain booting support later20:42
sorenvishy: I'm cool with that. I'd like to the the dude who made the Eucalyptus changes in on the discussion.20:43
vishysoren: good idea.  If he wants to submit a patch that would rock20:44
*** Cybodog has joined #openstack20:45
*** xtoddx has quit IRC20:49
dendrobatessoren: who are you talking about?20:50
ewanmelloramscanne: You have XenAPI questions?20:50
littleideadendrobates: so will you change all the blueprints that are marked 'austin' series in nova to have the name 'austin-'20:51
amscanneewanmellor: No, I actually have questions about the general acceptance or policies around implementing new "virt" adaptors20:51
*** Cybodog has quit IRC20:52
dendrobateslittleidea: only if the drafter tells me that they still intend for them to land in Austin20:52
ewanmelloramscanne: I don't think we have any specific policies for virt adaptors -- just the ones that apply to all code.20:52
littleideadendrobates: ahh, can we remove the 'austin' from the series for the ones that won't, I thought it was a little confusing.20:53
ewanmelloramscanne: What platform do you want to support?20:53
amscanneewanmellor: We have our own platform (, I like where openstack is going and I've implemented some toy integration code20:54
*** jonesy_ has quit IRC20:54
sorendendrobates: smoser20:55
dendrobatessoren: ah20:55
amscanneewanmellor: Going to formalize it a bit more, but I'm just wondering about the general policies and philosophy is for integrating support for other platforms20:55
dendrobateslittleidea: what blueprints do you mean?20:55
letterjdendrobates: I sent an email about a source for lab hardware.  Did you ever see it?20:55
dendrobatesletterj: no, to me personally?20:56
*** coredump|br has quit IRC20:57
littleideadendrobates: I'm not confused now, but all the blue prints that have 'austin' in the series on this page but aren't named austin
letterjNo I was told to send it to jpurrier20:57
dendrobateslittleidea: yeah you can't go by that, we made too many changes to how were doing things.20:59
dendrobatesthat is why I need everyone to either name them austin- or tell me.20:59
ewanmelloramscanne: Well, I can't speak for the project, but I personally have no problem with adding additional virt adapters for things like this.  As much unit testing as you can supply will reduce the chance of it breaking when other things are refactored.  That will be important, since presumably you will be the only ones testing it.20:59
dendrobatesalso a bunch of those blueprints are low hanging fruit that we would like people to do , but have no plan on tracking fo rthe release.21:00
littleideadendrobates: I recommend the ones that aren't in austin release get that series tag removed, its confusing if you aren't 100% up on what is happening.21:00
ewanmelloramscanne: You'll be responsible for making sure that it doesn't bitrot, of course.21:00
dendrobateslittleidea: that is part of my weekend project21:00
littleideadendrobates: gotcha21:00
dendrobateslittleidea: I've been doing a little clean up work every day.  there is a lot of it to do.21:01
*** Cybodog has joined #openstack21:01
amscanneewanmellor: Totally :)21:02
*** joearnold has joined #openstack21:02
dendrobateslittle the issue we had was that almost every dev was in a group that auto approved blueprints they created, so  many blueprints got auto approved and targeted21:03
*** ppetraki has quit IRC21:03
dendrobateslittleidea: the teams are fixed now, I just need to fix all the blueprints21:03
amscanne(Sorry for the uninformed question) What is the procedure for determining which blueprints are accepted into the upcoming release?21:04
*** joearnold has quit IRC21:04
dendrobatesamscanne: are you asking how they are approved or how you find out which ones are approved?21:05
*** jc_smith has quit IRC21:05
amscannedendrobates: I'm interested in supporting our virtualization platform through a new "virt" adaptor.. I haven't written a blueprint yet (though in traditional developer-style, I've written code) but I'm just curious about what the process looks like21:06
*** joearnold has joined #openstack21:07
dendrobateswere you at the beginning of the blueprints meeting earlier?21:07
*** jc_smith has joined #openstack21:07
*** lew has quit IRC21:07
dendrobatesamscanne: ^^^21:08
amscannedendrobates: the one today, yes21:08
dendrobatesamscanne: I outlined the way it is supposed to work, which is not how we did it today.21:09
amscannedendrobates: sorry, jus reviewing the minutes to see if I missed it21:09
dendrobateser this release21:09
dendrobatesamscanne: basically there are two approvals, one for the blueprint before a design summit21:10
dendrobatesamscanne: and one for the spec after the summit, to approve the general design21:10
dendrobatesamscanne: of course there are always other gates such as feature freeze and performance testing to get past21:11
dendrobatesamscanne: does that answer your question?21:12
dendrobatesamscanne: and what virtualization platform are you representing?21:12
amscannedendrobates: I'm from gridcentric (, I've just been played with interfacing with our management stack because I like where openstack is going21:13
amscannedendrobates: in the minutes, you say that both the blueprint and the spec are due on Tuesday for this upcoming release -- so no summit presentations for the austin release I'm guessing?21:14
dendrobateswe had the summit in July21:14
dendrobatesso this is a very odd release.21:15
*** sparky__ has joined #openstack21:15
dendrobatesit was our first summit and most participants did not know what to expect.21:15
amscanneOf course,.. the first release is guaranteed to be odd no matter what you do ;)21:16
*** jc_smith has quit IRC21:17
dendrobatespeople have very high expectations for us, so we will strive to meet them, but the pressure will be high.21:18
amscannedendrobates: What are the general feelings towards integrating with other virtualization management stacks (basically just a new "virt" module) -- basically completely isolated changes... my guess would be the more the merrier (provided you take care of the code)?21:18
amscannedendrobates: I think that OpenStack is on the right track, even though there's obviously a lot of work to be done...21:19
dendrobatesis you software an abstraction layer like libvirt?21:19
dendrobatesamscanne: yes think apache modules.  Some we will ship some we will not.21:19
*** Cybodog has quit IRC21:20
*** rnewson has quit IRC21:20
dendrobatesas new virt modules go ,I think the more the merrier as long as they are maintained21:20
dendrobatesamscanne: thanks, and welcome.21:21
amscannedendrobates: No, we have our own management stack and hypervisor (a spin of Xen).. we're a start-up commercializing virtualization research that enables efficient on-demand cloning of virtual machines (kind of a new way to think of, and manage virtualized applications)21:22
dendrobatesamscanne: that sounds like it would fit nicely.21:23
amscannedendrobates: great! I'll cook up a blueprint...21:24
*** ewanmellor has quit IRC21:25
*** greenisus has quit IRC21:29
*** greenisus has joined #openstack21:29
* soren calls it a week21:31
*** burris has quit IRC21:34
*** Cybodog has joined #openstack21:35
dendrobateslazy bastard21:35
dendrobatesit's not even midnight in DK, yet21:35
dendrobatessoren: have a good weekend.21:36
vishysoren: agreed, have fun!21:36
*** jasonb365 has quit IRC21:39
*** klord has quit IRC21:40
*** pharkmillups has quit IRC21:40
*** annegentle has left #openstack21:45
*** tobym has quit IRC21:51
*** tobym has joined #openstack21:53
*** Cybodog has quit IRC21:54
*** Podilarius has left #openstack21:56
*** greenisus_ has joined #openstack22:00
*** burris has joined #openstack22:02
*** blpiatt has quit IRC22:03
*** greenisus_ has quit IRC22:03
*** greenisus has quit IRC22:04
*** cloud0 has quit IRC22:05
*** zheng_li has quit IRC22:05
*** cloud0 has joined #openstack22:05
jbrycei'm heading out. don't labor too hard on labor day.22:06
*** gustavomzw has joined #openstack22:06
JimCurryhappy labor / labour day weekend openstack!  i have a rack of ribs at home waiting for my smoker.22:07
*** jbryce has quit IRC22:09
*** JimCurry has left #openstack22:09
*** gondoi has quit IRC22:16
*** sparky__ has quit IRC22:22
*** Cybodog has joined #openstack22:24
*** amscanne has quit IRC22:25
*** mrayzenoss has quit IRC22:35
*** tobym has quit IRC22:41
*** tobym_ has joined #openstack22:43
*** burris has quit IRC22:47
*** joearnold has quit IRC22:59
*** alekibango has quit IRC23:00
*** jonesy_ has joined #openstack23:02
*** jkakar has quit IRC23:02
*** joearnold has joined #openstack23:06
*** gazumps has joined #openstack23:07
*** metoikos has quit IRC23:10
*** Cybodog has quit IRC23:14
*** jkakar has joined #openstack23:15
*** jonesy_ has quit IRC23:19
*** jonesy_ has joined #openstack23:30
*** abecc has quit IRC23:31
*** alekibango has joined #openstack23:40
littleideadendrobates: Is there a time deadline for Tuesday? EOD?23:41
*** maplebed has quit IRC23:54
dendrobateslittleidea: 21:00 UTC23:58
littleideaok, by the meeting23:58

Generated by 2.14.0 by Marius Gedminas - find it at!