Tuesday, 2010-08-10

*** mtaylor has joined #openstack00:06
*** ChanServ sets mode: +v mtaylor00:06
*** mray has quit IRC00:17
*** silassewell has joined #openstack00:23
*** lbieber has joined #openstack00:23
*** silassewell has quit IRC00:31
*** silassewell has joined #openstack00:32
uvirtbotNew bug: #615637 in nova "rpc.call doesn't handle errors sufficiently" [Undecided,New] https://launchpad.net/bugs/61563700:35
*** heckj has joined #openstack00:55
heckjafternoon/evening - process question if someone's available00:56
heckjI submitted a review/merge request from a branch on cleaning - got feedback. Updated it just now. Do I just reply to that email thread that I've done so, or create a new merge request?00:57
*** heckj has quit IRC01:10
*** e1mer has joined #openstack01:25
*** e1mer has quit IRC01:29
mtaylorherki: there's a couple of ways you can go about this01:31
mtaylorherki: you'll find a link "resubmit merge proposal" on the merge request page01:32
mtaylorherki: oops. sorry - wrong person01:32
mtaylortab completion fail01:32
*** zul has joined #openstack01:45
*** e1mer has joined #openstack01:46
*** lbieber has quit IRC01:49
*** silassewell has quit IRC01:49
*** miclorb has quit IRC01:57
*** sophiap has quit IRC02:00
*** e1mer has quit IRC02:00
*** sophiap has joined #openstack02:00
*** sophiap has quit IRC02:05
*** sophiap has joined #openstack02:12
*** zul has quit IRC02:22
*** jamiew has quit IRC02:25
*** RobertLJ has quit IRC03:02
*** miclorb_ has joined #openstack03:03
*** dendrobates is now known as dendro-afk03:08
*** lbieber has joined #openstack03:10
uvirtbotNew bug: #615670 in glance "Develop backend adapter framework for Teller" [Undecided,New] https://launchpad.net/bugs/61567003:15
uvirtbotNew bug: #615671 in glance "Implement filesystem backend for Teller" [Undecided,New] https://launchpad.net/bugs/61567103:20
*** miclorb_ has quit IRC03:21
uvirtbotNew bug: #615672 in glance "Implement Swift backend for Teller" [Undecided,New] https://launchpad.net/bugs/61567203:25
*** e1mer has joined #openstack03:27
*** miclorb has joined #openstack03:27
*** e1mer has quit IRC03:31
*** e1mer has joined #openstack03:31
*** miclorb has quit IRC03:32
*** pvo has joined #openstack03:33
*** ChanServ sets mode: +v pvo03:33
*** miclorb has joined #openstack03:39
uvirtbotNew bug: #615675 in glance "Implement Parallax's image-lookup API call" [Undecided,New] https://launchpad.net/bugs/61567503:45
*** mtaylor has quit IRC03:53
*** mtaylor has joined #openstack03:55
*** ChanServ sets mode: +v mtaylor03:55
*** sirp2 has joined #openstack04:00
*** kashyapc has joined #openstack04:00
uvirtbotNew bug: #615682 in glance "Implement Parallax's register-image API call" [Undecided,New] https://launchpad.net/bugs/61568204:01
*** binaryWarrior has joined #openstack04:01
*** sirp1 has quit IRC04:03
*** abecc has quit IRC04:11
*** binaryWarrior has quit IRC04:11
*** nettok has joined #openstack04:14
*** sophiap has quit IRC04:19
*** sophiap has joined #openstack04:20
uvirtbotNew bug: #615687 in glance "Implement Serializer Adapater for Parallax" [Undecided,New] https://launchpad.net/bugs/61568704:21
*** _cerberus_ has quit IRC04:23
*** armmer has quit IRC04:23
*** miclorb has quit IRC04:24
uvirtbotNew bug: #615689 in glance "Implement Datastore Adapater for Parallax" [Undecided,New] https://launchpad.net/bugs/61568904:30
*** toddmorey has quit IRC04:43
uvirtbotNew bug: #615692 in glance "Implement Teller's fetch-image API call" [Undecided,New] https://launchpad.net/bugs/61569204:46
*** lbieber has quit IRC04:47
*** sophiap_ has joined #openstack05:00
*** sophiap has quit IRC05:00
*** sophiap_ is now known as sophiap05:00
*** miclorb has joined #openstack05:05
*** karmabot has quit IRC05:09
uvirtbotNew bug: #615699 in glance "Implement core Teller functionality" [Undecided,New] https://launchpad.net/bugs/61569905:15
*** randyb has joined #openstack05:19
*** lbieber has joined #openstack05:22
*** pvo has quit IRC05:29
*** sirp2 has quit IRC05:44
*** jdmaturen has joined #openstack05:45
*** nettok_ has joined #openstack05:54
*** nettok has quit IRC05:58
*** jdmaturen has quit IRC06:00
*** metoikos has joined #openstack06:08
*** rnewson has joined #openstack06:08
*** rnewson has joined #openstack06:08
*** nettok has joined #openstack06:18
*** nettok_ has quit IRC06:19
*** kim0 has joined #openstack06:32
*** gholt has quit IRC06:40
*** gholt has joined #openstack06:41
*** jtimberman has left #openstack06:50
*** jdmaturen has joined #openstack06:53
*** jdmaturen has quit IRC06:56
*** brd_from_italy has joined #openstack06:58
*** nettok has quit IRC07:07
*** randyb has quit IRC07:07
*** randyb has joined #openstack07:08
*** rajijoom has joined #openstack07:16
*** miclorb has quit IRC07:28
*** rajijoom has quit IRC07:29
*** ianweller has quit IRC07:31
*** btorch has quit IRC07:32
*** ianweller has joined #openstack07:32
*** btorch has joined #openstack07:33
*** rajijoom has joined #openstack07:46
*** calavera has joined #openstack07:46
*** zheng_li has joined #openstack07:53
*** Paul_Miller has joined #openstack07:58
*** allsystemsarego has joined #openstack08:11
*** kashyapc has quit IRC08:16
*** kashyapc has joined #openstack08:18
*** rajijoom has quit IRC08:18
*** randyb has quit IRC08:19
*** arcane has joined #openstack08:35
*** adjohn has joined #openstack08:43
*** ptremblett has quit IRC08:54
*** rajijoom has joined #openstack09:23
*** adjohn has quit IRC09:27
*** rajijoom has quit IRC09:29
*** rajijoom has joined #openstack09:45
*** adjohn has joined #openstack09:53
*** ptremblett has joined #openstack09:56
*** jonesy_ has quit IRC10:29
*** zul has joined #openstack10:50
alekibangoi dont think parallax is a good name. there is an (not much) active project of that name for smart storage system.11:08
alekibangoit for sure confused me when i seen the name11:10
alekibangoParallax Distributed Virtual Machine Storage11:11
alekibangothat one could be fun to try11:12
*** jonesy_ has joined #openstack11:13
alekibangoi like how they do snapshots, but first make sure the garbage collecting  of unused blocks works :)11:13
*** miclorb has joined #openstack11:14
*** jonesy_ has quit IRC11:15
*** adjohn has quit IRC11:27
*** ctennis has quit IRC11:29
*** zul has quit IRC11:41
*** zul has joined #openstack11:47
*** zul has quit IRC11:53
*** ctennis has joined #openstack11:54
*** miclorb has quit IRC12:17
*** toddmorey has joined #openstack12:22
*** pvo has joined #openstack12:33
*** ChanServ sets mode: +v pvo12:33
*** pvo has quit IRC12:36
*** pvo has joined #openstack12:37
*** ChanServ sets mode: +v pvo12:37
*** kashyapc has quit IRC12:44
*** armmer has joined #openstack13:02
*** zul has joined #openstack13:04
*** codejunkie has joined #openstack13:13
*** jdarcy has joined #openstack13:15
*** gundlach has joined #openstack13:23
*** e1mer has quit IRC13:28
*** lbieber has joined #openstack13:30
*** justinsheehy has quit IRC13:34
*** ctennis has quit IRC13:35
*** ctennis has joined #openstack13:37
*** aliguori has joined #openstack13:37
*** toddmorey has quit IRC13:40
uvirtbotNew bug: #615857 in nova ""nova-manage project create" isn't working" [Undecided,New] https://launchpad.net/bugs/61585713:41
*** e1mer has joined #openstack13:42
*** dendro-afk is now known as dendrobates13:42
*** lbieber has quit IRC13:42
*** kashyapc has joined #openstack13:47
*** aliguori has quit IRC13:50
*** e1mer has quit IRC13:52
zuldid someone mention doing a bof at linuxcon?13:58
jaypipesmtaylor: ping14:06
*** adjohn has joined #openstack14:07
mtaylorjaypipes: pong14:07
jaypipesmtaylor: heya, is there a bzr command to submit something for merge request?  or do I have to use bzr send --mail-to or something?14:08
*** alekibango has quit IRC14:09
mtaylorjaypipes: bzr lp-submit14:09
jaypipesmtaylor: cheers.14:09
mtaylorjaypipes: you need either pipelines installed or bzr 2.2 though14:09
jaypipesmtaylor: yeah, I already use pipeline14:09
jaypipesmtaylor: ha! well, that didn't work quite as expected :)14:14
jaypipesmtaylor: https://code.edge.launchpad.net/~jaypipes/nova/pylint/+merge/3220414:14
jaypipesmtaylor: I submitted the last pipe in the pipeline and it wants to merge the first pipe into the last. :)14:14
mtaylorjaypipes: what command did you do to submit?14:15
jaypipesmtaylor: bzr lp-submit -Rnova-core lp:~jaypipes/nova/nova-tests-apitest14:15
mtaylorjaypipes: ah - the  lp:~jaypipes/nova/nova-tests-apitest was unneeded14:15
jaypipesmtaylor: aha.14:16
mtaylorif you've pushed a branch, it knows what branch it is and looks up the dev focus ... also, it _should_ get -R right without you telling it14:16
jaypipesmtaylor: hmm, I was switch-piped to :first, instead of :last.  Had something to do with it?14:17
mtaylorjaypipes: well... both things had somethign to do with it14:17
*** allsystemsarego has quit IRC14:18
mtaylorjaypipes: a branch argument is always going to be the branch to submit _to_ ... the branch you're in is always going to be the branch to submit14:18
jaypipesmtaylor: hmm, so I branched locally (from my local trunk) to create this pylint branch.  So, I get this:14:18
jaypipes(pylint)jpipes@serialcoder:~/virtenvs/nova/pylint/pylint$ bzr lp-submit -Rnova-core14:18
jaypipesbzr: ERROR: file:///home/jpipes/repos/nova/trunk/ is not registered on Launchpad14:18
mtaylorhave you pushed to lp yet?14:18
mtaylorit's two discreet steps, sadly14:18
*** zul has quit IRC14:19
jaypipesmtaylor: OK, so I'm confused as to *what* I need to push to LP.  I have a jaypipes/nova/pylint branch already up there.  But, I started using pipelines locally, and have 2 pipes ready to submit based on that pylint branch.  All I have done is the lp-submit (which doesn't work, see above)14:20
mtaylorjaypipes: ok. do you want to submit both pipes?14:20
jaypipesmtaylor: yep14:20
jaypipesmtaylor: but separately of course... that's the whole point :)14:20
mtaylorjaypipes: do this "bzr switch :first ; bzr lp-submit lp:nova ; bzr switch :next ; bzr lp-submit lp:nova"14:20
*** abecc has joined #openstack14:21
jaypipesmtaylor: but my :first doesn't have any changes...14:21
mtaylorjaypipes: then there is no need to submit it14:21
jaypipesmtaylor: ok...sorry, I misunderstood you.14:21
mtaylorjaypipes: :)14:21
mtaylorit's early in the morning for me14:21
jaypipesmtaylor: btw, it's switch-pipe, not switch...2 different commands :)14:22
*** adjohn_ has joined #openstack14:22
mtaylorjaypipes: hehe. good call sorry14:22
*** theron has joined #openstack14:23
*** adjohn has quit IRC14:24
*** adjohn_ is now known as adjohn14:24
mtaylorjaypipes: yay. that seems to have gone better14:25
jaypipesmtaylor: indeed :)14:25
*** zul has joined #openstack14:26
jaypipesmtaylor: crap. https://code.edge.launchpad.net/~jaypipes/nova/nova-virt-connection/+merge/3220814:26
jaypipesmtaylor: the diff shows changes for the pipe after that one as well :(14:26
mtayloryou sure?14:27
*** allsystemsarego has joined #openstack14:27
* mtaylor thinks perhaps a push or a commit happenend in the wrong place?14:27
gholtDid somebody fiddle with the blog making it post all comments to planet.openstack.org? Or is it just my reader?14:28
jaypipesmtaylor: yeah, I must have...14:28
* jaypipes notes that bzr-pipeline is a terrible interface.14:28
*** johnpur has joined #openstack14:29
*** ChanServ sets mode: +v johnpur14:29
mtaylorgholt: looking - I haven't personally done anything to planet...14:29
*** adjohn has quit IRC14:30
gholtNo biggie, if you didn't do anything in that area don't worry about it. I was just curious.14:31
jaypipesmtaylor: I think there should just be a bzr lp-submit-all :)14:32
mtaylorjaypipes: yes. I have that on my todo list14:32
jaypipesmtaylor: that enables a dev to just write a commit for each pipe and it automatically handles things..14:32
jaypipesmtaylor: and the fact that bzr switch and bzr switch-pipe are so easily confused (and f**k up the commits if you mistake them) is, well, ass-baggery.14:33
mtaylorwell - switch has been around way before switch-pipe, but yeah14:34
jaypipesmtaylor: just saying it's very easy to screw up the commits because of the similarity :)14:35
jaypipesmtaylor: just realized there is an alias "swp" for switch-pipe.  That will help at least :)14:36
*** mrayzenoss has joined #openstack14:37
*** mrayzenoss is now known as mray14:37
*** mray has joined #openstack14:37
*** alekibango has joined #openstack14:41
*** sirp1 has joined #openstack14:43
*** sirp1 has joined #openstack14:43
*** mray has quit IRC14:44
*** sophiap has quit IRC14:45
*** g0rdy has left #openstack14:45
*** sophiap has joined #openstack14:46
*** lbieber has joined #openstack14:47
gundlacheday: so, you approved https://code.launchpad.net/~gundlach/nova/fix-install-venv/+merge/32155 -- does that mean that it will show up in trunk automatically?  when I run the "merge this branch" command explicitly, I get "nothing to do".14:53
sorengundlach: It's lacking a commit message.14:54
sorengundlach: Feel free to set it.14:54
*** _cerberus_ has joined #openstack14:54
sorengundlach: Once that's set, Tarmac will take care of the rest.14:55
gundlachsoren: ah.  ok, thanks.14:55
gundlachwas there anything informing me of that on the page (besides "set a commit message") or should I just learn to watch for that notification when a branch isn't merging?14:55
gundlachin other words, are there other types of things that can block a merge, and where should i look for those?14:55
*** mrayzenoss1 has joined #openstack14:56
*** mrayzenoss1 is now known as mray14:56
sorengundlach: That's about it, I think.14:57
gundlachsoren: thanks.14:57
sorengundlach: sure thing.14:57
mtaylorgundlach: if anything goes wrong with the merge, tarmac will report it back as a comment to the merge prop14:58
mtaylorgundlach: so if it either doesn't merge cleanly, or if it fails tests for what not14:59
gundlachmtaylor: good to know.  Looks like it merged, in this case.14:59
mtaylorgundlach: woot14:59
gundlachwould have been nice to have a comment saying "waiting for commit message", perhaps, since when it doesn't merge, people can then go check for messages from tarmac.14:59
gundlachthough maybe that's a tarmac-level thing instead of a bzr or launchpad-level thing.14:59
mtaylorit is - and it's something on my todo list to change15:00
*** justinsheehy_ has joined #openstack15:00
mtaylorthe blocking-on-commit message thing is silly ... if there's no commit message, it should just use the merge prop description15:00
*** sirp1 has quit IRC15:00
*** sirp1 has joined #openstack15:01
*** sirp1 has quit IRC15:01
*** sirp1 has joined #openstack15:01
*** justinsheehy_ has quit IRC15:05
*** justinsheehy has joined #openstack15:07
*** justinsheehy_ has joined #openstack15:13
*** justinsheehy has quit IRC15:14
*** justinsheehy_ is now known as justinsheehy15:14
uvirtbotNew bug: #615906 in nova "pip-requires and virtualenv don't play together nicely" [Undecided,New] https://launchpad.net/bugs/61590615:21
*** calavera has quit IRC15:27
*** rajijoom has quit IRC15:28
mtaylorgundlach: ok. done15:37
gundlachmtaylor: impressive :)15:37
mtaylorgundlach: got tired of this being a blocker - is now fixed and installed15:37
mtaylorsure thing!15:37
creihtmtaylor: about time! ;)15:37
mtaylornow - it may still be desirable for people to set commit messages- if that merge request description is a novel and stuff15:37
mtaylorbut I think that can be a yell-at-each-other sort of thing :)15:38
mtaylorcreiht: I know. I'm slack ;)15:38
*** zul has quit IRC15:39
*** zul has joined #openstack15:41
*** gundlach has quit IRC15:46
*** gundlach has joined #openstack15:46
*** kashyapc_ has joined #openstack15:53
*** binaryWarrior has joined #openstack15:53
*** kashyapc has quit IRC15:54
*** lbieber has quit IRC15:56
*** brd_from_italy has quit IRC15:56
*** lbieber has joined #openstack16:02
*** e1mer has joined #openstack16:02
*** e1mer has quit IRC16:04
*** PiotrSikora has quit IRC16:09
*** zheng_li has quit IRC16:11
*** justinsheehy has quit IRC16:13
*** PiotrSikora has joined #openstack16:14
*** e1mer has joined #openstack16:15
*** heckj has joined #openstack16:18
heckjhowdy - I have a merge request moved into approved state - wasn't sure if I needed to do anything to move it into the trunk. Didn't know if that was a person invoking it, or an automatic process.16:20
*** e1mer has quit IRC16:20
mtaylorheckj: nope. it's automated16:20
heckjOkay - out of curiousity, how does it work? I was going to continue cleaning, but that branch hasn't been merged into the trunk (that I can see) as yet.16:21
mtaylorheckj: ah - it's not actually in approved state - it's just had an approve vote cast for it16:21
heckjOkay - thanks!16:22
heckjIf I do more cleaning tonight, I'll just open a new branch to keep from confusing things16:22
mtaylorpoppiing it to approved real quick - it should be in trunk in like 5 minutes16:22
heckjcool - thanks.16:24
heckjWhat's proper ettiquette from here? Continue working on this branch, or close it and open a new one?16:24
mtayloreither way works - if it's still pylint cleanup, I say just keep working in the same branch16:25
*** toddmorey has joined #openstack16:33
*** zul has quit IRC16:33
*** kashyapc_ has quit IRC16:34
*** e1mer has joined #openstack16:35
*** bitmonk1 has joined #openstack16:38
*** e1mer has quit IRC16:39
jaypipesmtaylor: hope you don't mind I put you in touch with the Citrix folks about their Hudson stuff.16:39
mtaylorjaypipes: not at all - I haven't responded yet, but will16:40
jaypipesmtaylor: cheers16:40
*** binaryWarrior has quit IRC16:40
*** zheng_li has joined #openstack16:44
*** lew has joined #openstack16:45
*** e1mer has joined #openstack16:49
*** e1mer has quit IRC16:52
*** rajijoom has joined #openstack16:58
*** metoikos has quit IRC17:03
*** jamiew has joined #openstack17:06
gundlacheverybody: dubs sirp1 and i have been talking about the Right Way to support serialization to JSON/XML/other in a wsgi stack.17:07
gundlachthe API Endpoint system will need to support this, and it would be nice if the code were reusable in other parts of the system.17:07
gundlachwe thought of having a wsgi middleware that detects Accept: header (or URL suffix like .xml) on the request, gets the response from the wsgi app below it, then serializes to the proper format.17:08
gundlachbut that raises a question: in what format did the "wsgi app below it" return its results?  the wsgi spec requires that responses be an iterator of strings.17:08
gundlachso the wsgi stack can't just be passing a dictionary up until the serializer can turn that into json or XML.17:09
gundlachsirp1 suggested that the lowest-level app (the one that fetches models from the DB) would pickle the response, and let that percolate up to the serializer.17:09
gundlachthat feels a little weird to me -- passing an opaque pickle object through a WSGI stack -- but i can't come up with a good argument against it.17:10
gundlachanyone have a good reason why we shouldn't go this route?17:10
mtaylorspy: ping... see my note about dns?17:10
spymtaylor: yup, fixing to do it, got tied up a in few meetings this morn17:11
gundlachanother approach would be for the serialize middleware to record the Accept type in the environ during the request phase, and then the lowest-level app would call serialize_my_response_dict() which would serialize to the requested Accept type.  the downside is that each low-level app will have to add that line.17:11
mtaylorspy: meetings suck17:11
* mtaylor replaces all meetings in his day with vodka17:11
spymtaylor: yeah, i've been good at dodging them, but they occasionally catch me17:12
* mtaylor imagines people running through the castle whic large inflatable meetings float around trying to snag people17:12
spythat's not far from the truth :)17:13
spymtaylor: ok, dns updated, should take about 30 minutes or so17:14
mtaylorspy: rock17:16
mtaylorspy: thanks!17:16
mtaylorchromakode: ping17:17
mtaylorchromakode: I restarted the karmabot server and it's not running anymore- could you give me a note on how it starts/runs?17:17
*** tobym has joined #openstack17:21
*** holoway has quit IRC17:22
*** bitmonk has joined #openstack17:23
*** holoway has joined #openstack17:23
_cerberus_gundlach: that would mean that each middleware up the chain would have the deserialize, append, and reserialize, right?17:28
gundlach_cerberus_: yes, if it wanted to modify the response body.  but i don't know of too many that do.17:28
*** h1nchtastic has joined #openstack17:29
gundlachalso, as i described the problem to the channel, i realized that it sounds a lot like serializing to JSON/XML/whatever is really not the job of a WSGI middleware... and I can't find an example of a middleware-based serializer on the internet, which makes me think i'm right17:29
gundlachso i may be leaning at the moment toward just having the lowermost wsgi app do something like17:30
gundlachmyDict = go_calculate_the_response()17:30
gundlachreturn SerializerByAcceptType(request).serialize(myDict)17:30
gundlachaka, there's no middleware at all, and I provide instead a class that can figure out how to serialize.17:31
gundlachand maybe it'll be a decorator on a method instead, so the method can return the dict and the @serialize_by_accept_type decorator takes care of the rest.17:31
gundlach(which is kind of like the pattern in TurboGears, except that they can only serialize to JSON.)17:31
_cerberus_gundlach: I wasn't aware that TurboGears already had a solution of sorts. Is the only distinction that they only serialize to JSON?17:32
*** zul has joined #openstack17:34
gundlach_cerberus_: correct -- they don't read the accept header, and instead just provide a decorator that turns a dictionary into a JSON object.17:35
gundlach_cerberus_: incidentally re: building the testing framework that we talked about yesterday: i was pretty sure webob provided a default request environment so you wouldn't have to fire up an HTTP server to do your testing.  It's webob.Request.blank('http://google.com') -- the resulting Request is a full web request and its .environ is fully loaded.17:36
gundlachyou can pass it to the app to test and the app won't know it wasn't called by a web server.17:36
_cerberus_gundlach: The turbogears solution looks pretty cool.17:37
sirp1for comparison: swift handles serialization directly in the application itself (see account.py for example), no serializeByAcceptHeader or middleware17:37
_cerberus_And also, awesome find17:37
gundlachsee http://pythonpaste.org/webob/jsonrpc-example.html [search for 'blank Request']17:37
creihtgundlach: Sorry, have been at lunch, but I think you have come to the right conclusiong to not do serialization in middleware17:37
gundlachcreiht: cool -- care to elaborate?17:37
creihtgundlach: You might also look at restish to see how they handle differnt accepts17:38
gundlachcreiht: ty17:38
creihtat least I like how they do it :)17:38
gundlachsirp1: where is account.py?17:38
gundlach(viewing files in LP so no 'find' command available ;) )17:38
sirp1swift code base: swift/account/server.py actually :)17:39
gundlachah hah17:39
creihtgundlach: but most frameworks take the approach of serializing at that level (also like the TG example that you used)17:39
gundlachsirp1: ah, i see, swift is actually building XML by hand.17:41
gundlachcreiht: ok.17:41
notmynamegundlach: and you'll see that swift uses query parameters to determine the serialization format (rather than accept headers)17:42
gundlachnotmyname: cool.  i'd like the serializing function to read whatever it needs to -- accept headers, or suffix, or qsvars.17:43
gundlachfor nova's RS API, i'll hard-code it to check the accept headers and then the suffix (which is what the RS API specifies) but we should then extend it to work for swift as well.17:43
notmyname(I may be wrong about swift only using query args)17:43
gundlachnotmyname: looks like it tries the ?format param first, then falls back on webob.Request.accept17:44
notmynamejust looked at the code17:44
notmynameya, that17:44
notmynamethe worst part: I wrote at least some of that code....17:45
gundlachnotmyname: the best part: it works :)17:45
notmynamethat must have been the part I didn't write ;-)17:45
gundlachno one is expected to remember what they wrote the day after they wrote it17:45
gundlachthat's what comments are for :)  at least in my case anyway.17:45
notmynamehehe. true17:45
*** justinsheehy has joined #openstack17:47
gundlachcreiht: having trouble finding the Accept-related bits on http://ish.io/embedded/restish/ .  Can you point me to them, or just tell me how they handle serialization?18:06
*** toddmorey has quit IRC18:11
*** zul has quit IRC18:12
*** zul has joined #openstack18:12
creihtgundlach: http://ish.io/embedded/restish/resources.html#resource-url-handling18:13
creihtand search for content negotiation18:13
creihtYou would still have to call some serialization functionality at that point, but it handles the content negotiation piece18:14
creihtAnd then I had middleware that would override the accept header based on suffix, or query args18:15
holowaycreiht: yay for using Accept headers18:16
creihtThe above was in a previous life18:16
*** justinsheehy has quit IRC18:17
*** zul has quit IRC18:24
*** justinsheehy has joined #openstack18:25
*** dendrobates is now known as dendro-afk18:26
*** karmabot has joined #openstack18:33
_cerberus_gundlach: are you still working on the resource mapping or have you moved onto the specific controller components?18:38
gundlach_cerberus_: still doing resource mapping -- i have been reading Routes' instructions, because i can't quite see how it's supposed to work.18:38
*** dendro-afk is now known as dendrobates18:38
_cerberus_gundlach: gotcha18:38
gundlachwhy -- you want to grab it?18:38
_cerberus_Nope, curious mostly. Looking at starting the servers controller18:38
*** dendrobates is now known as dendro-afk18:39
*** dendro-afk is now known as dendrobates18:39
gundlach_cerberus_: cool -- just make it conform to the list on http://routes.groovie.org/restful.html#resource-options and i should be able to tie in.18:46
gundlachsorry that i haven't figured it out yet!18:46
_cerberus_gundlach: no problem dude. Just trying to figure out where we are. Obviously I want to avoid starting something and having to rewrite 90% to conform to what you're setting up18:46
gundlachin addition to the CRUD methods, i guess calling the other actions 'reboot()' and 'resize()' etc would work -- i'll just map those individually, manually.18:47
*** gustavomzw has joined #openstack18:51
*** metoikos has joined #openstack18:53
PythonPupHowdy.  I am about equally interested in Nova and Swift.  Which is easier to setup?  I'll start with that one.  I asked this once before, and I think someone answered, but it scrolled off before I got back to the chat program.18:59
*** mrayzenoss has joined #openstack18:59
*** mrayzenoss has joined #openstack18:59
holowayPythonPup: swift19:00
PythonPupThanks, holoway19:00
*** rajijoom has quit IRC19:00
*** mray has quit IRC19:02
PythonPupoops, I meant that for another channel.19:06
*** pvo_ has joined #openstack19:08
*** pvo has quit IRC19:08
*** pvo_ is now known as pvo19:08
*** ChanServ sets mode: +v pvo19:08
sorenWooo! Nova was accepted into Ubuntu Maverick.19:16
alekibangowhat will that mean for us?19:17
alekibangoi dont see it, where did you read this?19:18
*** mrayzenoss is now known as mray19:18
alekibangoah, found some19:18
sorenalekibango: It means that going forward, it doesn't need review by an archive administrator. I can just upload new version at will.19:23
mtaylorsoren: w00t19:23
sorenmtaylor: w00t²19:25
uvirtbotNew bug: #616024 in swift "Add metadata facility to account and container servers" [Wishlist,In progress] https://launchpad.net/bugs/61602419:25
*** ptremblett has quit IRC19:25
*** zul has joined #openstack19:36
alekibangosoren: thanks, good news19:48
*** mtaylor has quit IRC19:55
*** justinsheehy has quit IRC19:57
*** mray has quit IRC20:07
*** dendrobates is now known as dendro-afk20:10
*** mrayzenoss has joined #openstack20:10
*** mrayzenoss is now known as mray20:10
*** mray has joined #openstack20:10
*** mtaylor has joined #openstack20:13
*** ChanServ sets mode: +v mtaylor20:13
*** zul has quit IRC20:19
*** zul has joined #openstack20:20
* mtaylor is pleased to announce http://paste.openstack.org/ ... go nuts20:28
*** ChanServ sets mode: +o mtaylor20:28
*** mtaylor changes topic to "Find out more and get involved here: http://www.openstack.org | Nova Docs: nova.openstack.org | Swift Docs: swift.openstack.org | Wiki: wiki.openstack.org | http://wiki.openstack.org/LifeWithBzrAndLaunchpad | http://irclogs.openstack.org/logs | http://paste.openstack.org/"20:28
cory_ok, #2 got it20:33
*** lbieber_ has joined #openstack20:34
mtaylorstewart: https://code.edge.launchpad.net/~mordred/drizzle/rplugin-innobase/+merge/3225720:34
mtaylordamn. wrong channel20:34
notmynamemtaylor: your pastebin crashed20:35
creihtworks for me20:35
mtaylornotmyname: yeah? works for me?20:35
notmynameI get a 50020:35
* mtaylor doesn't want his pastebin to crash...20:35
notmynamecreiht saw it too (on my screen)20:36
creihtMust be because he is using a mac20:36
*** lbieber has quit IRC20:36
*** zul has quit IRC20:37
notmynameI insist on using a laptop bathed in unicorn tears20:37
mtaylornotmyname: ok. I'm running through the log and not seeing any tracebacks...20:37
redboworked fine on my mac.  He must be using safari.  My browser isn't bathed in unicorn tears.20:38
mtaylorI see a traceback - but it means very little to me at the moment...20:38
mtaylorInvalidRequestError: Could not evaluate current criteria in Python. Specify 'fetch' or False for the synchronize_session parameter.20:38
mtaylormacs always cause me grief :)20:38
creihtok now it isn't loading for me20:39
notmynameya, now it's hanging20:39
mtayloralright. I officially rescind the laungh :)20:39
redbowe've been using lodgeit with mod_wsgi for forever internally for swift and haven't had any problems with it20:39
mtaylorcreiht: try again?20:39
creihtok now it loads20:40
mtaylorok. I think that freeze was me having frozen stdout in screen :)20:40
mtaylorwhich is ghey - but whatever20:40
*** zul has joined #openstack20:40
mtaylorthe problem seems to be coming in fetch_replies - I'll see if I can figure out what madness is going on20:40
* mtaylor waves his fist at running software directly from version control20:41
*** ctennis has quit IRC20:42
mtaylormakes no sense... why would it matter what browser you're using. sigh20:42
* mtaylor hates the web20:42
notmynamemtaylor: Safari debug console shows the 500 on /favicon.ico20:44
mtaylorperhaps I should just add a favicon.ico20:45
notmynametouch favicon.ico <-- done20:45
gholtWell, the lack of a favicon doesn't blow up my Safari...20:46
*** ctennis has joined #openstack20:52
*** xtoddx has joined #openstack20:54
xtoddxcreiht: you around?20:55
creihtxtoddx: yo!20:55
*** toddmorey has joined #openstack20:56
xtoddxso i've been looking at moving auth to nova20:56
mtaylornotmyname: I'm sadly having to learn more about werkzeug than I really wanted to know - back in a jiff20:56
xtoddxthinking that account ring will be managed by another process and validated by the middleware20:56
xtoddxbut first issue I'm coming into is what url to return with auth, since nova and swift have potentially different endpoints20:57
creihtxtoddx: Is there a reason it can't return a compute endpoint and a storage endpoint?20:58
notmynamebut not in headers (please)20:58
xtoddxnotmyname: how will it return them then?20:59
notmynamein the body of the response. I only say that because the rackspace cloud auth uses header values, and it's not very future-friendly (what with multiple global swift clusters and all)21:00
xtoddxright now the specs for both expect headers: X-Storage-Url and X-Server-Management-Url21:00
creihtso lets back up a sec21:01
zulis nginx still required?21:01
creihtxtoddx: Are we talking about a more generalized auth solution for both, or hooking swift into the current auth in nova?21:02
xtoddxi think we need one endpoint that works for both systems21:02
*** jdarcy has quit IRC21:03
xtoddxSo I'm willing to change nova systems to use the rackspace way of auth.21:03
mtaylornotmyname: would you try again real quick for me?21:03
notmynamemtaylor: works21:04
mtaylornotmyname: woot21:04
creihtxtoddx: Are we sure that is the best idea?21:04
mtaylorstill haven't solve favicon.ico - but I care less about thtat now21:04
xtoddxcreight: i like it.  and as we move nova to the openstack API we need the same type of system, so why not make it the same system for both?21:05
creihtWell it has some drawbacks21:05
creihtplus side is that it is simple21:05
creihtminus side is that you can't do things like signed urls21:06
xtoddxwhats a signed url?21:06
creihtlike a one time use url21:07
creiht(for storage)21:07
creihtmaybe that is a bad example21:07
xtoddxi don't see why it would block that21:07
xtoddxit might just make the signature live in some other storage other than account/auth21:08
redbothe other main drawback is you have to use https, since you don't want anyone getting the storage token21:09
*** mtaylor has quit IRC21:10
redboWell, maybe less of a problem for you, but it'd be nice for us to offer http, but it'd be pretty insecure compared to request signing.21:10
xtoddxrequest signing is using a secret key to sign the body of the request and put the signature in the header?21:11
creihtI was thinking that more people would want something like oauth21:11
creihtxtoddx: more or less21:12
redboUsually you just do something like take the request headers and timestamp and secret key and hash them all and stick that in the headers.21:12
xtoddxi don't see why a central auth system would interfere with that, you can get from a session token to a user record, which will have the secret token21:12
redbookay, when you say "the rackspace way", you imply a lot of things to us that we don't necessarily like :)21:13
* bitmonk chuckles21:14
xtoddxha, okay.  so lets do something cool going forward21:14
*** Paul_Miller has quit IRC21:14
creihtNow that is the talk we like to hear :)21:14
*** toddmorey has quit IRC21:14
xtoddxso I should read up on oauth, and how to implement it in a command line / library environment?21:15
creihtI dunno... who decides what is the "right" way to do auth for openstack21:15
redbowhoever implements it first.  GO!21:16
creihtThe other option is that we could do the rackspace way right now, since that would be the easier/fastest thing until openstack decides which way to go with auth21:16
xtoddxyea, probably RS way for now, with a suggestion for going forward21:16
creihtAnd at that point actually, we could have to auth variants21:17
xtoddxwe'd probably have to make a blueprint and get it reviewed by the tech board?21:17
gholtIs it still true that OAuth 2 isn't yet final?21:17
* vishy jumps into the middle of the conversation21:17
creihtyay for colaboration :)21:17
*** dendro-afk is now known as dendrobates21:18
xtoddxoauth2 looks draft, but a late stage draft we could probably work from21:18
xtoddxthere are some implementations already21:18
gholtYeah, still draft, but... yeah, what xtoddx said, heh. http://oauth.net/2/21:18
*** mtaylor has joined #openstack21:19
*** ChanServ sets mode: +v mtaylor21:19
creihtit is more or less what google uses21:19
creihtthough they still use v1.021:20
gholtYeah, 2 is the way to go for new stuff, as the chatter goes.21:20
*** zul has quit IRC21:22
gholt"considered stable but with many features being changed or added", seems a bit of a contradiction there21:22
xtoddxpfff, logic, who needs it21:22
*** theron has quit IRC21:22
*** theron has joined #openstack21:22
gholtThe flag for me is that they indicate 1 did not scale all that well, but 2 should.21:23
*** xtoddx has left #openstack21:23
*** xtoddx has joined #openstack21:23
*** mtaylor has quit IRC21:27
*** theron has left #openstack21:27
creihtwell this is funny, turns out that OAuth WRAP is pretty close to what RS Auth does now :)21:32
creiht2.0 of course deprecates it21:33
*** mtaylor has joined #openstack21:34
*** ChanServ sets mode: +v mtaylor21:34
gholtDo we want to move away from token-based? Or continue to support it?21:34
holowaycan you use oauth outside of http?21:36
holowayI imagine not21:36
gholtNot with that spec.21:37
*** cglee has joined #openstack21:40
*** mtaylor has quit IRC21:41
gholtI'm having a hard time sifting it all fast, but I'm guessing OAuth1 = sign-per-request and OAuth2 = token-per-session required https.21:42
gholtDo we have anybody familiar with this stuff?21:42
*** pvo has quit IRC21:46
*** metoikos has quit IRC21:47
*** gustavomzw has quit IRC21:50
heckjI'm pretty familiar with OAuth - using it extensively21:59
heckjOauth1 (hammer spec) has a nice setup for service to service authorization that's really light, might be applicable21:59
heckjoauth2 is moving pretty fast with the spec, but is actually much simpler to use - a lot less complex signing requirements22:02
heckjoauth2 is at rev 10 on the spec last I looked - but not final22:02
gholtI'm wondering if having code for both is best? Some might want the extra security 1 provides. But I'm going on very limited knowledge of the two specs, heh.22:03
gholtEssentially, current Swift is oauth2-like. Authenticate with auth server for a token, use token as authorization with resource server, resource server validates (and caches) token with auth server.22:07
*** ptremblett has joined #openstack22:07
gholtAll over https, of course so the token isn't sniffed.22:07
redboso there's no signing with oauth2?22:08
gholtThere's signing with the authentication part to get the initial token, as I understand.22:08
gholtOr, there can be signing, I should say.22:08
edaywreese: hey, you around?22:14
*** cglee has quit IRC22:16
edaywreese: you still in the castle, and if so, have a min to chat? I'm over in the ozone area22:16
creihteday: he's on the other side of the partition from you if you want to walk on over :)22:17
gholtYou're sending him towards George probably... heh22:18
creihteven better :)22:18
gundlachout, bye22:18
*** gundlach has quit IRC22:18
gholtheckj: Is there an oauth server out there most folks use? Or does everybody just provide an oauth layer on top of their own user databases? or?22:21
heckjmostly it's a light oauth layer on top of your application - python-oauth2, etc.22:22
gholtSo, for OpenStack, currently just swift and nova, we'd make a new project for the oauth server they'd both use which could be ignored if the installer has their own oauth server already?22:24
creihtWell I think we have identified that we need a reference auth server for both swift and nova to use together22:26
creihtthe question is what does that look like? Oauth?22:26
heckjyeah, pretty much. We've done an oauth token store and an authorization service that responded with the Xauth headers from a plugable store22:27
creihtI have to admit, now that I have looked at it, the whole oauth landscape is a bit confusing22:27
heckjit is22:27
gholtYeah, I was hoping there was already an open source oauth 1 and 2 server out there we could just use.22:27
*** stewart has quit IRC22:28
*** lbieber has joined #openstack22:29
*** rnewson has quit IRC22:33
*** toddmorey has joined #openstack22:39
*** pvo has joined #openstack22:42
*** pvo has joined #openstack22:42
*** ChanServ sets mode: +v pvo22:42
*** mtaylor has joined #openstack22:42
*** ChanServ sets mode: +v mtaylor22:42
*** pvo has quit IRC22:43
*** toddmorey has quit IRC22:45
mtaylorI know it's not fully openstack related yet - but it's a real PITA to do both a .com and a .org that are the same thing in RS Cloud Sites/DNS ...22:47
*** toddmorey has joined #openstack22:53
*** mray has quit IRC22:53
*** miclorb has joined #openstack22:55
vishyyay for pylint22:57
mtaylorvishy: it's when it does something actually useful that all the rest of it is worth it :)22:59
vishymtaylor: i believe you23:00
*** RobertLJ has joined #openstack23:03
vishyis cbehrens on here?23:08
*** toddmorey has quit IRC23:09
*** toddmorey has joined #openstack23:09
*** codejunkie has quit IRC23:10
*** tobym has quit IRC23:10
*** codejunkie has joined #openstack23:12
*** lew has quit IRC23:16
*** hybridpollo has joined #openstack23:19
*** viewlogic has quit IRC23:21
*** viewlogic has joined #openstack23:22
*** jamiew has quit IRC23:25
*** gustavomzw has joined #openstack23:26
*** zheng_li has quit IRC23:26
*** hybridpollo has quit IRC23:27
*** lbieber has quit IRC23:30
*** tobym has joined #openstack23:34
*** allsystemsarego has quit IRC23:41
*** metoikos has joined #openstack23:42
*** johnpur has quit IRC23:52

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!