Monday, 2013-09-16

*** oz_akan_ has joined #openstack-marconi00:30
*** oz_akan_ has quit IRC00:31
*** oz_akan_ has joined #openstack-marconi00:32
*** oz_akan_ has quit IRC00:36
*** zyuan_ has quit IRC00:53
*** oz_akan_ has joined #openstack-marconi01:25
*** ayoung has quit IRC02:13
*** oz_akan_ has quit IRC03:50
*** oz_akan_ has joined #openstack-marconi03:51
*** oz_akan_ has quit IRC03:55
*** ekarlso- has quit IRC07:07
*** flaper87|afk is now known as flaper8707:15
*** ykaplan has joined #openstack-marconi07:35
*** gordonsim has joined #openstack-marconi08:10
*** cthulhup has joined #openstack-marconi08:21
*** cthulhup has quit IRC08:39
*** _alexr_ has joined #openstack-marconi09:26
al-maisanflaper87: Good morning -- are you joining us for the call at 17:00 CEST?09:27
flaper87al-maisan: #FuckYeah09:27
al-maisanhow is the rain in Como?09:28
flaper87al-maisan: It rained pasta & pizza yesterday but today we've got a beatiful, sunny, not so hot, perfect weather day09:29
flaper87Does that make you jealuos enough ?09:29
flaper87I spent my sunday coding at home09:29
al-maisannothing beats the fine Zurich drizzle ;)09:30
flaper87well, I'd have done the same even if it was sunny09:30
flaper87al-maisan: lol09:30
al-maisanthe "sunday coding at home" bit *does* make me jealous09:30
flaper87Mission Accomplished09:31
* flaper87 spent his sunday fixing Rust bugs and discussing `select` based timers09:32
flaper87al-maisan: do you work remotely or at RAX office?09:32
flaper87(assuming there's one in Zurich)09:32
al-maisanyeah, I wanted to ask you about Rust .. what is it that attracts you to it?09:32
* al-maisan is in the Zurich office today09:32
al-maisantomorrow: remote work09:33
flaper87al-maisan: Rust's memory management and type inference it's pure poetry09:38
flaper87The syntax is nice, there are a couple of things I'd change but it's under heavy development09:38
flaper87towards a 1.0 stable release and hopefully production ready09:38
al-maisanWow! After your glowing endorsement I'll need to take a look :)09:43
flaper87al-maisan: :D09:43
* flaper87 used to read poems in high school, he now reads code.09:44
flaper87actually, that last msg deserves a tweet09:44
al-maisanhmm .. hmm .. reading good code can be very enjoyable .. but maybe not at a poem level :)09:45
*** ykaplan has quit IRC10:29
*** ykaplan has joined #openstack-marconi10:52
*** ykaplan has quit IRC11:42
*** oz_akan_ has joined #openstack-marconi11:49
*** oz_akan_ has quit IRC11:53
*** tedross has joined #openstack-marconi11:55
*** ykaplan has joined #openstack-marconi11:55
*** ykaplan has quit IRC12:29
*** acabrera has joined #openstack-marconi12:30
acabreraGood morning. :)12:34
*** ykaplan has joined #openstack-marconi12:36
*** oz_akan_ has joined #openstack-marconi12:51
*** vkmc has joined #openstack-marconi13:08
*** amitgandhi has joined #openstack-marconi13:12
*** jergerber has joined #openstack-marconi13:32
*** oz_akan_ has quit IRC13:39
*** oz_akan_ has joined #openstack-marconi13:40
*** malini has joined #openstack-marconi13:40
flaper87yo yo yo13:44
acabreraflaper87: o/13:47
*** ayoung has joined #openstack-marconi13:49
flaper87acabrera: gooood morning13:51
acabreraI read the openmic that were featured in, flaper87. When I saw the gummy bears surrounding the coffee mug, I had no doubt it was you. :P13:52
acabreraAlso, I thought - "Man, that's an awesome wall paper!"13:52
flaper87gummy bears, it's like, like, like, like.... Can't even describe them. Superb... :D13:53
* flaper87 depends on those things13:53
flaper87btw guys, I submitted a patch for ddt
flaper87it just got approved13:54
flaper87I was suggested to use testscenarios which is what other projects use and seem to work better than ddt13:54
flaper87Can any of you take care of that migration before tests using ddt get bigger?13:54
acabreratestscenarios - hmmm...13:55
flaper87I've never used it13:55
flaper87so, I'd love to know what you guys think13:55
acabreraI need to read up on that a little more. The PyPI docs on it made me a little weary.13:55
maliniI am looking at it now13:55
acabreraThey were a bit scattered. :x13:55
acabreraIMHO, anyway.13:55
acabreramalini: awesome!13:56
malinithe testscenarios doc make ddt docs look like bible ^o)14:03
maliniflaper87: do you know of any openstack  history with pytest?14:05
flaper87malini: not at all14:06
* flaper87 is so waiting for that ddt patch to land14:11
flaper87damn it14:11
maliniis cool!!14:14
flaper87yeah, py.test is cool14:14
maliniBut I'll try testscenarios first we conform to OS14:15
flaper87it's not that we have to14:15
flaper87but, you know, other projects are using it, there's to be a why14:15
flaper87so to speak :D14:16
maliniyeah..thts what I thought too14:16
malinithere might be a reason why others arent using something common like pytest14:16
*** amitgandhi has quit IRC14:19
*** amitgandhi has joined #openstack-marconi14:20
acabreratestscenario was developed in 2009. It might have solved a problem that was really bad at that time. I'd love to hear what problem it solves now. It's definitely maintained, as I observed from the launchpad/PyPI page, and it supports Python 3 (+1), but if we don't need it...14:23
acabreraAs for pytest, I'm not sure on that one. I've done mild comparisons of it to nose in the past. IMO, they're on equal footing. If you need a test runner, pick one of the two and run with it. :)14:24
malinipytest seems to be have better support for tests that are beyond the scope of unit tests14:25
acabrera[pytest v. nose]: The reason one is chosen over the other might be familiarity, in that case. Or it might be plugin compatibility (does pytest work with testr?)14:25
flaper87acabrera: plus, nosetest has been used since the very beginning14:29
flaper87there's some hate around nose anyway14:29
flaper87but you know, migrating all projects would be a real adventure14:29
acabreranose hate, flaper87? I haven't heard of it. What's being said?14:29
* acabrera curious about pain points14:30
maliniI will play around with pytest & chk on testr compatibilty & stuff14:30
flaper87mmh, TBH, dunno what are the arguments against it but I know there's some14:30
acabreraahh, alright. If you knew the specifics, I'd love to hear them. :)14:31
flaper87acabrera: I know mordred doesn't like nose that much14:32
acabrerathanks - I may bug him about it at some point!14:34
acabreraFor the moment, I'm going to check the review queue.14:34
acabreraThere's some rebasing I need to do on proxy patches, and I'm sure there's things that need reviewing.14:35
*** ametts has joined #openstack-marconi14:38
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: marconi-proxy forwarding
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: define interfaces for proxy storage
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: WIP: add mongo driver for marconi proxy storage home ..
*** kgriffs_afk is now known as kgriffs14:43
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): partition related endpoints
kgriffsflaper87: did you already register a bug re ddt => testscenarios ?14:52
flaper87kgriffs: erm, nope :( sorry about that14:53
kgriffsok, just trying to keep track of things. Could you do that real quick?14:53
flaper87kgriffs: btw, dunno if you noticed it, I wen ahead and marked bugs with Fix Committed as Fix Released14:53
flaper87kgriffs: I'll do that14:53
kgriffsflaper87: yeah, saw that. I guess we can just do that until we have our first real release14:54
flaper87malini: ^14:55
flaper87kgriffs: correct14:55
maliniflaper87: thanks! I assigned me to it14:55
flaper87malini: awesome, thanks for taking care of it14:56
zyuanChanServ: what's the purpose of having multiple url for each node?14:56
zyuanacabrera: ^^14:56
acabrerazyuan: :P14:56
acabreraI'll explain that.14:56
acabreraThe reason for having multiple URLs registered for each partition is to improve availablity. For example, if those URLs were merely haproxy nodes, then marconi-proxy can remove a failed haproxy node from that nodes list and still have a fallback available while the deployer fixes the haproxy problem.14:58
zyuanit might be better to have some HA selection algo instead of having back-up urls...14:58
zyuani'm not familiar with these.  but  i got it.14:59
acabrerait's something to revisit later. Primarily, I'd love to leave complex HA logic out of marconi-proxy to the greatest extent possible, while allowing marconi-proxy to be pluggable into an HA deployment.15:00
flaper87al-maisan: kgriffs 2 secs, I'll get my ipad15:01
kgriffsflaper87: not seeing al-maisan online?15:06
amitgandhikgriffs: flaper87: you guys talking foghorn?15:06
kgriffswe were going to, but I am starting to wonder if the calendar invite got flubbed15:06
al-maisankgriffs: ready now15:07
kgriffso hai!15:07
al-maisanhello :)15:07
kgriffsbtw, I don't use my @rackspace account for google much - better to use my @gmail - sffirgk15:08
amitgandhimind if i join too?15:09
al-maisanamitgandhi: please do ..15:09
amitgandhiG+ ?15:09
al-maisankgriffs: gotcha15:09
*** openstackgerrit has quit IRC15:13
*** briancline has quit IRC15:14
*** briancline has joined #openstack-marconi15:17
*** openstackgerrit has joined #openstack-marconi15:30
*** openstackgerrit has quit IRC15:30
al-maisanamitgandhi: thanks!15:31
*** openstackgerrit has joined #openstack-marconi15:31
al-maisanflaper87: can you please give me the meetup link as well?15:31
amitgandhiit was just something i did as a concept for the last hackday so its very rough15:31
al-maisanamitgandhi: understood!15:32
al-maisanthanks flaper87 !15:32
* flaper87 will talk about MongoDB in Marconi15:32
flaper87The title of the outline is: "Steps out of hell"15:32
zyuanacabrera: found another tiny issue15:34
zyuanthe url can not have a trailing /15:35
zyuanlike http://...:port/15:35
kgriffsflaper87: LOL15:36
al-maisanflaper87: that's gotta be a hot talk ;)15:36
al-maisanor .. "a hell of a hot talk" :-P15:37
acabrerazyuan: it's unintentional.15:38
flaper87what the hell is going on with the gate15:38
zyuanhmm, and there is no way to clean up catalog after a test -- unless i do a flushdb() from redis15:39
zyuaneither i refactor it or just add the call for now...15:39
acabrerazyuan: That's right. You're going to need to make that part of teardown.15:39
zyuanacabrera: i usually send DELETE15:40
kgriffsreminder: teem mtg in 15 mins over in #openstack-meeting-alt15:45
acabreraflaper87: If I wanted to set an oslo.config option via tinkering in the CLI (to interact with oslo.cache), how would I achieve that? :x15:45
kgriffsis it just me, or is pypi acting up again?15:58
flaper87kgriffs: back, sorry15:58
Alex_Gaynorkgriffs: yes, it's know that pypi is acting up, see dstufft and I in -infa15:58
flaper87kgriffs: what Alex_Gaynor said15:59
flaper87kgriffs: re oslo.config15:59
flaper87kgriffs: Is conf.register_cli_opts() what you need?15:59
kgriffsflaper87: I believe acabrera was looking for that16:00
flaper87that'll register the option as a CLI parameter and as a config parameter16:00
flaper87acabrera: ^16:00
kgriffsi think he wants to set it on the fly from pdb, tho16:00
kgriffs(set a value)16:00
acabrerafrom python-cli, specifically.16:00
acabreraDoing a little experimentation with oslo.cache and oslo.cfg is getting in the way. D:16:00
kgriffsmtg time16:01
acabreraflaper87, kgriffs: I'll look at this some more after the meeting. I'm migrating marconi-proxy to use oslo.cache today.16:01
maliniflaper87: Some of the queue tests are missing cleanup/teardown eg. . This was unintentional, rt ?16:08
kgriffsmalini: ?16:10
*** flaper87 is now known as flaper87|afk16:12
*** _alexr_ has quit IRC16:28
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): catalog
acabrera+1, kgriffs16:47
acabreraIt solidifies our midterm goals16:47
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): catalog
*** ekarlso has joined #openstack-marconi16:49
kgriffsguys, I was just thinking that we won't be able to get pylru into global-requirements until icehouse opens up - there will be a freeze while havana is cut16:52
kgriffsfortunately, simplejson is already in there16:53
acabrerahmm, that's unfortunate...16:56
acabrerawhen does icehouse open up?16:57
acabrera(we might be able to embed pylru if we need to until then, as a temporary workaround)16:57
acabrerakgriffs: ^^16:57
kgriffsnot sure on icehouse… I am assuming late october16:59
acabreramakes sense, given that schedule.17:00
acabreracutting the ice. :)17:00
kgriffsembedding it i think would be an OK short-term workaround17:01
kgriffs(pylru is GPL)17:01
kgriffs"Yes eventlet works well on PyPy, both Marconi and Swift use it."17:03
kgriffswe REALLY need to load-test that17:04
*** vkmc has quit IRC17:04
acabreraoh yes.17:05
acabreraalso, regarding eventlet17:05
acabreraI'm sure I've mentioned it before. :P17:06
kgriffshmm, are they dropping support for 2.6 then?17:08
kgriffsafaik we still need that17:08
acabreranah, they shouldn't need to drop 2.6 support17:10
acabreraultimately, it'll be up to the contributor that implements it, but one can get py26, py27, py32, and py33 all running one one codebase.17:10
kgriffsfyi, just triaged:
openstackgerritKurt Griffiths proposed a change to openstack/marconi: chore: Update Falcon dep to allow version 0.1.7
*** ykaplan has quit IRC17:32
*** vkmc has joined #openstack-marconi17:56
acabreraI won't be making any more changes to the first proxy patches. I'm in agreement with zyuan that we need to start merging these with the caveat that they should not be used yet.18:27
acabrerakgriffs, flaper87|afk: ^^18:27
acabreraHaving a dependency chain that's >2 patches long is getting rather unwieldly.18:28
acabreraIf we pursue this route, I'll create an etherpad of issues that the proxy has as-is.18:29
kgriffslet's do that as a first step regardless18:30
acabrerakgriffs, zyuan: ^^18:34
acabreraI've noted the most prominent existing issues, as well as the patches I believe we should get merged soon.18:43
acabreraI'm tackling the select issues atm.18:47
acabreraclass RoundRobin, gooooo~18:47
maliniGET queues with a bad marker is returning  http 200 with the list of queues. I think we agreed on & used to return 204s in this case. Anybody knows if the 200s are intentional ?19:09
acabrerathat sounds like a bug19:12
zyuanit's not a bug19:12
zyuanbad marker is regarded as non-existing ones19:13
malinizyuan: thts my understanding as well19:13
malinishudnt it be a 204 in tht case?19:14
zyuanyou are not accesing the marker19:14
malinizyuan: can u explain further?19:14
zyuanwait, i need to take a look at the code. this might be changed19:16
maliniI think the code changed too..19:17
zyuanthe code currently behave as if the marker is not sent19:18
zyuanlike, disappeared in paramether19:18
maliniwhat prompted this change?It used to treat it as non existing & return 204.19:19
acabrerahmm... that's some weirdness.19:20
zyuani remembered19:22
zyuanyes it looks weird19:23
zyuanbut the thing it19:23
zyuanit can not be fixed19:23
zyuanbecause there is no way to recognize all invalid markers19:24
zyuanat least in mongo19:24
zyuanso i implemented it in the trivial way:19:24
zyuancompare the marker (None) with others without testing whether it's None or not19:25
zyuanand the result is all others are "larger" then None (in database, not Python, don't worry about python 3)19:25
zyuanan interesting "feature" related to this is:19:26
zyuanyou can use marker='a' (where there is no queue named 'a')19:27
zyuanto list queues starts with 'a'19:27
zyuansry, the marker=None one only applies to list messages19:28
malinizyuan: can you update page to reflect the non-existing markers19:29
kgriffsmalini: can you give an example of a "bad" marker?19:29
*** gordonsim has quit IRC19:29
zyuani just realize that no queue marker is "bad" -_-19:29
malinikgriffs: non existing marker19:29
amettskgriffs:  We get bad markers all the time when people leave the caps off after using the whiteboard.19:30
zyuanametts: -_-|||19:30
kgriffsoh man, that's a shame19:30
kgriffssounds like you guys need some more training19:30
zyuanoh, btw19:30
zyuanbiggest reason:19:30
zyuanqueue can be deleted19:30
zyuanand the marker should still work19:30
kgriffsmalini: so, i think we just do a range comparison19:30
kgriffsi.e., "give me everything with marker > 5'19:31
acabreralol ametts - bad markers...19:31
kgriffsthat being said, we could check to make sure it's an int19:31
zyuankgriffs: you mixed message markers with queue markers too -_-19:31
kgriffsah crap19:32
kgriffsyou're right19:32
zyuanqueue markers are just queue names....19:32
maliniI am cool with the current behavior. We just need to update the docs to reflect the code
kgriffsqueue "marker" is a name, alphabetical19:32
kgriffsbtw, what happens when someone lists messages with a non-integral marker?19:32
kgriffsshould be a 40019:32
malinineed to leave now, will be back online in an hour19:32
zyuandecode -> None19:32
kgriffsok, let me update that page19:32
zyuanid > None?19:33
zyuanthis one should be 204; i mixed queue markers and message markers at begining -_-19:33
*** malini is now known as malini_afk19:35
acabreraI'm almost done preparing the mongo proxy storage driver, from which we can branch and start working on oslo.cache integration.19:35
zyuankgriffs: it's not 400 since the message marker format is opaque; it just happens to be an integral19:35
zyuanametts: O-O19:35
zyuanacabrera: O-O19:36
acabreralol auto-complete...19:36
zyuani want you to immediately change the id back19:36
zyuani @ed my booooss!19:37
*** acabrera is now known as cppcabrera19:39
cppcabrerazyuan: :D19:39
zyuancppcabrera: thanks ^_^19:39
cppcabrerathat should solve the latest set of autocomplete bugs19:39
zyuani realized that i need to create a diamond dependency...19:49
cppcabrerazyuan: ?19:49
zyuanforward test code depends on 3 proxy patch and my 2 test patches19:50
zyuan1 2 319:50
zyuan\a b19:50
cppcabreraall the more reason to get things merged soon.19:51
zyuanmerge them!19:51
cppcabreraI want to run this plan by both kgriffs and flaper87|afk tomorrow morning.19:51
cppcabreraafter that, I think we can start alleviating this patchy headache.19:51
zyuanhere is what i want to achive:19:52
zyuanmerge the existing redis impl and tests19:52
zyuanto ensure them work!19:52
zyuanthen try to refactor and extend them19:52
cppcabreraI like the idea, and I'm in a bit of a tough spot, too, since there's proof of concept cruft I need to clean up.19:57
cppcabreraEspecially relating to how round robin selection works atm.19:57
cppcabreraI'm heading home for the day, though. I'll write an email about the situation before I go.19:57
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): partition related endpoints tested
zyuanretry jenkins20:01
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): catalog tested
zyuanretry jenkins20:05
cppcabreraI'm out for the night. Take care, everyone. :)20:05
*** cppcabrera has quit IRC20:06
openstackgerritZhihao Yuan proposed a change to openstack/marconi: fix(test): shutdown the server in functional test
openstackgerritZhihao Yuan proposed a change to openstack/marconi: test(proxy): catalog tested
*** oz_akan_ has quit IRC20:20
*** oz_akan_ has joined #openstack-marconi20:22
openstackgerritA change was merged to openstack/marconi: Update .gitreview for incubation org move
*** oz_akan_ has quit IRC21:16
*** tedross has quit IRC21:37
*** malini_afk is now known as malini22:01
*** amitgandhi has quit IRC23:04
*** kgriffs is now known as kgriffs_afk23:33
*** amitgandhi has joined #openstack-marconi23:52
*** kgriffs_afk is now known as kgriffs23:54
*** amitgandhi has quit IRC23:56
*** amitgandhi has joined #openstack-marconi23:56

Generated by 2.14.0 by Marius Gedminas - find it at!