Monday, 2013-10-07

*** nosnos has joined #openstack-marconi01:00
*** malini_afk is now known as malini01:53
*** malini is now known as malini_afk01:55
*** malini_afk is now known as malini01:56
*** amitgandhi has joined #openstack-marconi02:41
*** amitgandhi has quit IRC02:50
*** malini is now known as malini_afk03:36
*** nosnos has quit IRC05:36
*** nosnos_ has joined #openstack-marconi05:36
*** fifieldt_ has joined #openstack-marconi05:53
*** dafter has quit IRC06:26
*** dafter has joined #openstack-marconi06:26
*** dafter has quit IRC06:30
*** notmyname has quit IRC06:58
*** flaper87|afk is now known as flaper8706:58
*** notmyname has joined #openstack-marconi06:59
*** nosnos has joined #openstack-marconi07:02
*** nosnos_ has quit IRC07:02
*** nosnos has quit IRC07:19
*** nosnos has joined #openstack-marconi07:20
*** ykaplan has joined #openstack-marconi07:46
*** flaper87 is now known as flaper87|afk08:03
*** yassine has joined #openstack-marconi08:12
*** ykaplan has quit IRC08:21
*** fifieldt_ has quit IRC08:26
*** flaper87|afk is now known as flaper8708:27
*** ykaplan has joined #openstack-marconi08:27
*** fifieldt has joined #openstack-marconi08:41
*** ykaplan has quit IRC08:59
openstackgerritFlavio Percoco proposed a change to openstack/marconi: Use oslo.config directly instead of common.config  https://review.openstack.org/4955009:01
*** ykaplan has joined #openstack-marconi09:05
openstackgerritA change was merged to openstack/marconi: chore: Designate semi-private "friend" helper methods as such  https://review.openstack.org/4981209:07
openstackgerritFlavio Percoco proposed a change to openstack/python-marconiclient: Implement HTTP Transport  https://review.openstack.org/4978809:13
openstackgerritFlavio Percoco proposed a change to openstack/python-marconiclient: Make the request object API aware  https://review.openstack.org/4978709:13
openstackgerritFlavio Percoco proposed a change to openstack/python-marconiclient: Split get_transport into 2 different functions  https://review.openstack.org/4978609:13
*** reed has joined #openstack-marconi09:15
*** flaper87 is now known as flaper87|afk10:31
*** ykaplan has quit IRC11:13
*** ykaplan has joined #openstack-marconi11:28
*** flaper87|afk is now known as flaper8711:32
*** tedross has joined #openstack-marconi11:38
*** fvollero|gone is now known as fvollero12:26
*** flaper87 is now known as flaper87|afk12:30
*** flaper87|afk is now known as flaper8712:36
*** flaper87 is now known as flaper87|afk12:43
*** nosnos has quit IRC12:47
*** rustlebee is now known as russellb12:54
*** mpanetta has joined #openstack-marconi12:54
*** flaper87|afk is now known as flaper8712:57
*** fifieldt has quit IRC13:06
*** oz_akan_ has joined #openstack-marconi13:08
*** alcabrera has joined #openstack-marconi13:16
*** amitgandhi has joined #openstack-marconi13:16
alcabreraGood morning!13:19
flaper87alcabrera: HEYYYY!!!13:20
flaper87good morning13:20
flaper87hope you had an amazing w/e13:20
alcabreraI was pretty awesome, flaper87. :D13:21
alcabreraSpent a lot of time with my wife and house mate playing Terraria 1.2.13:21
alcabrera*It13:22
flaper87alcabrera: hehehe, cool!!!13:24
alcabreraI noticed you made several updates to rust-http. ;)13:26
flaper87alcabrera: hahaha, yeah!! :D13:34
flaper87alcabrera: didn't know you were following it13:34
flaper87:P13:34
flaper87I'm also struggling with the implementation of Select traits for the Timer13:34
flaper87basically, I hack on other projects while rust compiles :P13:35
flaper87and tests run13:35
*** ykaplan has quit IRC13:36
*** malini_afk is now known as malini13:38
alcabreraflaper87: reminds me of my c++ days. "Oh, you're compiling?" /me goes off to read tech articles13:38
flaper87:P13:40
flaper87yeah, basically that but with code13:40
flaper87:D13:40
alcabrera:)13:41
alcabreraI took a moment to clean up marconi-redis Sunday morning. I got rid of *a lot* of unnecessary cruft, so it's just the queues-storage impl. and the test stub.13:41
alcabreraI've also added marconi as a req using the -f <URL#egg> style.13:41
flaper87alcabrera: AWESOME, so, you could basically do: pip install marconi-redis-backend13:43
flaper87and that will install marconi as well, then tweak the config and then: $ marconi-server13:43
flaper87alcabrera: btw, I'd like to get the config patch merged ASAP13:45
flaper87I'm afraid that patch will conflict a lot13:45
flaper87with new patches13:46
alcabreraI'm looking into it. :)13:46
flaper87alcabrera: AWESOME!13:47
flaper87thanks a lot13:47
alcabreraI'm still 90% zombie this morning, haha.13:47
flaper87ahahahha13:47
flaper87need a coffee + nutella + 1kg of sugar13:47
alcabreraflaper87: "Use oslo.config directly instead of common.config"?13:47
flaper87and add a redbull13:47
flaper87alcabrera: yeah13:47
flaper87that one13:47
alcabreranooooo, that would kill me. :P13:47
alcabreraI've been avoiding sugars and starches lately - makes me feel more energized. :D13:48
* alcabrera reviews13:49
flaper87alcabrera: LOL, sorry, I'll take sugars for ya then!13:50
flaper87:S13:50
flaper87:D13:51
*** ykaplan has joined #openstack-marconi13:51
alcabreralol go ahead. :)13:52
* alcabrera is reminded of cocosettes13:52
alcabreraWhich in turn reminds me of...13:52
alcabrerafvollero: Good morning!13:52
alcabreraflaper87: reviewed - +2'd13:55
flaper87alcabrera: thank you sir, I now need kgriffs_afk bless13:56
alcabreramarconi/marconi-{queues, proxy}.conf and lots of oslo.config usage examples - lgtm.13:56
* flaper87 is having SO much fun with the client13:56
flaper87it's nice to see how things start fitting together13:57
flaper87alcabrera: next step is to rename marconi-server entry_point into marconi-queues13:57
flaper87so we can get rid of that prog='marconi-queues' keyword13:57
flaper87then, I'll get rid of the global instance13:57
*** jcru has joined #openstack-marconi13:59
*** dafter has joined #openstack-marconi14:01
*** flaper87 is now known as flaper87|afk14:05
fvolleroalcabrera: Morning to you :)14:07
*** dafter has quit IRC14:15
*** ayoung has joined #openstack-marconi14:18
*** kgriffs_afk is now known as kgriffs14:32
kgriffsgood afternoon/morning14:33
kgriffso/14:34
alcabrerakgriffs: o/14:34
kgriffsI have a feeling today is going to be a busy day14:34
alcabreravery much so.14:34
kgriffsneed some eyes on this14:41
kgriffshttps://review.openstack.org/#/c/49378/614:41
kgriffslast breaking change14:41
malinikgriffs: looks like that patch cud use some functional tests14:43
maliniwant me to add one?14:43
kgriffssure thing14:43
kgriffsi want to get this done today so we can lock down the API (finally)14:43
zyuanmalini: you only need to add the testing with wrong client_id14:44
zyuanmalini: because your functional tests starts with UUID, which works out of box14:44
maliniyeap..I'll add a test that will use client-id = 1234 & expect a 400 back14:44
*** ametts has joined #openstack-marconi14:44
*** alcabrera is now known as alcabrera|afk14:45
kgriffsametts: could you help review? https://review.openstack.org/#/c/49378/614:47
amettsok14:49
*** reed has quit IRC14:51
kgriffsthanks!14:52
*** flaper87|afk is now known as flaper8714:58
flaper87MAAAAAAAAAAAAAAAAAAAAAAAAAAAAN, I'm alive14:58
flaper87T_T14:58
flaper87I got blocked out my VPN for a while14:58
flaper87I had to take panic attack pills14:58
flaper87and stuff14:58
flaper87phew14:58
flaper87so good to be on-line again14:58
flaper87kgriffs: just to make that feeling true, could you help me getting https://review.openstack.org/#/c/49550/ in ?14:59
flaper87ping world ?15:02
* ametts hears everyone sing "We are the world..." in response to flaper87's ping15:10
flaper87ametts: LOOOL15:12
kgriffsflaper87: sure15:15
flaper87kgriffs: I'm afraid that patch is conflict-prone15:15
flaper87so, I'd like to see it landing asap15:16
*** alcabrera|afk has quit IRC15:21
*** briancline has quit IRC15:23
zyuanflaper87: can we reach it after done proxy? we are deploying.15:24
*** alcabrera has joined #openstack-marconi15:25
flaper87zyuan: that patch shouldn't break existing patches - not 100% sure - I'm more concern about new patches than existing ones.15:26
*** briancline has joined #openstack-marconi15:26
flaper87hope that makes sense15:26
zyuanflaper87: i wish it do not break existing conf files.  we are deploying, you know...15:27
flaper87zyuan: oooh gotcha. It doesn't, it uses the same config files and options as before15:28
flaper87marconi-proxy.conf and marconi-queues.conf15:28
flaper87no changes to the configuration params either15:28
amettszyuan: In https://review.openstack.org/#/c/49378/6 I see changes to transport and sqlite driver.  No changes are needed to mongo driver?  Just making sure....15:33
zyuanametts: mongo works out-of-box. it recognize UUID15:33
amettsOk - thx.15:33
flaper87meeting in 10mins15:48
flaper87kgriffs: zyuan alcabrera malini megan_w ametts amitgandhi ^15:48
kgriffsyep15:52
kgriffsI totally forgot to send out the email until just now15:52
* kgriffs ducks15:52
fvolleroHola fellas15:56
fvolleroflaper87: marconi meeting ? can i join as well ?15:56
kgriffseveryone's welcome!15:56
kgriffs#openstack-meeting-alt15:56
* fvollero hug kgriffs 15:56
flaper87fvollero: no, get the fuck out of here15:57
* flaper87 hides15:57
flaper87fvollero: you're always welcome, d00d15:57
* kgriffs dispatches sniper15:57
fvolleroflaper87: :P15:57
*** yassine has quit IRC15:57
flaper87kgriffs: thing is that fvollero has something I want, something I wish I had here, something that comes from my land!!!!15:57
flaper87and I'm hating him so bad because of that15:58
flaper87:P15:58
kgriffswow15:58
fvollerofellas, i'm working on the elasticsearch storage backend and a problem just pop out, how to garantee the FIFO on this storage, suggestion ?15:58
fvolleroflaper87: Fuck man, i'll send it through Ceska Posta!15:58
kgriffsmtg16:02
kgriffs#openstack-meeting-alt16:02
kgriffssee you there!16:02
*** tedross has quit IRC16:04
* alcabrera made it back from a deep dive debug session16:05
amitgandhidid you find any treasures?16:06
amitgandhioic =P16:07
alcabreraamitgandhi: yeah.16:07
alcabreranginx doesn't accept headers containing '_' by default.16:07
amitgandhistrange16:07
alcabreraso... two discoveries.16:07
alcabrera1) That^^16:07
alcabrera2) There's a bug in the proxy.16:07
*** flaper87 is now known as flaper87|afk16:07
alcabreraIt doesn't cannonicalize headers before sending them to marconi-queues.16:08
alcabreraI have a patch in the works.16:08
amitgandhiok16:08
amitgandhithats where the missing projectid issue came into play?16:08
amitgandhior clientid issue16:08
*** flaper87|afk is now known as flaper8716:09
alcabreraboth16:10
alcabreraflaper87: wb!16:10
alcabreraamitgandhi: ^16:10
flaper87ping ?16:10
flaper87fvollero: now we're talking :D16:10
flaper87fvollero: re ElasticSearch, there are no auto-increment there16:10
*** jarret has joined #openstack-marconi16:10
flaper87fvollero: and the id generated is just a hash16:10
flaper87fvollero: there's a trick you could use that basically increments the _version of the document16:10
flaper87but I'm not very happy with the way that trick works16:10
fvolleroflaper87: Uhm me neither16:11
flaper87fvollero: btw, something really cool about ES is that it has a built-in ttl support per document16:11
fvolleroflaper87: exactly16:11
flaper87fvollero: I think you can take advantage of the built-in timestamp property16:11
flaper87of the document16:11
flaper87(for FIFO, I mean)16:11
fvolleroflaper87: i was thinking about this as well, so that instead to set the ttl value and evict manually.16:11
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: bug: proxy headers aren't forwarded properly  https://review.openstack.org/5012916:12
fvolleroflaper87: yeah, I thought about it, but later we need to do a query to the recent elements based on timestamp16:12
alcabreraflaper87, kgriffs: bug fix, tiny review. ^^ :)16:13
kgriffskk16:15
openstackgerritA change was merged to openstack/python-marconiclient: Create v1 package and move messages there  https://review.openstack.org/4866416:16
flaper87fvollero: yeah, but that should be harmless, AFAIk, you can sort on that field16:16
*** tedross has joined #openstack-marconi16:17
*** ykaplan has quit IRC16:34
oz_akan_alcabrera: uwsgi --fastrouter 127.0.0.1:3017 --fastrouter-use-code-string 0:mapper.py:get16:48
oz_akan_uwsgi supports a py to decide where a request should be routed.16:49
oz_akan_and it accepts a parameter16:49
fvolleroflaper87: ok, going in that direction16:49
oz_akan_http://uwsgi-docs.readthedocs.org/en/latest/Fastrouter.html16:50
oz_akan_alcabrera: ^^16:50
alcabreraoz_akan_: I remembered something about that, but it was too limiting for the general case.16:50
alcabreraoz_akan_: Not everyone will deploy a marconi-proxy using uwsgi.16:50
*** acabrera has joined #openstack-marconi16:51
flaper87fvollero: double check the sort on timestamp think16:51
flaper87I'm not 100000000% sure16:51
oz_akan_alcabrera: not everything has to be modular :)16:52
oz_akan_for sure I see our implementation of proxy is more capable16:53
*** alcabrera has quit IRC16:54
*** acabrera is now known as alcabrera16:55
amitgandhialcabrera needs some reviews : https://review.openstack.org/50129  : kgriffs, zyuan, flaper8717:05
kgriffshttps://wiki.openstack.org/wiki/Marconi/Incubation/Graduation#To_Do17:05
kgriffsUpdated17:06
* flaper87 clicks on all those links17:06
flaper87I'll check those after dinner!17:06
flaper87brb17:06
*** tedross has quit IRC17:07
amettsDid alcabrera really give himself a  -1 on that patch?!17:08
fvolleroflaper87: http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/mapping-timestamp-field.html17:08
flaper87ametts: hahaha17:09
flaper87ametts: I've -2 mines17:09
alcabreralol17:09
flaper87:P17:09
amettsAh.  He made a typo.  Now I just look stupid for giving him a +1.17:09
alcabreraametts: yeah, sometimes I catch issues on my own patches after stepping away.17:09
flaper87fvollero: there you go17:09
flaper87and it's not_analyzed17:09
alcabreraAfter working on the proxy, I've learned to do little things to make reviews easier for everyone. :D17:09
flaper87btw, remeber to make all the fields not_analyzed17:10
flaper87you don't need that for marconi17:10
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: bug: proxy headers aren't forwarded properly  https://review.openstack.org/5012917:10
alcabreraflaper87, kgriffs: fixed ^^17:10
alcabreraametts: thanks for the review. :)17:11
ametts+1'd again.  Although I think you should lose points for inventing the world "canonicalize".... :)17:12
amettss/world/word/17:12
alcabreraametts: lol, fair enough. I've seen it around, but it's not acknowledged by any dictionary. :P17:13
*** tvb has joined #openstack-marconi17:14
*** tvb has quit IRC17:14
*** tvb has joined #openstack-marconi17:14
*** tvb is now known as dafter17:15
*** flaper87 is now known as flaper87|afk17:19
kgriffsalcabrera: didn't we add a .headers to falcon that does that for you?17:21
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: bug: proxy headers aren't forwarded properly  https://review.openstack.org/5012917:23
*** tedross has joined #openstack-marconi17:23
kgriffsactually, that would have been in 0.1.717:23
kgriffsdid that every make it into global reqs?17:23
alcabrerakgriffs: not yet. :/17:29
alcabreraI was just putting together a patch for openstack/requirements17:30
alcabrerato bump falcon up a version17:30
alcabrerakgriffs: zyuan just told me about ,headers, too. I've added a TODO on this patch to address that asap.17:30
alcabrerakgriffs, zyuan: https://review.openstack.org/5014017:32
alcabreraLet's see if I can get that in now, since the Havana code freeze is over.17:32
kgriffsok17:32
alcabreraIf I can, I'll update the bug fix patch accordingly.17:33
kgriffslet's allow this bug fix patch to land for now17:33
kgriffsjust make a note to update later17:33
alcabreraah, alright.17:33
kgriffsbtw, yours is a little faster since it uses list comprehension, not a generator17:33
kgriffsI am fixing that in falcon as we speak. :D17:33
alcabrerakgriffs: https://bugs.launchpad.net/marconi/+bug/123648417:34
alcabreraSweet. :)17:34
kgriffsthanks!17:35
fvollerouhmmm17:41
fvollerohow marconi 'assign' the ttl to the messages ?17:41
*** dafter has quit IRC17:42
alcabrerafvollero: When posting messages, the user specifies the TTL to use. Does that help?17:42
*** dafter has joined #openstack-marconi17:42
*** dafter has quit IRC17:42
*** dafter has joined #openstack-marconi17:42
fvolleroalcabrera: nop :)17:42
fvolleroalcabrera: i'm wondering in which format: day,seconds, ms, minutes,weeks ?17:43
alcabreraooohh17:43
alcabreraseconds17:43
fvolleroalcabrera: brilliant.17:43
fvolleroalcabrera: it's taking shape the elastic search storage engine :)17:43
alcabrerafvollero: awesome! :D17:43
alcabreramake it a real thing.17:44
kgriffsflaper87|afk - https://review.openstack.org/#/c/50129/17:44
*** dafter has quit IRC17:46
kgriffszyuan: https://review.openstack.org/#/c/49378/17:46
kgriffsone tiny change17:47
*** dafter has joined #openstack-marconi17:48
*** dafter has quit IRC17:48
*** dafter has joined #openstack-marconi17:48
*** tvb|afk has joined #openstack-marconi17:49
*** tvb|afk has quit IRC17:49
*** tvb|afk has joined #openstack-marconi17:49
*** dafter has quit IRC17:53
*** tvb|afk has quit IRC17:53
zyuankgriffs: i see. wsgi/utils has no function directly talks to falcon objects yet.  ok?17:54
zyuani'm fine with that.17:54
kgriffsthat's fine17:55
zyuanok17:55
*** dafter has joined #openstack-marconi17:55
kgriffsYou can pretty much cut-paste it over there17:55
kgriffsalcabrera: Blacklisting partition hosts that timeout? Does request raise an exception?17:57
kgriffson a socket timeout?17:58
alcabrerakgriffs: No blacklisting yet.17:58
alcabrerakgriffs: also, requests will raise an exception if the socket times out.17:59
kgriffsok, is there a bp somewhere for that?17:59
openstackgerritA change was merged to openstack/marconi: bug: proxy headers aren't forwarded properly  https://review.openstack.org/5012917:59
kgriffsand are you handling the timeout exception?17:59
alcabrerakgriffs: there are no bug reports or blueprints for either one, and I'm handling neither.17:59
kgriffsoh oh17:59
*** tvb|afk has joined #openstack-marconi17:59
alcabrerayup17:59
kgriffsI'd say, add a bp for blacklisting (temporary) on timeout to queues17:59
alcabrerawill do18:00
kgriffsand then ASAP add a try..except around those forwarded requests18:00
kgriffs(can return 500 to the client, I guess)18:00
kgriffsor...18:00
alcabreramatches my thoughts - HTTP 50018:00
kgriffsjust try another host18:00
alcabrerathat seems reasonable, too18:01
kgriffsmakes us look better if we "just work" ;D18:01
alcabreraI'll write up that bp, and then take a look at what I can make happen asap. :)18:01
kgriffscool, thanks!18:01
*** dafter has quit IRC18:01
kgriffsdid you see your patch got merged?18:02
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): Client-ID is now a real UUID  https://review.openstack.org/4937818:02
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): give validation resp a different title  https://review.openstack.org/4983718:02
kgriffsI made an executive decision18:02
*** kgriffs is now known as kgriffs_afk18:03
alcabrerakgriffs_afk: thanks!18:04
alcabrerakgriffs_afk: https://blueprints.launchpad.net/marconi/+spec/proxy-blacklist-nonresponsive-hosts18:05
*** dafter has joined #openstack-marconi18:09
*** tvb|afk has quit IRC18:12
*** vkmc has joined #openstack-marconi18:17
*** vkmc has quit IRC18:17
*** vkmc has joined #openstack-marconi18:17
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): Client-ID is now a real UUID  https://review.openstack.org/4937818:18
zyuannow fixed ^^18:18
amitgandhizyuan, kgriffs_afk: meeting18:31
*** kgriffs_afk is now known as kgriffs18:33
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): Client-ID as a real UUID  https://review.openstack.org/4937818:36
zyuankgriffs: alcabrera ^^18:41
*** flaper87|afk is now known as flaper8718:46
* flaper87 back18:49
flaper87I'll review the pending patches now18:49
flaper87sorry for the delay guys18:49
alcabreraflaper87: no worries, and thanks!18:51
*** zzs has joined #openstack-marconi18:52
flaper87zyuan: +2'd both18:58
flaper87kgriffs: can I get your bless here? https://review.openstack.org/#/c/49550/19:00
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): give validation resp a different title  https://review.openstack.org/4983719:01
zyuanflaper87: thanks19:01
flaper87zyuan: I think you just broke the dependency chain with that last commit19:02
flaper87:/19:02
*** oz_akan_ has quit IRC19:04
flaper87the last patch won't get merged unless the outdated commit it's depending on gets merged19:04
kgriffsflaper87: sure, I'll take a look at the oslo patch19:04
kgriffsoslo.config, I mean19:04
flaper87kgriffs: thank you :D19:05
zyuanflaper87: yes. but i assume jekins can try merge... anyway, let me rebase.19:05
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(api): Client-ID as a real UUID  https://review.openstack.org/4937819:06
flaper87zyuan: it wont :(19:06
flaper87if it depends on a patch, it doesn't even bother to try19:06
flaper87zyuan: I'll +2 both again19:07
flaper87thanks19:07
flaper87ah no need for the last one19:07
alcabreraI'll be back later guys. :)19:08
*** alcabrera has quit IRC19:08
kgriffszyuan: I think this is OK, but later I would like errors to be more specific, to help out the client developer19:10
kgriffshttps://review.openstack.org/#/c/49837/4/marconi/queues/transport/wsgi/exceptions.py19:10
kgriffsI mean, it should say "Invalid header value" and then in the description state *which* header is invalid19:10
kgriffsor "Invalid query parameter"19:10
kgriffsetc.19:10
kgriffsI filed a bug just so we don't forget to revisit19:11
kgriffshttps://bugs.launchpad.net/marconi/+bug/123654019:11
zyuankgriffs: input validation is 1 layer, and the layer is higher than low level things like header or body19:11
malinidid we have something change around queues marker recently?19:11
zyuanthat's my understanding...19:11
kgriffsmalini: I can't think of anything specific.19:12
zyuankgriffs: the error message body raise from input validation has those information19:12
kgriffsor is this going through the proxy?19:12
kgriffszyuan: gtk. I just want to take some time later to find everywhere we raise a 400 and make sure the response is consistent19:14
malinikgriffs: this is not thru proxy..I am doing a list queues with GET /v1/queues?marker=ZZZ, it is returning me queues with names starting with d etc. ( marker is the same as queue name, rt? )19:15
kgriffsyeah19:15
kgriffsis the same19:15
kgriffssounds like a bug to me19:16
kgriffsyou would expect 204, right?19:16
malinikgriffs: yeap..This is my pdb o/p http://paste.openstack.org/show/48038/19:17
openstackgerritA change was merged to openstack/marconi: feat(api): give validation resp a different title  https://review.openstack.org/4983719:17
malinikgriffs: I'll open a new bug19:18
*** dafter has quit IRC19:18
kgriffsok19:18
kgriffsI will work on it19:18
openstackgerritA change was merged to openstack/marconi: feat(api): Client-ID as a real UUID  https://review.openstack.org/4937819:20
zyuankgriffs: i think we discussed that19:20
zyuanit's not a bug...19:20
zyuanqueue marker is explicit and may not exist19:21
zyuanthe listing behaves like listing with prefix but it's not the purpose and can not be avoided19:21
zyuanthe purpose is that if you delete a queue from the middle, the listing still works.19:22
zyuanmalini: i hope you are not too surprised ^^19:22
malinizyuan: yeap we discussed this & the decision was https://wiki.openstack.org/wiki/Marconi/specs/api/v1/responsecodes#List_Queues19:24
malini'A marker always "exists" in the sense that we can always perform a range comparison with it. In other words, if you pass 'g' for the marker, Marconi will return all queue's whose names sort after 'g', lexicographically. If you pass a marker that sorts after all existing queues, you will receive "204 No Content" in response.'19:24
maliniI was expecting a 204 for the marker=ZZZZ, but ended up getting queues with names starting d etc.19:25
kgriffshah19:27
kgriffsnice catch malini19:27
kgriffsactually, something *did* change recently there19:28
kgriffslooks like a regression caused by joining the "project" and "queue" fields19:28
kgriffs(in the DB)19:28
kgriffsI have a repro19:28
malinicool..one of our functional tests was failing19:28
kgriffsnice19:28
malinitest_get_queue_malformed_marker, if u want a repro19:29
kgriffsI added a test for that to the unit tests as well, to make sure it doesn't sneak by my again. :p19:29
malinigtg now..will chk back later19:29
kgriffsttfn19:29
*** malini is now known as malini_afk19:30
*** dafter has joined #openstack-marconi19:35
*** dafter has quit IRC19:35
*** dafter has joined #openstack-marconi19:35
*** vkmc_ has joined #openstack-marconi20:00
*** vkmc_ has quit IRC20:13
*** vkmc has quit IRC20:14
*** vkmc has joined #openstack-marconi20:15
*** vkmc has quit IRC20:15
*** vkmc has joined #openstack-marconi20:15
*** oz_akan_ has joined #openstack-marconi20:43
*** oz_akan_ has joined #openstack-marconi20:44
*** flaper87 has quit IRC20:48
*** flaper87 has joined #openstack-marconi20:52
*** flaper87 is now known as flaper87|afk21:02
*** malini_afk is now known as malini21:02
kgriffsmalini: o/21:12
malini\o21:13
kgriffshey, so about that bug21:15
kgriffstuns out that Z sorts before 'q'21:16
kgriffsso, 'q' is greater than 'Z21:16
maliniwhat abt 'z' ?21:17
kgriffslowercase z is fine21:17
malinihmm..I was getting 200s with 'z' as well :(21:18
kgriffsuppercase sorts that way because in the character table, uppercase has lower integers than lowercase21:18
kgriffsmalini: yes, I would expect that when there is more than one project ID21:18
kgriffsis that the case?21:18
kgriffsbecause....21:18
kgriffsthat is another bug I discovered21:18
maliniI am getting a different set of queues with lower case 'z'21:19
maliniBut this time I get queues starting with 'U'21:20
kgriffsif I'm right, then those other queues actually belong to a different project than the one you asked for21:20
kgriffsI know for a fact there is a bug there. Whether there are others, is TBD21:21
malinithis is the response with ?marker=zzz , http://paste.openstack.org/raw/48051/21:21
maliniit is probable that these are for a different project, since it looks like the Unit Tests create these q's21:22
kgriffsok21:24
kgriffslet me finish this patch and then you can see if it fixes the problem21:24
maliniok21:24
ekarlsoaloha :p21:24
kgriffsmalini: btw, I found some big holes in the unit test for listing queues while doing this.21:25
kgriffsekarlso: guten Tag!21:26
malinikgriffs: that is concerning :( we have coverage in functional tests, but we probably need to revisit all the API's if we are missing anything else21:26
kgriffswell, there wasn't a test anywhere detecting returning queues from the wrong project, for example21:27
kgriffs(I added it to the unit test)21:27
kgriffs(just now)21:27
kgriffsAlso, the unit test was still attempting to "PUT" metadata when creating a queue21:27
kgriffsand not really checking that metadata was being returned in listings21:27
malinithat must be a cruft from the old design..21:28
kgriffsyep21:28
malinithe functional tests do not validate the project id part either..will add that to my todo21:28
kgriffssounds good21:28
maliniGuess we'll need the project-id validation for other parts of our api as well21:29
*** tedross has quit IRC21:33
*** notmyname has quit IRC21:34
*** zzs has quit IRC21:34
*** openstackgerrit has quit IRC21:34
*** notmyname has joined #openstack-marconi21:35
*** zzs has joined #openstack-marconi21:35
kgriffsmalini: ok, I am doing a final test on my fix21:35
kgriffswill be submitting shortly21:35
kgriffsdid you ever create a bug for this?21:36
malinino..I will21:36
malinido it now21:36
kgriffsok21:36
malinior wud you rather do it? since this is diff from the original one?21:36
kgriffsbasically, the bug turned out to be that we would return queues from other projects instead of stopping at the end21:37
kgriffsi can do it21:37
kgriffsstand by21:37
kgriffshttps://bugs.launchpad.net/marconi/+bug/123660521:39
malinikgriffs: does this potentially afect other listings as well ? like messages?21:41
amitgandhido we have security tests, to ensure we cant get other user queues/messages21:41
kgriffsnope21:41
kgriffsmessage listing markers are independent of project ID21:42
kgriffsI mean, we are already explicitly checking project ID when listing messages21:42
kgriffsgood thought, tho!21:43
kgriffsbugs do tend to cluster21:43
maliniamitgandhi: we dont have security tests..but we need those & Ihave it in my todo list now21:45
amitgandhicool21:45
*** vkmc has quit IRC21:47
kgriffsmalini: https://review.openstack.org/#/c/50176/21:51
*** kgriffs is now known as kgriffs_afk22:01
amitgandhikgriffs: have you ever run into this issue installing marconi on mac osx22:14
amitgandhi Unknown distribution option: 'setup_requires'22:15
amitgandhiUnknown distribution option: 'pbr'22:15
amitgandhii've been getting that lately when i try to run python setup.py develop22:15
amitgandhi on a fresh install of marconi in a venv22:15
amitgandhihmm interesting22:19
amitgandhiafter pip install d2to1 i get a diff error now for a bunch of other dists22:19
mpanettaOh man...22:23
mpanettaI think alej knows that one.22:23
mpanettadtry pip install --upgrade instead22:23
mpanetta*try22:23
mpanettaI think that is what I ended up doing.22:23
mpanettaamitgandhi: ^^22:24
*** jcru has quit IRC22:24
amitgandhiwhat did you upgrade?22:25
amitgandhiive tried pip install -U setuptools22:25
mpanettaIt may have been d2to122:25
mpanettait was either that or pbr, I had issues with both.22:26
amitgandhiwhen i installed d2to1 it got me around the pbr issue22:26
amitgandhibut now it doesnst know about entry_points, install_requires, etc22:26
mpanettatry reinstalling it with upgrade option.22:27
amitgandhii did22:27
amitgandhi=/22:27
amitgandhiafter forcing an upgrade of setuptools...22:29
amitgandhi…im back to the pbr issue22:29
mpanettaOh wonderful...22:29
mpanettaCd to your marconi code and do pip install -U -r requirements.txt22:30
mpanettaThat is another workaround we use at times...22:30
amitgandhi…is installing a bunch of stuff22:31
amitgandhicool22:32
amitgandhii think that worked22:32
amitgandhiyou rock =)22:32
mpanettayeah.  Probably every single requirement.22:32
mpanettaNah, I just listen :P22:32
mpanettaAnd ocasionally remember lol22:32
amitgandhithanks for your help22:32
mpanettaNo problem22:32
*** amitgandhi has quit IRC22:33
*** mpanetta has quit IRC22:34
*** openstackgerrit has joined #openstack-marconi23:01
*** dafter has quit IRC23:05
*** vkmc has joined #openstack-marconi23:07
*** fifieldt has joined #openstack-marconi23:26
*** tedross has joined #openstack-marconi23:39
*** amitgandhi has joined #openstack-marconi23:42
*** zzs has left #openstack-marconi23:44
*** amitgandhi1 has joined #openstack-marconi23:47
*** amitgandhi has quit IRC23:47

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