Wednesday, 2013-10-23

*** reed has quit IRC00:41
*** nosnos has joined #openstack-marconi00:43
*** tedross has joined #openstack-marconi01:15
*** amitgandhi has quit IRC01:26
*** openstack has joined #openstack-marconi01:34
*** amitgandhi has joined #openstack-marconi02:03
*** tedross has quit IRC02:19
*** amitgandhi has quit IRC02:37
*** amitgandhi has joined #openstack-marconi02:40
*** malini is now known as malini_afk02:55
*** openstackgerrit has quit IRC03:16
*** openstackgerrit has joined #openstack-marconi03:17
*** vkmc has quit IRC03:24
*** amitgandhi has quit IRC03:37
*** amitgandhi has joined #openstack-marconi04:07
*** metral has quit IRC04:12
*** amitgandhi has quit IRC04:24
*** metral has joined #openstack-marconi04:38
*** amitgandhi has joined #openstack-marconi04:49
*** amitgandhi has quit IRC05:07
*** amitgandhi has joined #openstack-marconi05:32
*** amitgandhi has quit IRC05:34
*** amitgandhi1 has joined #openstack-marconi05:34
*** amitgandhi1 has quit IRC05:51
*** amitgandhi has joined #openstack-marconi06:16
*** amitgandhi1 has joined #openstack-marconi06:18
*** amitgandhi has quit IRC06:18
*** amitgandhi1 has quit IRC06:46
*** amitgandhi has joined #openstack-marconi07:11
*** amitgandhi1 has joined #openstack-marconi07:14
*** amitgandhi has quit IRC07:14
*** amitgandhi1 has quit IRC07:33
*** amitgandhi has joined #openstack-marconi07:57
*** yassine has joined #openstack-marconi08:13
*** amitgandhi has quit IRC08:17
*** amitgandhi has joined #openstack-marconi08:43
*** amitgandhi has quit IRC08:54
*** nosnos has quit IRC09:12
*** nosnos has joined #openstack-marconi09:13
*** nosnos has quit IRC09:17
*** amitgandhi has joined #openstack-marconi09:49
*** amitgandhi has quit IRC09:54
*** amitgandhi has joined #openstack-marconi10:50
*** amitgandhi has quit IRC10:55
*** amitgandhi has joined #openstack-marconi11:50
*** tedross has joined #openstack-marconi12:09
*** amitgandhi has quit IRC12:25
*** alcabrera has joined #openstack-marconi12:28
alcabreraGood morning!12:28
*** oz_akan_ has joined #openstack-marconi12:29
*** jcru has joined #openstack-marconi12:34
*** vkmc has joined #openstack-marconi12:37
*** amitgandhi has joined #openstack-marconi12:56
*** amitgandhi has quit IRC13:01
*** fvollero|gone is now known as fvollero13:11
*** malini_afk is now known as malini13:17
*** mpanetta has joined #openstack-marconi13:21
mpanettaMorning all13:28
mpanettakgriffs_afk: when you get in, I would like to continue where we left off yesterday.13:28
malinigood morning!!13:28
mpanettamalini: Morning!13:29
mpanettaAnd yes, it will be good :)13:29
maliniof course!!13:30
alcabrerampanetta, malini: Morning. :)13:33
malinihowdy alcabrera?13:35
*** amitgandhi has joined #openstack-marconi13:35
mpanettaMorning alcabrera!13:38
*** ametts has joined #openstack-marconi13:59
mpanettamorning ametts14:00
amettsHowdy.14:00
alcabreraametts: hey! :)14:05
amettsHowdy otra vez14:07
alcabrera¬°gracias!14:08
*** amitgandhi has quit IRC14:16
*** amitgandhi has joined #openstack-marconi14:16
*** jergerber has joined #openstack-marconi14:26
*** tedross has quit IRC14:33
alcabrerazyuan, kgriffs_afk, flaper87|afk, ametts: pre-review completed on the admin API feature line and the catalogue storage driver. Reviews needed for those. Let me know if you have questions. :D14:35
*** whenry has joined #openstack-marconi14:43
*** amitgandhi has quit IRC14:43
*** whenry has quit IRC14:45
*** whenry has joined #openstack-marconi14:46
malinialcabrera: Can you post the links for all the patches ready for review?15:04
alcabreramalini: sure thing15:05
alcabreraHere they come!15:05
alcabrera(in order, too)15:05
alcabrera1. https://review.openstack.org/#/c/50701/15:05
alcabrera2. https://review.openstack.org/#/c/50702/15:05
alcabrera3. https://review.openstack.org/#/c/50721/15:06
alcabrera4. https://review.openstack.org/#/c/50815/15:06
alcabrera5. https://review.openstack.org/#/c/50998/15:06
alcabrera6. https://review.openstack.org/#/c/52160/15:06
alcabreramalini: that's all of them.15:06
malinithx alcabrera15:06
alcabrerathe first five are the admin API feature line, and the last one is the catalogue storage driver.15:06
*** kgriffs_afk is now known as kgriffs15:08
alcabrerakgriffs: o/15:08
malinialcabrera: where do the sharding logs live? dont they need an entry in marconi-queues.conf-sample ?15:09
alcabreramalini: they're logged as part of marconi-queues.conf15:09
alcabrerathe sharding component is a new feature on the core marconi-queues project, so they live together. :)15:09
maliniSo it'll be the same location as marconi server logs ?15:10
alcabrerayup15:10
alcabreragood question, though! Thanks for considering that. Logging wasn't something that came to mind beyond making sure that the transpoprt resource was indicating what was going on.15:11
*** tedross has joined #openstack-marconi15:11
alcabrera**came to my mind...15:11
*** whenry has quit IRC15:15
kgriffshey there. reviewing ur patches now15:16
alcabrerakgriffs: awesome! I'm working on the eom:bastion and responding to everett's questions now. :)15:16
kgriffsrock on15:17
alcabreraWe're getting great feedback. I15:17
alcabreraI'll add his thoughts to the wiki when I'm down.15:17
alcabreraapi.next~15:17
alcabrera*done15:17
malinieverett's feedback caught be my surprise..15:18
maliniI thought we meant to return 404 when queue is not found for GET messages15:18
malinias in here https://wiki.openstack.org/wiki/Marconi/specs/api/v1/errors#Get_Messages15:18
alcabrerahmmm15:19
malinidid we change something to intentionally return 204 on GET message non existing queue?15:19
alcabreranot quite15:19
alcabreraI see what happened15:19
maliniThe only valid case we return 204 on non existing is delete15:19
alcabreraYes, we return 404 when we do GET /v1/queues/not_exists15:19
alcabreraHowever15:19
alcabreraWe return 204 for the case of GET /v1/queues/not_exists/messages15:20
kgriffsoops15:20
kgriffssounds like a bug15:20
malinito me too15:20
alcabreraI think I also recall the rationale for this.15:20
maliniI was abt to open one, but the docs explanation confused me15:20
alcabreraIt hjad something to do with treating that endpoint as a filtering endpoint.15:20
alcabreraEither way15:20
alcabrera404 seems like the right call there15:20
maliniyes..let me open tht bug now15:21
kgriffsi haven't caught up on my email - was that the feedback, that we should return 404 there?15:21
malinikgriffs: yes15:21
kgriffsok15:21
alcabrerayup, that's the gist of it.15:21
kgriffsmalini, have you confirmed the bug yourself?15:21
maliniGET /messages on a non existing queue should return 404. we currently return 20415:21
alcabreraNow that  I've recalled the rationale and understand the situation better, writing this email will be a breeze. :D15:22
maliniyes..I confirmed it15:22
kgriffsok15:22
alcabreraawesome15:22
kgriffsi will give this high priority to fix15:22
kgriffswe are getting behind on bugs, btw15:22
malinikgriffs: I didnt have a test for tht :-$15:22
kgriffsneed to get sharding done!15:22
maliniwill add a test as well15:22
alcabreramalini: thanks!15:22
kgriffsmalini: i'm sure you will remedy that. ;)15:22
alcabreraWhoever writes the fix should also add a unit test for this scenario.15:23
alcabreraI'll volunteer once I roll out the bastion.15:23
alcabreraUnless someone gets to it first. ;D15:23
malinialcabrera: are you responding to Everett?15:25
alcabrerakgriffs, malini: are either of you filing the bug report atm?15:26
alcabreramalini: yes :)15:26
malinihttps://bugs.launchpad.net/marconi/+bug/124375215:26
maliniPlz include the bug in the response15:26
alcabreramalini: Thanks!15:26
alcabreraWas about to, heh. :)15:26
malinithanks!!15:26
zyuankgriffs: there was no such a bug, but flavio wanted to make it 204 due to performance, and we agreed15:28
kgriffsproblem is it is confusing users15:28
kgriffswe need to be careful when balancing implementation details against UC15:29
kgriffss/UC/UX15:29
zyuanfrom an API point of view, i +1 for 40415:29
kgriffslet's try to get it done without sacrificing performance. TBD.15:29
zyuanbut practically i just +0 for 204 instead of -115:29
zyuanlet me take a look15:31
kgriffskk15:31
kgriffsdon't forget to assign yourself to the bug15:31
alcabreraemail sent - back to eom:bastion. :)15:32
kgriffsthx15:36
kgriffsalcabrera: re https://review.openstack.org/#/c/50701/12/marconi/queues/transport/wsgi/admin/driver.py15:40
kgriffsis the health endpoint only exposed in admin mode?15:41
alcabrerakgriffs: no, it's exposed in both modes.15:41
alcabreraIn the final patch, I actually unify this. Now I realize I double-health'd the app.15:41
alcabreralooking for context...15:42
alcabrerakgriffs: https://review.openstack.org/#/c/50998/5/marconi/queues/transport/wsgi/admin/driver.py15:42
mpanettadouble healthed?15:43
alcabrerampanetta: I add the /v1/health route *twice* when running in admin mode.15:43
mpanettaOops15:43
mpanettaAh ok15:43
alcabreraIt's harmless, but silly.15:43
*** amitgandhi has joined #openstack-marconi15:44
kgriffscan you make a note to fix that?15:45
alcabreraWill do. I made in the final patch.15:45
alcabreraI'll make another in the initial patch.15:45
alcabrerakgriffs: noted, thanks!15:46
alcabrera**I made one in...15:47
zyuankgriffs: look at this change c1a564b415:52
*** jergerber has quit IRC15:53
*** mpanetta is now known as mpanetta_afk15:53
zyuanif claim does not return 404, how messages do?15:53
malinizyuan: I was abt to update the bug report - 204s for claims on non existing queues :D16:03
zyuan...16:04
zyuani go lunch and handle leasing stuff; come back no early then 1:3016:05
alcabrerabrb, grabbing lunch and going for a walk.16:06
alcabrerazyuan: enjoy, and good luck!16:06
*** amitgandhi has quit IRC16:06
*** alcabrera is now known as alcabrera|afk16:06
*** amitgandhi has joined #openstack-marconi16:06
*** yassine has quit IRC16:16
*** alcabrera|afk is now known as alcabrera16:22
*** mpanetta_afk is now known as mpanetta16:53
*** tacuddy2000 has joined #openstack-marconi16:57
*** tacuddy2000 has quit IRC16:57
*** tacuddy2000 has joined #openstack-marconi16:58
*** amitgandhi has quit IRC17:19
kgriffsalcabrera: I just finished reviewing this one: https://review.openstack.org/#/c/50702/17:22
alcabrerakgriffs: I just received the email on it!17:22
alcabreraThanks!17:22
alcabreraI'll look it over very shortly. I'm responding to flaper87|afk's email.17:23
*** kgriffs is now known as kgriffs_afk17:31
*** fvollero has quit IRC17:35
*** ametts has quit IRC17:40
*** amitgandhi has joined #openstack-marconi17:41
*** reed has joined #openstack-marconi17:41
*** alcabrera is now known as alcabrera|afk17:45
*** asalkeld has joined #openstack-marconi17:59
asalkeldhi, is there anyway of working around this: https://bugs.launchpad.net/marconi/+bug/124384518:01
*** alcabrera|afk is now known as alcabrera18:01
alcabreraasalkeld: let's see...18:01
alcabrerahmmm18:01
alcabreraThis error is new to me. All of the interesting scenarios. I'm trying to think of a workaround.18:02
asalkelda server shouldn't be doing termio18:02
alcabreraI have one idea.18:02
alcabreraIt's possible to launch marconi using an external WSGI container, say, 'uwsgi' or 'gunicorn'.18:03
alcabreraThat might avoid the termio silliness.18:03
asalkeldmaybe18:03
asalkeldok, I'll look into that, or fixing it18:03
*** kgriffs_afk is now known as kgriffs18:03
alcabrerathanks!18:04
alcabreraI'd appreciate if you updated the ticket with your findings. :)18:04
asalkeldsure18:04
*** whenry has joined #openstack-marconi18:08
alcabrerakgriffs: responded on that review - greatly appreciated! :D18:11
*** amitgandhi has quit IRC18:21
openstackgerritAngus Salkeld proposed a change to openstack/marconi: Don't enable_echo when we don't have a tty  https://review.openstack.org/5341518:28
*** asalkeld is now known as asalkeld_eating18:29
alcabrerakgriffs: can we get a quick review on this: https://review.openstack.org/#/c/53415/ (Closes a recently discovered bug, +5 LOC)18:31
*** tacuddy2000 has quit IRC18:33
zyuankgriffs: ping18:39
*** alcabrera has quit IRC18:46
*** amitgandhi has joined #openstack-marconi18:51
*** whenry has quit IRC18:59
*** alcabrera has joined #openstack-marconi19:00
alcabrerakgriffs: I like messages.Item + messages.List19:01
alcabreraThanks for the suggestions!19:02
*** JRow has joined #openstack-marconi19:06
*** reed has quit IRC19:14
*** asalkeld_eating is now known as asalkeld19:16
*** JRow has left #openstack-marconi19:18
*** malini is now known as malini_afk19:32
kgriffskk19:35
zyuankgriffs: have you saw the Claim change?  so is the 204 when listing messages when there is no queue considered as a bug?19:36
zyuan(i noticed that ralational db can also benefit from this "shortcut")19:37
*** asalkeld has quit IRC19:37
kgriffsit is considered a bug, based on user feedback19:38
kgriffs404 was expected19:38
zyuanbut, if change this, many places need to be changed as well19:39
zyuando it all over the places?19:39
openstackgerritA change was merged to openstack/marconi: Don't enable_echo when we don't have a tty  https://review.openstack.org/5341519:39
* kgriffs is looking19:39
zyuan(we can't have something partially just to address a user feedback)19:39
zyuan(all or nothing)19:39
zyuanpartially done*19:40
alcabrerait'll require some changes to how we handle the storage aspect.19:40
alcabreracurrently, we just issue queries over the message collection.19:41
alcabreraw/o regard to the queue collection.19:41
alcabrerathat's why we don't check for existence.19:41
*** asalkeld has joined #openstack-marconi19:41
alcabreraso there's an overlap in the case that a queue has no messages and a queue just doesn't exist.19:41
openstackgerritDennyZhang proposed a change to openstack/marconi: update quickstart document for marconi  https://review.openstack.org/5343519:43
kgriffsso, now that we have oslo.cache, what about having a queue exists check in the storage driver?19:44
zyuanbut this overlap does not seem to be harmful19:44
zyuanit's already there19:45
kgriffsi mean, modifying it to use cache, and then calling it from message listing19:45
alcabrerakgriffs: hmmm, good point. I hadn't considered leveraging oslo.cache.19:45
zyuanbut... you know, more checking, higher chance of race condition in other storage19:45
zyuanwhile the result from 1 query is more "correct"19:46
kgriffsyou would have to purge the cache when queue is deleted, but I thing it would be OK, practically speaking, if once in a while someone gets a 204 when they should have gotten 40419:47
zyuanit's rare, because queue are not exptected to be deleted often19:47
*** whenry has joined #openstack-marconi19:48
alcabreraI'm +1 for the cache idea. I also believe the race condition is benign.19:50
kgriffsdo we want to base this fix on alcabrera's patches since those set up cache in the bootstrap?19:51
kgriffs(iirc)19:51
kgriffsspeaking of which, does the cache backend just default to in-memory?19:52
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: add shard management resource  https://review.openstack.org/5070219:54
alcabrerakgriffs: yes, it defaults to in-memory.19:54
kgriffsok19:54
kgriffsi would hate give up our super easy "kick-the-tires" install, e.g.,19:54
alcabreraso that'd be a per-process queues cache - which is actually a problem.19:55
alcabreraUnless we added TTLs.19:55
kgriffspip install; marconi-server19:55
kgriffsis it really a problem? Real deployments would use memcached or something anyway, right?19:56
kgriffsbbl - got a meeting19:56
alcabrerakgriffs: well, it's more of a gotcha, really. :P19:56
alcabrerak19:56
openstackgerritDennyZhang proposed a change to openstack/marconi: update quickstart document for marconi  https://review.openstack.org/5343519:57
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: shards storage controller interface  https://review.openstack.org/5072119:59
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: shards mongodb driver + tests  https://review.openstack.org/5081519:59
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: integrate shard storage with transport  https://review.openstack.org/5099820:03
*** whenry has quit IRC20:06
*** amitgandhi has quit IRC20:08
*** amitgandhi has joined #openstack-marconi20:08
*** kgriffs is now known as kgriffs_afk20:09
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: integrate shard storage with transport  https://review.openstack.org/5099820:10
openstackgerritAlejandro Cabrera proposed a change to openstack/marconi: feat: add catalogue storage driver for queues  https://review.openstack.org/5216020:11
alcabrerarebasing complete20:11
alcabreraI'm headed home. I'll continue work on eom:bastion tomorrow.20:11
zyuansee you20:11
alcabrerazyuan: :D20:11
*** alcabrera has quit IRC20:12
*** amitgandhi has quit IRC20:22
*** amitgandhi has joined #openstack-marconi20:27
*** asalkeld has quit IRC20:28
*** asalkeld has joined #openstack-marconi20:28
*** reed has joined #openstack-marconi20:39
*** kgriffs_afk is now known as kgriffs20:42
*** asalkeld has quit IRC21:08
*** asalkeld has joined #openstack-marconi21:08
*** jergerber has joined #openstack-marconi21:15
openstackgerritZhihao Yuan proposed a change to openstack/marconi: feat(health): make health API storage-aware  https://review.openstack.org/5345721:26
*** alcabrera has joined #openstack-marconi21:38
*** whenry has joined #openstack-marconi21:42
*** amitgandhi has quit IRC21:46
*** oz_akan_ has quit IRC21:52
*** amitgandhi has joined #openstack-marconi21:52
*** amitgandhi has quit IRC21:57
*** tedross has quit IRC21:58
*** kgriffs is now known as kgriffs_afk22:09
*** alcabrera has quit IRC22:14
*** mpanetta has quit IRC22:15
*** whenry has quit IRC22:16
asalkeldis the marconi client functional?22:35
asalkeldand where is the cli?22:36
asalkeldI asked this months ago, I hope there has been progress22:36
*** openstack has joined #openstack-marconi22:47
*** asalkeld is now known as asalkeld_lunch23:20
*** jergerber has quit IRC23:23
*** jcru has quit IRC23:32
*** whenry has joined #openstack-marconi23:57

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