| openstackgerrit | wangxiyuan proposed openstack/zaqar: Let v2 tempest base on the base tests. https://review.openstack.org/324161 | 01:10 |
|---|---|---|
| flwang | Eva-i: ping | 02:33 |
| flwang | Kevin_Zheng: ping | 02:53 |
| Kevin_Zheng | hi | 02:54 |
| Kevin_Zheng | flwang: pong | 02:59 |
| flwang | Kevin_Zheng: are you still working on the message/claims api ref? | 03:00 |
| Kevin_Zheng | Sorry, I was busy on 2 nova BPs | 03:01 |
| Kevin_Zheng | the deadline is pretty close in nova | 03:01 |
| Kevin_Zheng | for non-priorities | 03:02 |
| Kevin_Zheng | I will work on it after this week, | 03:02 |
| flwang | Kevin_Zheng: ok, then how about i take over the 'claim' part? | 03:02 |
| Kevin_Zheng | Ok | 03:02 |
| flwang | you can still work on the 'message' part | 03:02 |
| Kevin_Zheng | Sure | 03:02 |
| flwang | Kevin_Zheng: cool | 03:02 |
| -openstackstatus- NOTICE: static.openstack.org (which hosts logs.openstack.org) is currently migrating due to a hardware failure. It should be back up shortly. | 03:11 | |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Add Claims for api ref https://review.openstack.org/331516 | 04:43 |
| *** flwang1 has quit IRC | 05:04 | |
| -openstackstatus- NOTICE: static.openstack.org (which hosts logs.openstack.org and tarballs.openstack.org among others) is currently being rebuilt. As jobs can not upload logs they are failing with POST_FAILURE. This should be resolved soon. Please do not recheck until then. | 05:23 | |
| *** ChanServ changes topic to "static.openstack.org (which hosts logs.openstack.org and tarballs.openstack.org among others) is currently being rebuilt. As jobs can not upload logs they are failing with POST_FAILURE. This should be resolved soon. Please do not recheck until then." | 05:23 | |
| *** rcernin has joined #openstack-zaqar | 06:13 | |
| *** pcaruana has joined #openstack-zaqar | 06:24 | |
| *** jtomasek has joined #openstack-zaqar | 07:11 | |
| *** Ravikiran_K has joined #openstack-zaqar | 07:54 | |
| *** wxy123 has joined #openstack-zaqar | 07:54 | |
| *** AAzza has quit IRC | 08:35 | |
| *** AAzza has joined #openstack-zaqar | 08:35 | |
| *** flwang1 has joined #openstack-zaqar | 08:42 | |
| *** flwang1 has quit IRC | 08:53 | |
| *** wxy123 has quit IRC | 09:08 | |
| *** wxy123 has joined #openstack-zaqar | 09:08 | |
| *** flwang1 has joined #openstack-zaqar | 09:13 | |
| *** shakamunyi has quit IRC | 09:48 | |
| *** shakamunyi has joined #openstack-zaqar | 09:56 | |
| *** rcernin is now known as rcernin|lunch | 10:37 | |
| *** ChanServ changes topic to "OpenStack Queuing and Notification Service || Smile :D || Meetings every Monday @ 21:00 UTC/15:00 UTC || Wiki: https://wiki.openstack.org/wiki/Zaqar || Paste: http://paste.openstack.org/ || Send messages and make some noise :D" | 10:40 | |
| -openstackstatus- NOTICE: static.openstack.org is back up. If you have POST_FAILURE and are missing logs from your CI jobs, please leave a 'recheck'. | 10:40 | |
| *** Ravikiran_K has quit IRC | 10:58 | |
| *** rcernin|lunch is now known as rcernin | 11:53 | |
| *** ryansb has quit IRC | 12:20 | |
| *** ryansb has joined #openstack-zaqar | 12:20 | |
| *** ryansb has quit IRC | 12:20 | |
| *** ryansb has joined #openstack-zaqar | 12:20 | |
| *** mdnadeem has quit IRC | 12:28 | |
| *** mdnadeem has joined #openstack-zaqar | 12:30 | |
| *** Ravikiran_K has joined #openstack-zaqar | 12:45 | |
| -openstackstatus- NOTICE: OpenID login from review.o.o is experiencing difficulties, possibly due to transatlantic network performance issues. Things are being investigated | 13:00 | |
| *** ChanServ changes topic to "OpenID login from review.o.o is experiencing difficulties, possibly due to transatlantic network performance issues. Things are being investigated" | 13:00 | |
| therve | Hum, zaqar-server got broken | 13:07 |
| therve | Fails with ArgsAlreadyParsedError | 13:07 |
| therve | https://review.openstack.org/#/c/327268/ presumably | 13:08 |
| *** ChanServ changes topic to "OpenStack Queuing and Notification Service || Smile :D || Meetings every Monday @ 21:00 UTC/15:00 UTC || Wiki: https://wiki.openstack.org/wiki/Zaqar || Paste: http://paste.openstack.org/ || Send messages and make some noise :D" | 13:14 | |
| -openstackstatus- NOTICE: OpenID logins are back to normal | 13:14 | |
| openstackgerrit | Thomas Herve proposed openstack/zaqar: Fix config parsing in zaqar-server https://review.openstack.org/331668 | 13:21 |
| therve | flwang, flaper87, vkmc, Eva-i ^^^ please have a look | 13:22 |
| therve | I thought we had a test for the websocket service | 13:22 |
| therve | Ah it's the integration tests we don't run anywhere | 13:24 |
| therve | Sigh | 13:24 |
| flaper87 | therve: +2 | 13:24 |
| *** mdnadeem_ has joined #openstack-zaqar | 13:30 | |
| *** mdnadeem has quit IRC | 13:33 | |
| *** eggmaster has joined #openstack-zaqar | 13:51 | |
| openstackgerrit | Thomas Herve proposed openstack/zaqar: Remove leftovers on websocket disconnections https://review.openstack.org/326897 | 14:25 |
| *** wxy has quit IRC | 14:30 | |
| *** srwilkers has joined #openstack-zaqar | 14:38 | |
| *** srwilkers has quit IRC | 14:53 | |
| *** srwilkers has joined #openstack-zaqar | 15:03 | |
| *** adu has joined #openstack-zaqar | 15:16 | |
| *** srwilkers has quit IRC | 15:18 | |
| *** srwilkers has joined #openstack-zaqar | 15:19 | |
| *** srwilkers has quit IRC | 15:33 | |
| *** srwilkers has joined #openstack-zaqar | 15:36 | |
| *** pcaruana has quit IRC | 15:45 | |
| *** Ravikiran_K has quit IRC | 15:47 | |
| *** rcernin has quit IRC | 15:55 | |
| *** srwilkers has quit IRC | 15:56 | |
| *** adu has quit IRC | 16:24 | |
| *** rcernin has joined #openstack-zaqar | 16:53 | |
| *** rcernin has quit IRC | 17:18 | |
| *** srwilkers has joined #openstack-zaqar | 17:26 | |
| openstackgerrit | Merged openstack/zaqar-ui: Ability to add and remove subscriptions https://review.openstack.org/311852 | 17:27 |
| *** tqtran has joined #openstack-zaqar | 17:29 | |
| *** srwilkers has quit IRC | 17:37 | |
| *** ametts has joined #openstack-zaqar | 17:39 | |
| flwang1 | therve: ping | 17:45 |
| flwang1 | reminder: meeting in 15 mins | 17:45 |
| therve | flwang1, Hey | 17:50 |
| flwang1 | therve: i'm looking at this patch https://review.openstack.org/331668 | 17:50 |
| flwang1 | but i don't really understand what's the problem and why it can pass all the tests | 17:51 |
| therve | Because we don't start zaqar-server in the tests? | 17:51 |
| flwang1 | therve: even in the tempest test? | 17:52 |
| *** rcernin has joined #openstack-zaqar | 17:52 | |
| therve | Well we start it but don't rely on it | 17:52 |
| therve | flwang1, http://logs.openstack.org/97/326897/3/check/gate-tempest-dsvm-mongodb-zaqar/0ec42cf/logs/screen-zaqar-websocket.txt.gz is the error | 17:52 |
| therve | The tempest tests probably use devstack so uwsgi | 17:53 |
| therve | We would need to run the integration tests somewhere, but we don't | 17:53 |
| therve | (which is probably on me as I introduced them) | 17:53 |
| flwang1 | therve: ah, i see. | 17:53 |
| flwang1 | we're using uwsgi to start app directly | 17:54 |
| therve | Yeah. But websocket needs zaqar-server | 17:55 |
| flwang1 | yep, i see. so it only impact websocket | 17:55 |
| flwang1 | and in tempest, we're not testing websocket :( | 17:56 |
| therve | "only" !!!!! | 17:56 |
| therve | :D | 17:56 |
| flwang1 | sorry, i shouldn't use 'only' :D | 17:56 |
| therve | Well it impacts everything that uses zaqar-server, which only ought to be websocket | 17:56 |
| therve | Basically the patch merged was completely broken, which is unfortunate | 17:57 |
| flwang1 | therve: that's my fault | 17:58 |
| therve | Only partly :) | 17:58 |
| flwang1 | one thing weird is tripleo is using websocket | 17:59 |
| flwang1 | how can the author submit pass the test locally? | 18:00 |
| flwang1 | meeting, ttyl | 18:00 |
| therve | I don't think the author run tripleo tests | 18:01 |
| therve | And I don't know if tripleo has websocket tests areadly, it probably doesn't | 18:01 |
| flwang1 | therve: tripleo is using zaqar's websocket, so i assume he did test before posting | 18:04 |
| flwang1 | anyhow | 18:04 |
| *** jtomasek has quit IRC | 18:09 | |
| * flwang1 feel lonely since only me in the openstack-meeting-3 channel :( | 18:09 | |
| therve | Heh | 18:15 |
| flwang1 | i have cancelled the meeting so i'm free for any chat :D | 18:17 |
| *** shakamunyi has quit IRC | 18:17 | |
| openstackgerrit | Merged openstack/zaqar: Fix config parsing in zaqar-server https://review.openstack.org/331668 | 18:24 |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Add Claims for api ref https://review.openstack.org/331516 | 18:26 |
| *** srwilkers has joined #openstack-zaqar | 18:37 | |
| flwang1 | eggmaster: ping | 18:45 |
| *** srwilkers has quit IRC | 19:16 | |
| eggmaster | flwang1: pong | 19:34 |
| *** flwang1 has quit IRC | 19:39 | |
| *** rcernin has quit IRC | 20:31 | |
| *** Eva-i has quit IRC | 20:41 | |
| flwang | therve: back? | 21:13 |
| *** flwang1 has joined #openstack-zaqar | 21:22 | |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Fix debug log missing issue https://review.openstack.org/331854 | 21:40 |
| flwang | therve: https://review.openstack.org/331854 | 21:45 |
| therve | flwang, Yeah | 21:57 |
| flwang | therve: see above patch | 21:57 |
| flwang | that's another regression issue after we moving the cli option | 21:57 |
| flwang | all the debug log have gone | 21:57 |
| therve | flwang, So moving the log.setup after the conf parsing seems to work too | 21:59 |
| therve | It may be a better idea to keep it contained in the the server cmd | 22:00 |
| flwang | you mean put it at line 49 https://review.openstack.org/#/c/331854/1/zaqar/cmd/server.py ? | 22:00 |
| flwang | it should work, i think | 22:00 |
| flwang | let me give it atry | 22:00 |
| therve | Yep I mean that | 22:01 |
| flwang | yep, it works, i will upload a new patchset | 22:02 |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar: Fix debug log missing issue https://review.openstack.org/331854 | 22:02 |
| flwang | therve: would you mind +1ing it? | 22:02 |
| therve | Done | 22:02 |
| flwang | therve: cool. btw, are you interested in proposing a breakout session about the aodh + zaqar + mistral + heat + senlin? | 22:03 |
| therve | I don't know | 22:04 |
| therve | What's a breakout session | 22:04 |
| flwang | a session on openstack summit | 22:04 |
| flwang | https://www.openstack.org/summit/barcelona-2016/call-for-presentations/ | 22:05 |
| therve | Not really | 22:07 |
| flwang | therve: ok, no problem | 22:07 |
| flwang | therve: feel free let me know if you need help for your patch https://review.openstack.org/#/c/329520/ | 22:09 |
| therve | I'm okay I think, just need a good amount of tests and cleanups | 22:10 |
| flwang | therve: cool | 22:10 |
| flwang | for now, when aodh trigger the webhook of heat, does heat need a token? | 22:11 |
| therve | aodh has a trust webhook as well | 22:11 |
| therve | So yeah it gets a token | 22:12 |
| flwang | therve: ok, i see. so can you see any benefit if we use zaqar between aodh and heat for autoscaling? | 22:13 |
| flwang | and should we still prompt heat's autoscaling if you guys are planing to move it out? | 22:14 |
| therve | I don't think we plan to move it out | 22:14 |
| therve | We still need some kind of trigger | 22:14 |
| flwang | 'move out' maybe not a good word | 22:15 |
| therve | Having zaqar between aodh and heat could have some benefit if we rewrote some of the logic to take into account message that we lose | 22:15 |
| therve | BTW zaqar support in aodh is totally dumb right now | 22:15 |
| therve | I'm not sure we want to advocate it | 22:16 |
| flwang | last time, when i asked this topic with stevebaker and Zane bitter. I was told auto scaling wont' be the focus on heat | 22:16 |
| therve | Righ | 22:17 |
| therve | Right | 22:17 |
| therve | Moving it out is another topic | 22:17 |
| stevebaker | well, we'd rather delegate to a dedicated autoscaling service | 22:17 |
| flwang | i see. forgive my bad english | 22:17 |
| flwang | therve: what do you mean 'dumb'? | 22:18 |
| flwang | therve: do you mean it has bug so totally doesn't work? | 22:18 |
| therve | flwang, It uses the service user | 22:18 |
| therve | So the queue is just not accessible | 22:18 |
| therve | stevebaker, Yeah but we probably need some knobs in Heat anyway | 22:19 |
| flwang | therve: ah, yes. but IIRC, it's using correct tenant, i will take a look | 22:19 |
| flwang | therve: stevebaker: IMHO, zaqar could be a good broker for the auto scaling in openstack | 22:20 |
| therve | AFAICT it's not | 22:20 |
| flwang | therve: ok, thanks for the heads up, i will take a look at that | 22:20 |
| flwang | what i want to figure out is what's the correct direction for the auto scaling in openstack | 22:21 |
| flwang | so that i can know how can zaqar/i help | 22:21 |
| flwang | can you guys give me some ideas? | 22:21 |
| therve | I'd say talk to the Senlin guys | 22:22 |
| therve | I tried working on autoscaling for a year and barely manage to do anything, so it's somewhat beyond me :) | 22:23 |
| flwang | therve: senlin is still using a hook which even doesn't need a token | 22:24 |
| therve | They use ec2 signature? | 22:24 |
| flwang | use a confidential UUID | 22:25 |
| therve | Meh that's fine | 22:26 |
| flwang | ok | 22:27 |
| *** ametts has quit IRC | 22:27 | |
| flwang | therve: i just reviewed the aodh zaqar notifier code again | 22:31 |
| flwang | and i can remember now | 22:31 |
| flwang | yep, it's using service tenant and user, but it's creating queues for different tenants | 22:31 |
| therve | What do you mean? | 22:33 |
| flwang | ok, it's not accurate | 22:34 |
| flwang | it's creating queues with alarm-id + topic | 22:34 |
| flwang | and i understand your concern | 22:36 |
| flwang | from the end user's view, they can't create an alarm and subscribe the queue for mistral | 22:37 |
| flwang | since they don't know the queue name | 22:37 |
| therve | flwang, They can't access the queue anyway? | 22:37 |
| therve | Even if they knew the name it's not in their project | 22:37 |
| flwang | yes | 22:37 |
| flwang | if we can get the project name/id for the alarm, i think we can create the correct client with the project info | 22:38 |
| flwang | the author discussed with me and they prefer to use this way | 22:39 |
| flwang | it works for current alarms distribution | 22:39 |
| flwang | but it's not enough for the aodh -> zaqar -> mistral | 22:39 |
| flwang | is that your concern? | 22:40 |
| therve | It seems that it uses the subscription mechanism to work around that | 22:40 |
| therve | It may actually be broken because it uses strange configuration names | 22:41 |
| therve | I wonder if anyone ever tested it :) | 22:41 |
| flwang | so technically, if we can create an alarm in aodh with mistral url as the subscriber, then it should work i think | 22:43 |
| therve | Probably not because you can't specify the post body | 22:43 |
| flwang | the only thing is it doesn't support the post body | 22:43 |
| flwang | yes | 22:43 |
| flwang | that's what i'm talking :) | 22:44 |
| flwang | i can propose a fix for that and test it in one shot :) | 22:44 |
| flwang | which could be a follow up patch along with your patch | 22:45 |
| therve | You can't do that in aodh though | 22:46 |
| therve | What would be better, IMO, is 1) Don't generate the queue name just post straight to the given queue | 22:47 |
| therve | 2) Use signed url so that we post to the correct project | 22:47 |
| therve | 3) Remove that subscription stuff which is too limited anyway | 22:47 |
| therve | If the queue is not dynamic, you can now create a proper subscription with a post body with my patch | 22:48 |
| flwang | therve: but the problem is they(aodh) don't want to change any existing code | 22:52 |
| therve | Don't they? If we can tell them that it doesn't work that could convince them :) | 22:53 |
| flwang | 1) when you say "don't geneate queue name", did you mean passing a queue name when creating alarm? | 22:54 |
| therve | Yeah instead of using the topic as postfix, use it directly | 22:55 |
| flwang | that could be a pre-signed url you mentioned in 2), is it? | 22:55 |
| flwang | when you say remove the 'subscription stuff', do you mean user have to create subscriptions in zaqar manually? | 22:56 |
| therve | Yes | 22:56 |
| flwang | hmm... | 22:57 |
| flwang | if so, the alarm creation is not completed from the aodh's view | 22:58 |
| flwang | am i missing something? | 22:58 |
| therve | Why? | 22:58 |
| flwang | that said, after created the alarm, user's targets can't be notified | 22:59 |
| flwang | unless user create the subscriptions in zaqar manually | 22:59 |
| flwang | maybe my understanding is wrong, not an aodh expert | 23:00 |
| therve | I don't see why that's a problem to create the subscription | 23:00 |
| therve | You don't even have to do it after creating the alarm, you can do it before if you want to remove the race condition | 23:00 |
| flwang | when creating alarm, user will pass in the webhook/email/whatever they want to be notified/triggered, right? | 23:01 |
| therve | No they will pass the zaqar queue | 23:01 |
| flwang | ok, i know that works | 23:01 |
| flwang | my point is | 23:01 |
| flwang | is it consistent with current alarms behaviour? | 23:01 |
| therve | Why not? | 23:02 |
| therve | OK got to go | 23:06 |
| flwang | ok, thanks therve, ttyl | 23:08 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!