| *** amitgandhinz has joined #openstack-zaqar | 00:03 | |
| *** amitgandhinz has quit IRC | 00:07 | |
| *** boris-42 has joined #openstack-zaqar | 00:30 | |
| *** Qiming has joined #openstack-zaqar | 01:29 | |
| *** jasondotstar has quit IRC | 01:30 | |
| *** Qiming_ has joined #openstack-zaqar | 01:30 | |
| *** Qiming has quit IRC | 01:34 | |
| *** Qiming_ is now known as Qiming | 01:35 | |
| *** jasondotstar has joined #openstack-zaqar | 01:59 | |
| *** Yanyanhu has joined #openstack-zaqar | 01:59 | |
| *** jasondotstar has quit IRC | 02:00 | |
| *** jasondotstar has joined #openstack-zaqar | 02:00 | |
| *** jasondotstar has quit IRC | 02:06 | |
| *** jasondotstar has joined #openstack-zaqar | 02:06 | |
| *** AAzza_ has joined #openstack-zaqar | 03:29 | |
| *** AAzza has quit IRC | 03:31 | |
| *** AAzza has joined #openstack-zaqar | 03:33 | |
| *** AAzza_ has quit IRC | 03:34 | |
| *** AAzza_ has joined #openstack-zaqar | 03:34 | |
| *** AAzza- has joined #openstack-zaqar | 03:36 | |
| *** AAzza has quit IRC | 03:37 | |
| *** AAzza- is now known as AAzza | 03:37 | |
| *** AAzza_ has quit IRC | 03:38 | |
| *** jasondotstar has quit IRC | 03:47 | |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar-specs: Add spec for email notification https://review.openstack.org/188662 | 03:59 |
|---|---|---|
| *** achanda has joined #openstack-zaqar | 04:13 | |
| *** achanda has quit IRC | 04:30 | |
| openstackgerrit | Digambar proposed openstack/python-zaqarclient: Implements CLI for pool. https://review.openstack.org/196674 | 04:56 |
| *** AAzza has quit IRC | 05:37 | |
| *** AAzza has joined #openstack-zaqar | 05:40 | |
| *** AAzza_ has joined #openstack-zaqar | 05:43 | |
| *** AAzza has quit IRC | 05:44 | |
| *** AAzza_ is now known as AAzza | 05:44 | |
| *** achanda has joined #openstack-zaqar | 05:48 | |
| openstackgerrit | Digambar proposed openstack/python-zaqarclient: Implements CLI for pool https://review.openstack.org/196674 | 05:58 |
| *** achanda has quit IRC | 06:04 | |
| *** achanda has joined #openstack-zaqar | 06:26 | |
| flaper87 | therve: I forgot I had an appoitment this morning but I'll be back and ping you for our push-notifications chat | 06:37 |
| flaper87 | sorry about that | 06:37 |
| therve | flaper87, Coolio | 06:37 |
| *** boris-42 has quit IRC | 07:02 | |
| *** dguerri` is now known as dguerri | 07:02 | |
| *** achanda has quit IRC | 07:03 | |
| *** dguerri is now known as dguerri` | 07:07 | |
| *** diga_ has quit IRC | 07:49 | |
| openstackgerrit | Christian Schwede proposed openstack/zaqar: Update development-environment.rst guide https://review.openstack.org/196557 | 07:54 |
| *** diga has joined #openstack-zaqar | 08:00 | |
| *** yfujioka has joined #openstack-zaqar | 08:01 | |
| -openstackstatus- NOTICE: OpenStack CI is down due to hard drive failures | 08:50 | |
| *** ChanServ changes topic to "OpenStack CI is down due to hard drive failures" | 08:50 | |
| *** bradjones has quit IRC | 08:55 | |
| *** bradjones has joined #openstack-zaqar | 08:57 | |
| *** bradjones has quit IRC | 08:57 | |
| *** bradjones has joined #openstack-zaqar | 08:57 | |
| *** dguerri` is now known as dguerri | 09:14 | |
| *** Yanyanhu has quit IRC | 10:15 | |
| *** Qiming has quit IRC | 10:15 | |
| *** exploreshaifali has joined #openstack-zaqar | 11:13 | |
| *** Qiming has joined #openstack-zaqar | 11:15 | |
| kragniz | flwang1: rather late pong | 11:37 |
| diga | flaper87: Hi | 11:39 |
| diga | vkmc: hi | 11:39 |
| vkmc | hi diga! | 11:39 |
| diga | can you please review this patch - https://review.openstack.org/#/c/196674/ | 11:40 |
| vkmc | sure, I'll check it out later | 11:44 |
| diga | ok | 11:50 |
| *** c_soukup has quit IRC | 11:55 | |
| *** jyuso1 has quit IRC | 12:07 | |
| *** yfujioka has quit IRC | 12:33 | |
| *** dynarro has joined #openstack-zaqar | 12:42 | |
| *** csoukup has joined #openstack-zaqar | 12:44 | |
| openstackgerrit | Doraly Navarro proposed openstack/zaqar-specs: Replaced 'capabilities' with 'notifications' https://review.openstack.org/196666 | 13:02 |
| exploreshaifali | mooooooorning! :D | 13:05 |
| *** sriram has joined #openstack-zaqar | 13:21 | |
| *** boris-42 has joined #openstack-zaqar | 13:31 | |
| dynarro | exploreshaifali: good morning! | 13:35 |
| exploreshaifali | \o/ | 13:36 |
| *** mpanetta has joined #openstack-zaqar | 13:40 | |
| *** diga has quit IRC | 13:51 | |
| *** malini has joined #openstack-zaqar | 13:51 | |
| *** malini has quit IRC | 14:05 | |
| *** malini has joined #openstack-zaqar | 14:10 | |
| *** amitgandhinz has joined #openstack-zaqar | 14:30 | |
| *** kgriffs|afk is now known as kgriffs | 14:38 | |
| openstackgerrit | Doraly Navarro proposed openstack/python-zaqarclient: wip: Added support to notifications https://review.openstack.org/196652 | 14:39 |
| flaper87 | therve: here when you are, sorry for the delay. We can move it to tomorrow morning if that works best for you | 14:41 |
| flaper87 | mornings are peaceful... normally | 14:41 |
| therve | flaper87, Good now! | 14:41 |
| therve | I don't have that much to talk about really | 14:41 |
| flaper87 | therve: shoot | 14:42 |
| therve | Basically, I was wondering if you had thoughts on making notifications work for websocket | 14:43 |
| therve | As a quick glance, it would mean to me that all nodes are able to talk to each other to be able to propagate messages | 14:43 |
| therve | Because IIUC notifications are managed by the frontend, and not by the storage layer | 14:44 |
| flaper87 | the goal is to get there, for sure. I haven't done much research on that yet but AFAICT, we'll need to add something more to Zaqar to get there. Probably rely on redis pub+sub to notify other nodes or let nodes integrate better with the storage | 14:45 |
| flaper87 | for example | 14:45 |
| therve | Ah, interesting | 14:46 |
| flaper87 | If we're using redis for the data, the redis driver could rely on pub+sub | 14:46 |
| flaper87 | if we're using mongodb, we could rely on tailed cursors + oplog | 14:46 |
| therve | flaper87, Isn't that backend specific though? Could we have a bandaid solution for mongo? | 14:46 |
| therve | OK I don't know about those | 14:46 |
| flaper87 | I'm not sure how that'd work with swift, for example :/ | 14:46 |
| therve | Well we don't have to support all backends for sure | 14:47 |
| flaper87 | right | 14:48 |
| flaper87 | another option is: | 14:48 |
| flaper87 | Have the websocket "push" feature in a different service (instead of zaqar-server it'd be zaqar-notify or zaqar-push). To that service, we can add an endpoint that gets the notifications pushed to | 14:49 |
| flaper87 | That endpoint would be registered in zaqar-server as a notifiyable webhook | 14:49 |
| flaper87 | s/that/those/ | 14:50 |
| flaper87 | not sure how scalable is that, OTOH | 14:50 |
| flaper87 | that is* | 14:50 |
| therve | Yeah that's one possibility too | 14:50 |
| flaper87 | but it's generic and it should be easy to add more nodes | 14:50 |
| therve | It's presumably the easiest solution | 14:51 |
| ryansb | flaper87: yeah, that's not going to be very happy with swift | 14:51 |
| flaper87 | or instead of using HTTP, we could use amqp 1.0, which is cheaper and we can send 1 message from Zaqar and let other services like dispatch distribute it | 14:51 |
| ryansb | may be able to ghetto something together with AMQP | 14:51 |
| flaper87 | I think I like the second option better for now | 14:52 |
| therve | Which was more or less what I thought first | 14:52 |
| flaper87 | awesome :D | 14:52 |
| therve | flaper87, Wait which second option? pubsub? :) | 14:52 |
| flaper87 | mmh, the separate zaqar-notify service. I like the pubsub one except for the fact that redis would become always required and a SPOF for this feature | 14:53 |
| *** 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" | 14:55 | |
| -openstackstatus- NOTICE: The log volume was repaired and brought back online at 14:00 UTC. Log links today from before that time may be missing, and changes should be rechecked if fresh job logs are desired for them. | 14:55 | |
| therve | Okay cool :) | 14:56 |
| therve | I guess it relates to some other question I have, on whether or not we want to support both transport in the same process | 14:57 |
| therve | It's doable, but I suppose we don't | 14:57 |
| therve | If not, we should think about having a different endpoint, especially in keystone, to be discoverable | 14:57 |
| *** exploreshaifali has quit IRC | 15:01 | |
| therve | flaper87, ^^^ | 15:03 |
| flaper87 | mmhh, that's actually a deployers call, I guess. The recommended way to deploy zaqar is by using an external wsgi container. If you have nginx routing http calls to the container and wsgi calls to a zaqar-server instance running websocket, I'd say we can't do much there | 15:05 |
| flaper87 | If you'd ask me, I'd recommend for having them separate, though. | 15:05 |
| flaper87 | and that's what we should probably recommend in the deployment docs | 15:06 |
| *** malini has quit IRC | 15:06 | |
| *** malini has joined #openstack-zaqar | 15:06 | |
| flaper87 | therve: ^ | 15:07 |
| therve | Hum... | 15:10 |
| therve | flaper87, I don't really understand how that would work. Do we have a doc or something? | 15:12 |
| flaper87 | we don't, yet but I can work on one that explains this but it'll really depend on the deployment | 15:13 |
| therve | If you have just one deployment in mind that would be fine :) | 15:14 |
| therve | If nginx is the endpoint, I don't see how it can push http calls on one backend and websocket on another | 15:15 |
| therve | Unless it can make routing decision on connection upgrade? | 15:15 |
| ryansb | it can do path routing though | 15:16 |
| ryansb | part of the API could redirect to /subscription/ws/ID | 15:16 |
| ryansb | so it'd then get all the subscribed messages over that websocket | 15:16 |
| therve | A different path is like a different endpoint | 15:17 |
| *** Qiming has quit IRC | 15:17 | |
| ryansb | What I'm saying is we keep the endpoint http | 15:17 |
| ryansb | then when a websocket-capable client connects, they get pointed by zaqar to the websocket path | 15:18 |
| therve | Right, that would assume nginx can do routing using connection upgrade | 15:18 |
| flaper87 | therve: why do you need connection upgrade? | 15:19 |
| flaper87 | nginx would just act as a proxy to the underlying websocket service | 15:19 |
| therve | flaper87, I mean that's what you do when you're using websocket | 15:19 |
| therve | Sure but it needs to differentiate regular http calls from websocket calls | 15:19 |
| flaper87 | ah nvm, I know what you mean | 15:19 |
| flaper87 | I've seen this done by just adding: "proxy_set_header Connection "upgrade";" | 15:20 |
| flaper87 | in other words, nginx supports it, AFAIR | 15:20 |
| therve | OK fair enough | 15:20 |
| therve | flaper87, So with that in mind, I'm tempted to have a different process for running a websocket zaqar service | 15:24 |
| therve | And not have the transport option in the config file | 15:24 |
| *** bradjones has quit IRC | 15:30 | |
| *** bradjones has joined #openstack-zaqar | 15:32 | |
| *** bradjones has quit IRC | 15:32 | |
| *** bradjones has joined #openstack-zaqar | 15:32 | |
| *** malini has quit IRC | 15:47 | |
| *** dynarro has quit IRC | 15:48 | |
| flaper87 | therve: mmh, I kinda agree. That'll also make it easier to deploy and understand | 15:56 |
| *** achanda has joined #openstack-zaqar | 15:56 | |
| *** malini has joined #openstack-zaqar | 15:58 | |
| *** achanda has quit IRC | 16:00 | |
| *** dontalton has joined #openstack-zaqar | 16:29 | |
| *** achanda has joined #openstack-zaqar | 16:35 | |
| *** openstackgerrit has quit IRC | 16:50 | |
| *** openstackgerrit has joined #openstack-zaqar | 16:51 | |
| *** malini has quit IRC | 16:51 | |
| *** dguerri is now known as dguerri` | 17:25 | |
| *** malini has joined #openstack-zaqar | 17:38 | |
| *** achanda_ has joined #openstack-zaqar | 17:38 | |
| *** achanda has quit IRC | 17:40 | |
| *** achanda has joined #openstack-zaqar | 17:42 | |
| *** achanda_ has quit IRC | 17:44 | |
| *** malini has quit IRC | 17:46 | |
| *** malini has joined #openstack-zaqar | 18:08 | |
| *** achanda has quit IRC | 18:21 | |
| *** davideagnello has quit IRC | 18:39 | |
| *** davideagnello has joined #openstack-zaqar | 18:44 | |
| *** malini has quit IRC | 18:54 | |
| *** malini has joined #openstack-zaqar | 19:06 | |
| *** achanda has joined #openstack-zaqar | 19:10 | |
| *** dguerri` is now known as dguerri | 19:11 | |
| *** dguerri is now known as dguerri` | 19:15 | |
| *** malini has quit IRC | 19:42 | |
| *** AAzza has quit IRC | 19:45 | |
| *** AAzza has joined #openstack-zaqar | 19:45 | |
| *** malini has joined #openstack-zaqar | 19:46 | |
| *** AAzza has quit IRC | 19:50 | |
| *** AAzza has joined #openstack-zaqar | 19:50 | |
| *** malini has quit IRC | 19:51 | |
| *** malini has joined #openstack-zaqar | 19:52 | |
| *** achanda has quit IRC | 19:54 | |
| *** achanda has joined #openstack-zaqar | 19:55 | |
| *** achanda has quit IRC | 19:59 | |
| *** malini has quit IRC | 20:16 | |
| *** malini has joined #openstack-zaqar | 21:09 | |
| *** mpanetta has quit IRC | 21:20 | |
| *** sriram has quit IRC | 21:22 | |
| *** malini has quit IRC | 21:24 | |
| *** fifieldt has quit IRC | 21:39 | |
| *** amitgandhinz has quit IRC | 21:46 | |
| *** fifieldt has joined #openstack-zaqar | 21:51 | |
| *** csoukup has quit IRC | 21:57 | |
| *** davideag_ has joined #openstack-zaqar | 22:09 | |
| *** davideagnello has quit IRC | 22:13 | |
| *** kgriffs is now known as kgriffs|afk | 22:14 | |
| *** dontalton has quit IRC | 22:37 | |
| *** Qiming has joined #openstack-zaqar | 23:39 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!