| *** jeffrey4l has quit IRC | 00:16 | |
| *** jeffrey4l has joined #openstack-zaqar | 00:26 | |
| *** flwang has quit IRC | 00:33 | |
| *** cpallares has quit IRC | 00:41 | |
| *** flwang has joined #openstack-zaqar | 00:52 | |
| *** jeffrey4l has quit IRC | 01:40 | |
| *** jeffrey4l has joined #openstack-zaqar | 01:53 | |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/python-zaqarclient: CLI support for message CRUD operations https://review.openstack.org/120966 | 01:54 |
|---|---|---|
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/python-zaqarclient: CLI support for claim CRUD operations https://review.openstack.org/122611 | 02:40 |
| *** vkmc has quit IRC | 02:41 | |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Adds encapsulation control to the client https://review.openstack.org/129988 | 03:18 |
| *** flwang has quit IRC | 03:37 | |
| *** flwang has joined #openstack-zaqar | 03:53 | |
| *** flwang has quit IRC | 04:26 | |
| *** sgotliv has joined #openstack-zaqar | 05:02 | |
| *** reed has quit IRC | 05:33 | |
| *** flwang has joined #openstack-zaqar | 05:46 | |
| *** sgotliv has quit IRC | 05:55 | |
| *** flwang has quit IRC | 05:55 | |
| *** sgotliv has joined #openstack-zaqar | 06:08 | |
| *** yfujioka_ is now known as yfujioka | 06:27 | |
| *** yfujioka has quit IRC | 07:28 | |
| *** yfujioka has joined #openstack-zaqar | 07:41 | |
| *** exploreshaifali has joined #openstack-zaqar | 08:10 | |
| *** X019 has joined #openstack-zaqar | 08:46 | |
| *** X019 has quit IRC | 08:52 | |
| *** flwang has joined #openstack-zaqar | 08:58 | |
| flwang | flaper87: ping | 08:59 |
| flaper87 | flwang: pong | 09:00 |
| flwang | do you have a few minutes to discuss the notification service? | 09:00 |
| flaper87 | flwang: yes sir | 09:04 |
| flaper87 | flwang: well, actually | 09:04 |
| flaper87 | flwang: just if you promise you'll be at the summit | 09:04 |
| flaper87 | :P | 09:04 |
| flwang | flaper87: haha, I will be there if you promise a beer :) | 09:05 |
| flaper87 | flwang: Sold! | 09:05 |
| flwang | as for the protocol, any priority in your mind? | 09:05 |
| flwang | I mean which one is the first we should focus? webhook, email, etc ? | 09:05 |
| flaper87 | flwang: I'd say email | 09:06 |
| flaper87 | erm | 09:06 |
| flaper87 | sorry | 09:06 |
| flaper87 | I meant, webhook | 09:06 |
| flwang | reason? | 09:07 |
| flwang | i was thinking email | 09:09 |
| flwang | so I would like to get some insight about why webhook | 09:09 |
| *** jeffrey4l has quit IRC | 09:51 | |
| flaper87 | flwang: ops, just saw this | 10:05 |
| flaper87 | flwang: you should really mention me | 10:05 |
| flaper87 | :P | 10:05 |
| flaper87 | otherwise I'll just forget about conversations | 10:05 |
| flaper87 | flwang: the reason is that it's useful for more projects right now | 10:06 |
| * flwang is sleeping. Zzzzz.... | 10:06 | |
| flaper87 | I think getting emails done after the webhooks are done would be really easy | 10:06 |
| flaper87 | flwang: wake up | 10:06 |
| flaper87 | >.> | 10:06 |
| flwang | flaper87: IMUC, a simple webhook is most like a URL which can be triggered with a POST action, is it? | 10:07 |
| flwang | s/IMUC/IIUC | 10:08 |
| *** jeffrey4l has joined #openstack-zaqar | 10:08 | |
| *** malini has joined #openstack-zaqar | 10:18 | |
| *** malini has quit IRC | 10:18 | |
| *** malini has joined #openstack-zaqar | 10:19 | |
| *** sgotliv has quit IRC | 10:22 | |
| flaper87 | flwang: right | 10:30 |
| flaper87 | flwang: that's the simpliest way to describe it | 10:30 |
| flaper87 | the implementation is not that easy, though | 10:31 |
| flwang | flaper87: got, I'm reading the api doc of SNS to get some inspirations | 10:31 |
| flaper87 | flwang: awesome, did you read the spec I wrote ? | 10:37 |
| flaper87 | flwang: let me know what you think | 10:37 |
| flaper87 | I'd really really like to get this done in Juno | 10:37 |
| flaper87 | flwang: how much time do you think you'll be able to put on this? | 10:37 |
| flwang | flaper87: I have read that | 10:40 |
| flwang | I will put my comments on that in/after the summit :) | 10:40 |
| flwang | did you mean Kilo? | 10:40 |
| flwang | I think that's my focus in Kilo among all my community work | 10:41 |
| flaper87 | flwang: awesome, good to know | 10:41 |
| flwang | I think I can put 15-20 hours per week, if that's you're asking for | 10:41 |
| flwang | you know, there is not too much interesting stuff in Glance, IMHO | 10:42 |
| flwang | so I need some fun at here :) | 10:42 |
| flaper87 | flwang: looooooooooooool, I'll save this log for later :P | 10:44 |
| flaper87 | flwang: 20h per week sounds great | 10:44 |
| flaper87 | I'm looking forward to help with as much as I can if you are willing to take the lead on that work | 10:44 |
| flwang | don't tell the secret to those glance guys :) | 10:45 |
| flwang | flaper87: i'm happy to take it and I will cry aloud at the channel if I need some help | 10:46 |
| flaper87 | flwang: there are things missing in that spec. For example, how are we going to call the subscribed webhooks concurrently? | 10:48 |
| flaper87 | flwang: I dropped a comment about this in the review | 10:50 |
| flwang | it may deserve a dedicated daemon service to do that | 10:51 |
| flwang | taskflow is most like a workflow management lib, I'm not really sure if it's good at parallel running something. | 10:53 |
| flwang | will talk with Joshua H to get some insight | 10:53 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/zaqar: Document mongodb's driver deployments https://review.openstack.org/125938 | 10:59 |
| flaper87 | flwang: I'm not sure about adding extra workers | 11:01 |
| flaper87 | to have separate workers you need to rely on a messaging service to talk to them | 11:01 |
| flaper87 | that'd be weird | 11:01 |
| flaper87 | with taskflow we could use the parallel executor which should be based on eventlet | 11:02 |
| flaper87 | after all, talking to the subscribers will be all about IO | 11:02 |
| *** flwang has quit IRC | 11:02 | |
| *** flwang has joined #openstack-zaqar | 11:03 | |
| *** boris-42 has quit IRC | 11:11 | |
| *** vkmc has joined #openstack-zaqar | 11:22 | |
| vkmc | morning! | 11:27 |
| exploreshaifali | Good morning vkmc | 11:29 |
| exploreshaifali | please can you give any thoughts for bug https://bugs.launchpad.net/zaqar/+bug/1373464 | 11:30 |
| vkmc | hey exploreshaifali! | 11:32 |
| vkmc | that is a complex one | 11:32 |
| exploreshaifali | vkmc: yeah, but there is no other bug which is easy yet unassigned | 11:33 |
| vkmc | we use a function to set some configs at runtime | 11:33 |
| vkmc | I see | 11:33 |
| vkmc | have you checked the client? | 11:33 |
| exploreshaifali | vkmc: client? | 11:34 |
| vkmc | https://bugs.launchpad.net/python-zaqarclient/ | 11:36 |
| exploreshaifali | vkmc okay, let me see there too :) Thanks! | 11:36 |
| vkmc | np | 11:39 |
| *** exploreshaifali has quit IRC | 11:45 | |
| *** X019 has joined #openstack-zaqar | 11:47 | |
| vkmc | flaper87, hey d000d | 11:47 |
| vkmc | did you see the websockets-amqp bindings I shared yesterday? | 11:48 |
| vkmc | wanna know your comments on that | 11:48 |
| *** X019 has quit IRC | 11:58 | |
| *** sgotliv has joined #openstack-zaqar | 12:00 | |
| *** X019 has joined #openstack-zaqar | 12:00 | |
| flaper87 | o/ | 12:07 |
| flaper87 | GOOOOOOOOOOOOOOOOD MORNING | 12:07 |
| flaper87 | vkmc: ah yeah, I saw it | 12:07 |
| flaper87 | I actually had seen it before and then completely forgot | 12:07 |
| flaper87 | that'd be *very* interesting to explore | 12:08 |
| flaper87 | vkmc: great finding | 12:08 |
| flaper87 | too bad it's still a draft, though. | 12:08 |
| vkmc | flaper87, yeah I saw it yesterday night and it was like finding a gummybear | 12:08 |
| vkmc | yeah, its fairly new | 12:08 |
| vkmc | but there is some work on that direction | 12:08 |
| vkmc | and it would make things easier for us! | 12:08 |
| flaper87 | vkmc: I'm asking the experts as we speak | 12:09 |
| flaper87 | lets see if I can get more info about it | 12:09 |
| vkmc | flaper87, you rock! | 12:09 |
| vkmc | it would be cool to have more info about it before the persistent transport session | 12:10 |
| flaper87 | vkmc: indeed | 12:10 |
| vkmc | so we can give some extra points to adding a websocket driver :D | 12:10 |
| flaper87 | somehow, I keep missing exploreshaifali lately | 12:10 |
| vkmc | flaper87, ok you know you rock... but you can be more humble about it | 12:10 |
| flaper87 | vkmc: LOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOL | 12:11 |
| flaper87 | ROFL | 12:11 |
| vkmc | 'you rock' -> 'indeed' | 12:11 |
| vkmc | lol | 12:11 |
| flaper87 | That was obviously meant for your last message | 12:11 |
| vkmc | I know, I'm messing with ya | 12:11 |
| flaper87 | ROFL | 12:11 |
| vkmc | ah well | 12:11 |
| vkmc | there is also some work with mqtt and stomp | 12:11 |
| vkmc | which may not be our target now... buuuuuut | 12:12 |
| flaper87 | vkmc: I know sebasmagri was interested in doing something with mqtt but he keeps bailing on us | 12:12 |
| flaper87 | ;) | 12:12 |
| vkmc | oh cool, didn't know about it | 12:13 |
| vkmc | hey another thing | 12:14 |
| vkmc | what's the deal with this change? https://review.openstack.org/#/c/122591/ | 12:14 |
| vkmc | flaper87, ^ | 12:14 |
| vkmc | should we remove grace update from the API as well? | 12:15 |
| *** flwang has quit IRC | 12:15 | |
| flaper87 | vkmc: mmmh | 12:19 |
| flaper87 | vkmc: I gotta take a better look at that but I think the change is correct | 12:19 |
| flaper87 | we don't allow updating the grace in the claims update | 12:20 |
| flaper87 | AFAIK | 12:20 |
| vkmc | we kinda do | 12:21 |
| vkmc | I dunno if we should | 12:21 |
| vkmc | but we do | 12:21 |
| vkmc | haha | 12:21 |
| flaper87 | LOL | 12:22 |
| flaper87 | sounds consistent /sarcasm | 12:22 |
| vkmc | of course, we produce the finest software | 12:23 |
| vkmc | since 2013 | 12:23 |
| vkmc | or 2012_ | 12:23 |
| vkmc | ?* | 12:23 |
| flaper87 | vkmc: 1984 | 12:24 |
| vkmc | ah well, Zaqar was there even before you born? | 12:24 |
| flaper87 | vkmc: Nope, I was there before I was born | 12:26 |
| vkmc | ok NO, I don't need another ride across your brain internals | 12:27 |
| vkmc | lol | 12:27 |
| *** jeffrey4l has quit IRC | 12:32 | |
| *** jeffrey4l has joined #openstack-zaqar | 12:53 | |
| *** X019 has quit IRC | 12:54 | |
| vkmc | ok, brb | 13:02 |
| *** sriram has joined #openstack-zaqar | 13:03 | |
| *** malini has quit IRC | 13:04 | |
| flaper87 | vkmc: http://svn.apache.org/viewvc/qpid/proton/branches/fadams-javascript-binding/proton-c/bindings/javascript/ | 13:28 |
| flaper87 | vkmc: I was told the ws support in proton is going to land soon | 13:29 |
| flaper87 | I guess we can't depend on it just yet | 13:29 |
| flaper87 | but we should totally keep an eye on that | 13:29 |
| *** jchai has joined #openstack-zaqar | 13:29 | |
| *** jeffrey4l has quit IRC | 13:33 | |
| flaper87 | kgriffs|afk: ping when you're around | 13:35 |
| flaper87 | kgriffs|afk: I'd like to discuss something w.r.t mongodb's FIFO support | 13:35 |
| *** sgotliv has quit IRC | 13:48 | |
| *** sgotliv has joined #openstack-zaqar | 13:48 | |
| flaper87 | kgriffs|afk: vkmc malini|afk https://review.openstack.org/#/c/125938/ | 13:53 |
| flaper87 | pls | 13:53 |
| *** malini has joined #openstack-zaqar | 13:53 | |
| *** cpallares has joined #openstack-zaqar | 13:58 | |
| *** amitgandhinz has joined #openstack-zaqar | 13:59 | |
| openstackgerrit | A change was merged to openstack/zaqar: Updated from global requirements https://review.openstack.org/131126 | 14:15 |
| *** jeffrey4l has joined #openstack-zaqar | 14:19 | |
| *** jchai is now known as jchai_afk | 14:40 | |
| jeffrey4l | flaper87, What's the requirement to add storage backend at the running time( i.e. The register pool interface, which will connect a backend storage at the running time). Why not configuring it through the config file? Then restart/reload the service. | 14:53 |
| *** jchai_afk is now known as jchai | 14:58 | |
| jeffrey4l | I think there should be also bug/issue in there. if the storage driver already created, it will be never updated when updating the pool's option. | 14:59 |
| flaper87 | jeffrey4l: operators hate restarting services ;) | 14:59 |
| jeffrey4l | flaper87, ^ | 14:59 |
| flaper87 | jeffrey4l: I'm not sure I understand what you're referring to | 14:59 |
| flaper87 | is it about the pool stuff ? | 14:59 |
| jeffrey4l | yup. | 14:59 |
| jeffrey4l | i can describe it more detail. | 15:00 |
| jeffrey4l | 1. Register a pool ( for example named p1) | 15:00 |
| jeffrey4l | 2. When user create queue, it may use the p1 pool. | 15:01 |
| jeffrey4l | 3. Then the user can do what ever he want. But remember the backend driver in the p1 pool is created. | 15:02 |
| jeffrey4l | 4. The admin want to re-config the options of p1 pool. Then he can update the pool. | 15:03 |
| jeffrey4l | 5. Bug is here. The created backend driver in step 3 will never be updated by using the latest option configured in step 4. | 15:04 |
| jeffrey4l | flaper87, clearly? | 15:04 |
| flaper87 | jeffrey4l: what options do you think the operator would change? connection uri? | 15:10 |
| flaper87 | I don't think the operator should be allowed to do that, not if there are queues assigned to that pool | 15:10 |
| flaper87 | instead, the operator should create a different pool and then migrate the queue | 15:10 |
| flaper87 | Queue's migrations were discussed at the ATL summit and we were waiting for flavors to land to start working on that | 15:10 |
| jeffrey4l | I think the operator may change any options. Because there are many option existence in mongo/redis connection param. | 15:15 |
| *** sgotliv has quit IRC | 15:15 | |
| flaper87 | jeffrey4l: oh, you mean just the uri options | 15:16 |
| jeffrey4l | yep | 15:16 |
| flaper87 | jeffrey4l: mmh, I'm torn there, though. I mean, on whether that's a good thing to allow or not | 15:17 |
| flaper87 | I think I'd lean towards letting the operator update the URI and just have a way to refresh the connection of the already loaded driver | 15:17 |
| flaper87 | It should be pretty trivial | 15:18 |
| flaper87 | jeffrey4l: could you create a bug issue for that | 15:18 |
| flaper87 | ? | 15:18 |
| jeffrey4l | flaper87, ok. | 15:18 |
| jeffrey4l | I just hate to change the system config through the api. :p | 15:25 |
| *** jeffrey4l_ has joined #openstack-zaqar | 15:27 | |
| jeffrey4l_ | I just hate to change the system config through the api. :p | 15:27 |
| flaper87 | It's not really a system config. FWIW, the system config is where the pool's config are stored | 15:27 |
| jeffrey4l_ | ok. | 15:30 |
| jeffrey4l_ | flaper87, so what's the right direction for this bug? | 15:31 |
| *** jeffrey4l has quit IRC | 15:31 | |
| *** reed has joined #openstack-zaqar | 15:33 | |
| jeffrey4l_ | 1) stop the operator to update the options if the queue is using. OR 2) reload the driver after the options is updated. | 15:35 |
| jeffrey4l_ | flaper87, ^ | 15:38 |
| flaper87 | jeffrey4l_: I'm leaning towards 2 | 15:39 |
| flaper87 | lets discuss this with other folks too | 15:39 |
| flaper87 | kgriffs|afk: vkmc malini ^ | 15:39 |
| *** jchai is now known as jchai_afk | 15:40 | |
| *** jeffrey4l_ has quit IRC | 16:04 | |
| *** jchai_afk is now known as jchai | 16:05 | |
| *** reed has quit IRC | 16:15 | |
| *** mpanetta has joined #openstack-zaqar | 16:31 | |
| *** mpanetta has joined #openstack-zaqar | 16:32 | |
| *** zhiyan has quit IRC | 16:36 | |
| *** reed has joined #openstack-zaqar | 16:43 | |
| AAzza | flaper87: hi | 16:56 |
| AAzza | flaper87: yes, the only thing left is decide to what to visit on the conference and in Paris) | 16:58 |
| *** mpanetta has quit IRC | 17:20 | |
| *** mpanetta has joined #openstack-zaqar | 17:21 | |
| *** amitgandhinz has quit IRC | 17:25 | |
| *** exploreshaifali has joined #openstack-zaqar | 17:27 | |
| *** fifieldt has quit IRC | 17:43 | |
| *** zhiyan has joined #openstack-zaqar | 17:59 | |
| *** mpanetta has quit IRC | 18:02 | |
| *** mpanetta has joined #openstack-zaqar | 18:03 | |
| *** jchai is now known as jchai_afk | 18:09 | |
| *** amitgandhinz has joined #openstack-zaqar | 18:16 | |
| *** jchai_afk is now known as jchai | 18:24 | |
| *** exploreshaifali has quit IRC | 18:31 | |
| *** jchai is now known as jchai_afk | 18:39 | |
| *** kgriffs|afk is now known as kgriffs | 18:50 | |
| *** jchai_afk is now known as jchai | 18:50 | |
| flaper87 | AAzza: I'm pretty sure you have Tuesday packed :P | 19:00 |
| flaper87 | AAzza: Tuesday is Zaqar all day long | 19:00 |
| * flaper87 is so exited, we'll all be there | 19:01 | |
| *** flwang has joined #openstack-zaqar | 19:05 | |
| *** jchai is now known as jchai_afk | 19:08 | |
| kgriffs | flaper87: regarding jeffrey41's question | 19:12 |
| kgriffs | I think it makes sense to let the operator update some options, say retry timeouts, ssl options, etc. | 19:13 |
| kgriffs | today you would do that by updating the pool catalog | 19:13 |
| kgriffs | and then HUP your web heads | 19:14 |
| kgriffs | something we should document in any case (add to your documentation TODO list) | 19:14 |
| flaper87 | mmh, I'm not a fan of HUP when we can invalidate the loaded driver | 19:34 |
| flaper87 | I mean, I think updating options is find | 19:34 |
| *** malini has quit IRC | 19:34 | |
| flaper87 | kgriffs: while I have you here: Was there any reason not to increment the counter right away? https://github.com/openstack/zaqar/blob/master/zaqar/queues/storage/mongodb/messages.py#L517 | 19:35 |
| *** exploreshaifali has joined #openstack-zaqar | 19:35 | |
| flaper87 | I don't remember | 19:36 |
| * kgriffs looking | 19:36 | |
| kgriffs | flaper87: I remember thinking about it a lot but can't remember the exact "why". :p | 19:37 |
| kgriffs | i.e., it wasn't an arbitrary decision | 19:37 |
| kgriffs | not saying it can't be revisited, but needs to be thought through carefully if we do | 19:37 |
| kgriffs | let me see | 19:38 |
| kgriffs | suppose we increment it right away | 19:38 |
| kgriffs | I remember | 19:38 |
| flaper87 | by N where N is the number of messages to insert | 19:38 |
| kgriffs | then you have two parallel requests, say | 19:38 |
| kgriffs | they won't collide | 19:38 |
| kgriffs | however if the one with the larger basis marker inserts first (wins the race) | 19:39 |
| kgriffs | then a client lists messages | 19:39 |
| kgriffs | it gets the higher marker value | 19:39 |
| kgriffs | then the other request to post messages completes | 19:39 |
| flaper87 | ah right | 19:39 |
| kgriffs | now the client won't see the last batch | 19:39 |
| flaper87 | now I remember, I should probably write that down | 19:39 |
| kgriffs | d00d, some days I am tempted to saw "forget the extra messaging patterns" and just do claim-based message retrieval only. it would be soooo much simpler. :p | 19:40 |
| *** AAzza is now known as AAzza_afk | 19:40 | |
| kgriffs | less useful, but a lot easier. | 19:41 |
| kgriffs | s/saw/say | 19:41 |
| kgriffs | thing is, there is still a need for stuff like pub-sub, and once you do that you have solved most of the implementation and adding on claims is a fairly straightforward extension. but I digress... | 19:42 |
| *** flwang has quit IRC | 19:43 | |
| kgriffs | flaper87: re updating options - yeah, it would be nice if Zaqar would periodically check last update time for a given pool and reload the driver automagically. | 19:44 |
| *** jchai_afk is now known as jchai | 19:53 | |
| *** cpallares has quit IRC | 20:04 | |
| *** flwang has joined #openstack-zaqar | 20:16 | |
| flaper87 | kgriffs: I think we can do that during the update | 20:51 |
| flaper87 | Update the pool -> reload the driver | 20:52 |
| kgriffs | on all the web heads? | 20:52 |
| flaper87 | kgriffs: oh, mmh. that's a very good point | 20:52 |
| * flaper87 slaps himself with the "distributed" guide | 20:52 | |
| kgriffs | :) | 20:52 |
| kgriffs | I was thinking we could just cache the "last update" timestamp as well as a "last checked" | 20:53 |
| flaper87 | kgriffs: right, and we go back to our last discussion re Redis. Is it fair to ask for it to *always* be installed ? | 20:53 |
| kgriffs | if it is just for caching, I don't think we have to require it assuming we are using a pluggable caching library | 20:54 |
| *** jchai is now known as jchai_afk | 20:54 | |
| kgriffs | speaking of which | 20:55 |
| kgriffs | oslo cache library - what's the status? | 20:55 |
| kgriffs | because right now | 20:55 |
| kgriffs | zaqar only supports in-process (memory) cache | 20:56 |
| flaper87 | kgriffs: http://doughellmann.com/2014/10/26/oslo-goals-for-kilo-cycle.html#cache-configuration-library | 20:56 |
| flaper87 | kgriffs: yeah | 20:56 |
| kgriffs | flaper87: I think this is actually somthing that fell through the cracks for us | 20:58 |
| kgriffs | by the end of kilo we really need to support memcached and/or redis for caching | 20:59 |
| kgriffs | that means either migrating to use dogpile directly, or getting that cache spec done | 20:59 |
| *** jchai_afk is now known as jchai | 21:01 | |
| *** sriram has quit IRC | 21:02 | |
| *** jchai has quit IRC | 21:15 | |
| *** amitgandhinz has quit IRC | 21:41 | |
| *** exploreshaifali has quit IRC | 21:43 | |
| *** openstackgerrit has quit IRC | 21:50 | |
| *** cpallares has joined #openstack-zaqar | 21:53 | |
| *** cpallares has quit IRC | 21:53 | |
| *** kgriffs is now known as kgriffs|afk | 22:45 | |
| *** mpanetta has quit IRC | 23:23 | |
| *** jeffrey4l_ has joined #openstack-zaqar | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!