| *** openstackgerrit has quit IRC | 00:21 | |
| *** openstackgerrit has joined #openstack-zaqar | 00:21 | |
| *** kgriffs is now known as kgriffs|afk | 00:35 | |
| *** achanda has quit IRC | 00:56 | |
| *** achanda has joined #openstack-zaqar | 01:00 | |
| *** openstackgerrit has quit IRC | 01:05 | |
| *** openstackgerrit has joined #openstack-zaqar | 01:05 | |
| *** amalagon_ has quit IRC | 01:19 | |
| *** amalagon has joined #openstack-zaqar | 02:06 | |
| *** achanda has quit IRC | 02:09 | |
| *** achanda has joined #openstack-zaqar | 02:10 | |
| *** openstackgerrit has quit IRC | 02:20 | |
| *** openstackgerrit has joined #openstack-zaqar | 02:21 | |
| *** wpf has quit IRC | 02:36 | |
| *** wpf has joined #openstack-zaqar | 02:37 | |
| *** openstackgerrit has quit IRC | 03:20 | |
| *** openstackgerrit has joined #openstack-zaqar | 03:20 | |
| *** achanda has quit IRC | 03:32 | |
| *** wpf has quit IRC | 03:43 | |
| *** achanda has joined #openstack-zaqar | 03:56 | |
| *** wpf has joined #openstack-zaqar | 04:00 | |
| *** achanda has quit IRC | 04:06 | |
| *** wpf has quit IRC | 04:10 | |
| *** boris-42 has quit IRC | 04:13 | |
| *** wpf has joined #openstack-zaqar | 04:23 | |
| *** flwang has quit IRC | 04:27 | |
| *** amalagon has quit IRC | 04:30 | |
| *** achanda has joined #openstack-zaqar | 04:45 | |
| *** achanda has quit IRC | 04:46 | |
| *** achanda has joined #openstack-zaqar | 04:57 | |
| *** echevemaster has quit IRC | 05:31 | |
| *** wpf has quit IRC | 06:19 | |
| *** wpf has joined #openstack-zaqar | 06:21 | |
| *** reed has quit IRC | 06:26 | |
| *** achanda has quit IRC | 06:53 | |
| *** achanda has joined #openstack-zaqar | 06:58 | |
| *** achanda has quit IRC | 07:19 | |
| *** dynarro has joined #openstack-zaqar | 08:12 | |
| *** dynarro has quit IRC | 09:01 | |
| *** dynarro has joined #openstack-zaqar | 09:07 | |
| *** flwang has joined #openstack-zaqar | 09:14 | |
| flwang | flaper87: poing | 09:17 |
|---|---|---|
| flaper87 | flwang: poing ? | 09:17 |
| flaper87 | :P | 09:17 |
| flaper87 | or I should say | 09:17 |
| flaper87 | piong ? | 09:17 |
| flwang | flaper87: you're smart :) | 09:17 |
| flwang | are u available now? | 09:17 |
| flaper87 | yes | 09:17 |
| flwang | as for your comments, https://review.openstack.org/#/c/142432/8/zaqar/transport/validation.py | 09:19 |
| flwang | the TTL of subscription | 09:19 |
| flaper87 | ah yeah | 09:20 |
| flaper87 | shoot | 09:20 |
| *** achanda has joined #openstack-zaqar | 09:21 | |
| flwang | http://specs.openstack.org/openstack/zaqar-specs/specs/kilo/notification-api.html | 09:21 |
| flwang | I implemented TTL based on the spec | 09:21 |
| flaper87 | yes, I'm now questioning the spec | 09:22 |
| flaper87 | :P | 09:22 |
| flwang | flaper87: TBH, i was wondering it when I coded it | 09:23 |
| flwang | but I saw SNS has the similar design, iirc | 09:23 |
| flwang | let me check | 09:23 |
| flaper87 | flwang: nice, that means probably the right thing to do is either remove ttl or make it infinite by default | 09:23 |
| flaper87 | flwang: I think they don't have ttl | 09:23 |
| flaper87 | or at least it's not the default | 09:23 |
| flaper87 | flwang: brb | 09:26 |
| flwang | flaper87: checking sns doc... | 09:26 |
| flaper87 | back | 09:34 |
| flaper87 | flwang: anything interesting? | 09:34 |
| flwang | flaper87: nope, seeems my memory is not correct | 09:36 |
| flwang | arn:aws:sns:us-west-2:429486775985:alarmsqsarn:aws:sqs:us-west-2:429486775985:beijingarn:aws:sns:us-west-2:429486775985:alarm:77be0caf-126d-43a7-ab71-229f4e9f6523 | 09:36 |
| flwang | Topic Protocol Endpoint Subscription ID | 09:36 |
| flwang | this is the attributes of SNS's subscription | 09:37 |
| flaper87 | right, no ttl | 09:38 |
| flaper87 | so, I think ttl is a nice thing to have | 09:38 |
| flaper87 | but the default should be no ttl | 09:38 |
| flwang | flaper87: agree | 09:39 |
| flwang | I will drop another patch to fix that | 09:39 |
| flaper87 | ok | 09:39 |
| flaper87 | I think you could fix the API in this one and then fix the storage later | 09:40 |
| flwang | flaper87: yep, that's what I want to do | 09:40 |
| flwang | flaper87: any status about the graduation? | 09:42 |
| flaper87 | no graduation needed | 09:44 |
| flaper87 | Zaqar is now an official openstack project | 09:44 |
| flaper87 | since we all live under the same big tent now | 09:44 |
| flaper87 | I'm waiting for the tags thing to be resolved so we can apply to that | 09:44 |
| flaper87 | I gotta remove all the "incubated" tags now from everywhere | 09:44 |
| flwang | flaper87: sorry ? | 09:48 |
| flwang | flaper87: so there is no graduation thing in OpenStack, is it? | 09:48 |
| *** exploreshaifali has joined #openstack-zaqar | 09:48 | |
| flaper87 | flwang: nope, no more graduations | 09:48 |
| flwang | is there an official announcement from TC? | 09:49 |
| flwang | I didn't see that from the mail list | 09:49 |
| flaper87 | yeah, there's an active discussion on the tags | 09:52 |
| flaper87 | plus all the governance reviews for the big tent thing | 09:52 |
| flaper87 | but I think the final official announcement hasn't been made yet | 09:53 |
| flwang | flaper87: ok, so we won't be care about the incubated/graduation stuff anymore, right? | 09:54 |
| flaper87 | flwang: exactly | 09:56 |
| flaper87 | we're officially in | 09:56 |
| flwang | cool | 09:56 |
| flaper87 | now we need to get people to use Zaqar :/ | 09:56 |
| *** dynarro has quit IRC | 09:58 | |
| flwang | flaper87: as for the subscriber driver, any thoughts about where to put those code? | 10:21 |
| flwang | flaper87: how about like zaqar/common/subscriber/webhook.py? | 10:22 |
| flaper87 | mmh, is it something that should go in common ? | 10:23 |
| flaper87 | what about under notifications ? | 10:23 |
| flaper87 | or a "notifiers" package | 10:23 |
| *** exploreshaifali has quit IRC | 10:28 | |
| flwang | flaper87: did you mean create a new notifications package under zaqar? at the same level with transport, storage? | 10:30 |
| flaper87 | flwang: yup, probably notifiers | 10:33 |
| flaper87 | mmh | 10:33 |
| flaper87 | I'm really bad at naming things | 10:33 |
| flaper87 | so :P | 10:33 |
| flwang | flaper87: see https://github.com/openstack/zaqar/tree/master/zaqar/common | 10:34 |
| flwang | we're putting transport, storage stuff under common | 10:34 |
| flwang | and you know, in glance | 10:34 |
| flwang | we also put the async task driver and those scripts under common | 10:35 |
| flaper87 | yeah but those are helpers | 10:35 |
| flwang | flaper87: yep | 10:35 |
| flaper87 | what you're working on now are the actual drivers that will push stuff to clients | 10:35 |
| flwang | i can see your point | 10:35 |
| flaper87 | and they won't be used throughout zaqar but just the notification feature | 10:35 |
| flwang | flaper87: so how about just keep a new subscriber package under zaqar? | 10:36 |
| flwang | just like storage/transport | 10:36 |
| flwang | and we can refactor it in the future if it's necessary | 10:36 |
| flaper87 | flwang: sounds good | 10:37 |
| *** exploreshaifali has joined #openstack-zaqar | 10:39 | |
| flwang | cool | 10:40 |
| flaper87 | flwang: can I get some reviews from you here? https://review.openstack.org/#/q/status:open+project:openstack/zaqar,n,z | 10:43 |
| flaper87 | we need to move some patches forward | 10:43 |
| flwang | flaper87: it's on my todo list :) | 10:43 |
| flwang | I will to that tonight | 10:43 |
| flwang | to/do | 10:43 |
| flaper87 | flwang: thank you very much | 10:44 |
| flwang | flaper87: sorry, i have a lot of internal work to do recently | 10:44 |
| flaper87 | flwang: no worries at all | 10:45 |
| flaper87 | that's why I'm here, to make sure you get fired for taking care of things you probably shouldn't | 10:45 |
| flaper87 | :) | 10:46 |
| openstackgerrit | Flavio Percoco proposed openstack/zaqar: Use pool's capabilities for flavors in v2 https://review.openstack.org/148566 | 10:46 |
| flwang | :P | 10:46 |
| flaper87 | exploreshaifali: looking at your patch in more depth now | 10:47 |
| *** exploreshaifali has quit IRC | 11:08 | |
| *** achanda has quit IRC | 11:22 | |
| *** exploreshaifali has joined #openstack-zaqar | 11:23 | |
| exploreshaifali | flaper87: Thanks for that :) | 11:24 |
| vkmc | morniiiiiiiiing | 11:36 |
| flwang | flaper87: still around? | 11:39 |
| exploreshaifali | vkmc: morning!!!!!!!11 | 11:40 |
| vkmc | how are you all? | 11:43 |
| vkmc | hey flwang! | 11:43 |
| flwang | vkmc: morning | 11:43 |
| vkmc | flwang, +1 to the subscription package | 11:45 |
| flwang | thanks :) | 11:46 |
| vkmc | although | 12:06 |
| vkmc | regarding the level | 12:06 |
| vkmc | subscriptions could be considered as part of transport | 12:06 |
| *** exploreshaifali has quit IRC | 12:13 | |
| *** dynarro has joined #openstack-zaqar | 12:15 | |
| *** dynarro has quit IRC | 12:20 | |
| flaper87 | flwang: I'm now | 12:26 |
| *** exploreshaifal has joined #openstack-zaqar | 12:32 | |
| exploreshaifal | vkmc: if you have a bit of time please can we disucss about the issue of message_controller which I am facing right now ? | 12:33 |
| vkmc | exploreshaifal, sure, could you please update me with the current situation? | 12:36 |
| exploreshaifal | vkmc: yes, current task is moving QueueController storage cofigurations into Control plane from data plane | 12:37 |
| exploreshaifal | that is basicallly shifting queue_controller method https://github.com/openstack/zaqar/blob/master/zaqar/storage/base.py#L213 from DataDriver to ControllerDriver | 12:39 |
| exploreshaifal | okay | 12:39 |
| exploreshaifal | so what I did is rather than shifting queue_controller method, I changed it defination | 12:39 |
| exploreshaifal | such that it will access QueueController from ControlDriver | 12:40 |
| vkmc | exploreshaifal, cool, I follow that | 12:40 |
| exploreshaifal | so now QueueController is instantiated througn ControlDriver and not through DataDriver | 12:40 |
| exploreshaifal | vkmc: here is my current patch https://review.openstack.org/#/c/144340/ | 12:41 |
| exploreshaifal | see storage/mongodb/driver.py file | 12:41 |
| exploreshaifal | vkmc: so now the issue is that ControlDriver is instantiating QueueController, ControlDriver does not have message_controller | 12:43 |
| exploreshaifal | but few method of QueueController need message_controller | 12:43 |
| exploreshaifal | like https://github.com/openstack/zaqar/blob/master/zaqar/storage/mongodb/queues.py#L281 | 12:43 |
| exploreshaifal | previously it was working well because DataDriver was instantialing QueueController, and DataDriver have messageController | 12:44 |
| exploreshaifal | now we can't add message_controller to ControlDriver becaues it won't be correct | 12:44 |
| exploreshaifal | so how to make this work? | 12:44 |
| vkmc | ohhhh I see | 12:45 |
| vkmc | so definitely we will have to change a little the design | 12:45 |
| vkmc | I mean, we changed our POV about the place queues have in Zaqar | 12:45 |
| vkmc | so we should also change how we manage other entities in Zaqar | 12:45 |
| flaper87 | vkmc: any thoughts on that? | 12:46 |
| flaper87 | this one be an easy thing to model | 12:46 |
| flaper87 | FWIW, I really hate that the queue controller depends on the message_controller | 12:46 |
| vkmc | yeah | 12:46 |
| flaper87 | we could probably move stats down there | 12:46 |
| vkmc | I think we should remove that dependency | 12:46 |
| vkmc | that is what we wanted to do in the first place | 12:46 |
| vkmc | so... for what I see, delete and stats use it | 12:46 |
| flaper87 | redis uses it in other places | 12:47 |
| flaper87 | IIRC | 12:47 |
| vkmc | oh | 12:47 |
| vkmc | let's check redis | 12:47 |
| vkmc | no.. better | 12:48 |
| vkmc | lets focus on mongo first | 12:48 |
| vkmc | one thing we should figure it out is how we make this change in a restful way | 12:50 |
| flaper87 | well, not much will change from what we have | 12:50 |
| flaper87 | I mean, if you delete a queue, the existing behavior should remain | 12:50 |
| flaper87 | the queue and its messages should be deleted | 12:50 |
| flaper87 | I think we should have some event based pipeline | 12:51 |
| vkmc | yeah, I was thinking on... what happens if we delegate the remove messages to the message controller directly | 12:51 |
| flaper87 | Right now, we've a pipeline implementation that calls a method for each object in the pipeline | 12:51 |
| flaper87 | vkmc: mmh, not sure I follow. The only reason a queue would call message deletion is because it's to drop the queue | 12:52 |
| vkmc | flaper87, in our lazy way of viewing things, a queue exist only if it has messages | 12:53 |
| flaper87 | right | 12:53 |
| vkmc | so if you delete all the messages == you dropped the queue | 12:53 |
| flaper87 | unless the queue has been created explicitly | 12:54 |
| flaper87 | as in, there's a metadata | 12:54 |
| vkmc | hmm | 12:54 |
| vkmc | good point | 12:54 |
| vkmc | so, we depend on the queue controller knowing the message controller | 12:55 |
| vkmc | because, in the case of a deletion, if the queue is empty and only has metadata... ok, just drop it... but if it has messages, then it has to delete every message there | 12:56 |
| vkmc | and that is something the message controller does | 12:56 |
| vkmc | and that is because we are considering the queue as a container of message, not as an attribute only | 12:56 |
| vkmc | flaper87, what is your plan with the pipeline? | 12:57 |
| exploreshaifal | can we shift the delete and stats to message_controller | 12:58 |
| exploreshaifal | brb in 10 min | 13:00 |
| *** exploreshaifal has quit IRC | 13:04 | |
| flaper87 | vkmc: sooooooooooooooooooooo | 13:11 |
| flaper87 | this is just a fresh thought | 13:11 |
| flaper87 | We've been (IIRC) talking about *always* using a pipeline | 13:12 |
| flaper87 | this would be useful for the notifications thing | 13:12 |
| flaper87 | so, what if we would use this same pipeline for something more than that | 13:13 |
| * flaper87 working out the idea as he writes | 13:13 | |
| * vkmc tries to follow flaper87's train of thoughts | 13:13 | |
| flaper87 | What if we add the message controller to the queue's pipeline so it's called whenever a queue is deleted | 13:14 |
| flaper87 | That means we need to rename the `delete` method to `delete_queue` or something like that | 13:14 |
| * flaper87 checks the code | 13:14 | |
| vkmc | hmm | 13:15 |
| flaper87 | so, yeah, we could do something really nasty here | 13:16 |
| flaper87 | We can creare a MessageQueuecontroller | 13:16 |
| flaper87 | that has a `_delete` method with the same signature as the QueueController | 13:16 |
| flaper87 | and we'd register it in the queue's pipeline | 13:16 |
| flaper87 | if only I remembered how the pipeline works | 13:17 |
| flaper87 | 2s, checking code again | 13:17 |
| flaper87 | so yeah | 13:20 |
| flaper87 | I think we've most of what we need to do that | 13:21 |
| flaper87 | the pipeline loading process may need some polishment | 13:21 |
| flaper87 | vkmc: ^ thoughts ? | 13:22 |
| vkmc | flaper87, I'm thinking what is the advantage of doing that | 13:24 |
| vkmc | instead of having the dataplane and controlplane tightly couple (aka keeping references to the controllers we need) | 13:24 |
| flaper87 | It'll be really hard to keep the references because now we have a circular dependency | 13:25 |
| flaper87 | FWIW, the data driver will still have a reference to the control driver | 13:26 |
| flaper87 | but not the other way around | 13:26 |
| flaper87 | and the other benefit is that when changes like this one need to happen, it's easier to do them | 13:26 |
| *** exploreshaifali has joined #openstack-zaqar | 13:27 | |
| * vkmc checks the pipeline code | 13:28 | |
| *** boris-42 has joined #openstack-zaqar | 13:31 | |
| flaper87 | flwang: 'x<=0' <- SERIOUSLY ? >.> | 13:33 |
| flaper87 | :P | 13:33 |
| flaper87 | let me add the docstring | 13:33 |
| vkmc | ok so | 13:34 |
| vkmc | when the enqueing in the pipeline ocurrs? | 13:34 |
| vkmc | we are using the pipeline in the storage layer | 13:34 |
| flaper87 | vkmc: but we're using it for pooling | 13:35 |
| flaper87 | not for the general case | 13:35 |
| flaper87 | the pipeline loads entry_points when creating instances for the many controllers | 13:35 |
| flaper87 | if we add an entry point for the queue_pipeline that delets messages, then we should be good to go | 13:35 |
| vkmc | flaper87, I'm checking out this pipeline over here https://github.com/openstack/zaqar/blob/master/zaqar/storage/pipeline.py | 13:38 |
| openstackgerrit | Flavio Percoco proposed openstack/zaqar: Forbid adding stores with mismatching capabilities https://review.openstack.org/145783 | 13:38 |
| openstackgerrit | Flavio Percoco proposed openstack/zaqar: Use pool's capabilities for flavors in v2 https://review.openstack.org/148566 | 13:38 |
| openstackgerrit | Flavio Percoco proposed openstack/zaqar: Split MongoDB's default implementation from FIFO's https://review.openstack.org/148297 | 13:38 |
| flaper87 | vkmc: flwang ^ | 13:38 |
| flaper87 | vkmc: yup | 13:38 |
| flaper87 | vkmc: https://github.com/openstack/zaqar/blob/master/zaqar/storage/pipeline.py#L44-L81 | 13:39 |
| flaper87 | we should probably move that to `common/pipeline.py` | 13:39 |
| flaper87 | so we can use it in other places | 13:39 |
| flaper87 | That said | 13:39 |
| flaper87 | vkmc: https://github.com/openstack/zaqar/blob/master/zaqar/storage/pipeline.py#L71 | 13:39 |
| flaper87 | vkmc: https://github.com/openstack/zaqar/blob/master/zaqar/storage/pipeline.py#L110 | 13:39 |
| flaper87 | See how the pipeline for queue's is loaded? | 13:39 |
| flaper87 | at the very end of the "dynamic" pipeline, the queue_controller is appended | 13:40 |
| flaper87 | vkmc: https://github.com/openstack/zaqar/blob/master/zaqar/storage/pipeline.py#L111 | 13:40 |
| vkmc | flaper87, yeah | 13:40 |
| flaper87 | if we add entrypoints for other things we want to call on each pipeline | 13:40 |
| vkmc | flaper87, so... what I was trying to say, we are not using the pipe for pooling only | 13:40 |
| flaper87 | vkmc: oh really? | 13:40 |
| flaper87 | NIIIIIIIIIICE | 13:41 |
| * vkmc tries to detect sarcarsm | 13:41 | |
| flaper87 | https://github.com/openstack/zaqar/blob/master/zaqar/bootstrap.py#L96-L108 | 13:41 |
| flaper87 | no no, I really didn't know that | 13:41 |
| vkmc | aahahahaha | 13:41 |
| flaper87 | I had forgotten we weren't using it just for pooling | 13:41 |
| flaper87 | that makes this even easier | 13:41 |
| vkmc | yeah, that's the point | 13:41 |
| vkmc | but but ... we add the message controller endpoint in the pipeline | 13:42 |
| vkmc | already | 13:42 |
| vkmc | so adding the delete method should be enough *I think* | 13:42 |
| flaper87 | nope because that's the *message* pipeline | 13:42 |
| flaper87 | we need to add a "message controller" to the queue's pipeline | 13:42 |
| vkmc | we use the words message and queue too much | 13:46 |
| * flaper87 reminds vkmc we're working on a messaging system that happens to have queues | 13:47 | |
| flaper87 | ok that was a joke | 13:48 |
| flaper87 | I couldn't hold it back | 13:48 |
| flaper87 | :P | 13:48 |
| vkmc | aahahaha I know but | 13:48 |
| vkmc | re read this sentence 'nope because that's the *message* pipeline - we need to add a "message controller" to the queue's pipeline' | 13:49 |
| vkmc | it sounded like this in my brain 'thats the message we need to add the message to the queue' | 13:49 |
| flaper87 | vkmc: if you're confused, that was the goal | 13:50 |
| flaper87 | jokes apart | 13:50 |
| vkmc | ok, seriously... in bootstrap.py we are returning a pipeline with the DataDriver https://github.com/openstack/zaqar/blob/master/zaqar/bootstrap.py#L108 | 13:50 |
| flaper87 | We need to create in mongodb.messages a `QueueMessageHandlerThingy` | 13:50 |
| flaper87 | then we need to add that to the entry-points and let the pipeline do its job | 13:50 |
| vkmc | we want to do something similar for control() | 13:51 |
| vkmc | ok exploreshaifali, seems like there is some consensus on what to do | 13:54 |
| exploreshaifali | vkmc, yeah, I am following whole discussion here | 13:54 |
| vkmc | flaper87, in the case of stats... we could proceed the same way? | 13:55 |
| flaper87 | vkmc: I'd assume so, yes. | 13:56 |
| flaper87 | not sure how we can "merge" the results, though. | 13:56 |
| flaper87 | Unlike deletes, for stats we need to unify the results of each call | 13:56 |
| flaper87 | I think this should be hanlded by the transport itself | 13:57 |
| flaper87 | not sure | 13:57 |
| vkmc | stats are generated on the fly so... we could delegate that to the transport | 13:57 |
| vkmc | exploreshaifali, where is the etherpad you were working on? let's write some notes there | 14:02 |
| exploreshaifali | vkmc, https://etherpad.openstack.org/p/exploreshaifali-opw-split-layers | 14:03 |
| *** sriram has joined #openstack-zaqar | 14:13 | |
| *** exploreshaifali has quit IRC | 14:32 | |
| *** mpanetta has joined #openstack-zaqar | 14:33 | |
| *** jasondot_ has joined #openstack-zaqar | 14:34 | |
| *** jasondot_ has quit IRC | 14:45 | |
| openstackgerrit | Merged openstack/zaqar: Refactor Api classes to enhance consistency https://review.openstack.org/148265 | 15:00 |
| *** kgriffs|afk is now known as kgriffs | 15:04 | |
| *** amitgandhinz has joined #openstack-zaqar | 15:26 | |
| *** achanda has joined #openstack-zaqar | 15:31 | |
| *** achanda has quit IRC | 15:41 | |
| *** diga has joined #openstack-zaqar | 15:42 | |
| *** diga has quit IRC | 16:09 | |
| *** amalagon has joined #openstack-zaqar | 16:16 | |
| *** ametts has joined #openstack-zaqar | 16:40 | |
| *** exploreshaifali has joined #openstack-zaqar | 16:48 | |
| *** cpallares has joined #openstack-zaqar | 17:23 | |
| exploreshaifali | vkmc, around? | 17:30 |
| *** bradjones has quit IRC | 17:38 | |
| *** bradjones_ has joined #openstack-zaqar | 17:39 | |
| *** bradjones_ is now known as bradjones | 17:39 | |
| *** bradjones has joined #openstack-zaqar | 17:39 | |
| *** reed has joined #openstack-zaqar | 18:00 | |
| *** ff has joined #openstack-zaqar | 18:11 | |
| *** achanda has joined #openstack-zaqar | 18:29 | |
| *** exploreshaifali has quit IRC | 18:34 | |
| *** openstackgerrit has quit IRC | 18:50 | |
| *** openstackgerrit has joined #openstack-zaqar | 18:51 | |
| *** kgriffs is now known as kgriffs|afk | 18:53 | |
| *** kgriffs|afk is now known as kgriffs | 18:53 | |
| *** flwang has quit IRC | 18:56 | |
| *** achanda has quit IRC | 19:10 | |
| *** achanda has joined #openstack-zaqar | 19:18 | |
| *** reed has quit IRC | 19:27 | |
| *** amitgandhinz has quit IRC | 19:44 | |
| *** amitgandhinz has joined #openstack-zaqar | 19:45 | |
| *** amitgandhinz has quit IRC | 19:52 | |
| *** amitgandhinz has joined #openstack-zaqar | 19:53 | |
| *** flwang has joined #openstack-zaqar | 19:56 | |
| * vkmc lurks | 19:58 | |
| *** openstackgerrit has quit IRC | 20:04 | |
| *** openstackgerrit has joined #openstack-zaqar | 20:05 | |
| *** kgriffs is now known as kgriffs|afk | 20:05 | |
| *** amitgandhinz has quit IRC | 20:07 | |
| *** exploreshaifali has joined #openstack-zaqar | 20:14 | |
| *** fifieldt has quit IRC | 20:24 | |
| *** fifieldt has joined #openstack-zaqar | 20:24 | |
| *** achanda has quit IRC | 20:55 | |
| *** miqui has quit IRC | 21:18 | |
| *** achanda has joined #openstack-zaqar | 21:19 | |
| *** miqui has joined #openstack-zaqar | 21:21 | |
| flwang | flaper87: ping | 21:31 |
| *** flwang has quit IRC | 21:41 | |
| *** flwang has joined #openstack-zaqar | 21:41 | |
| *** achanda has quit IRC | 22:03 | |
| *** kgriffs|afk is now known as kgriffs | 22:04 | |
| *** kgriffs is now known as kgriffs|afk | 22:10 | |
| *** achanda has joined #openstack-zaqar | 22:14 | |
| *** achanda has quit IRC | 22:14 | |
| ff | flwang: pong | 22:15 |
| *** ff has quit IRC | 22:15 | |
| flaper87 | flwang: pong | 22:16 |
| *** kgriffs|afk is now known as kgriffs | 22:17 | |
| *** achanda has joined #openstack-zaqar | 22:17 | |
| *** sriram has quit IRC | 22:20 | |
| flwang | flaper87: as for your comments https://review.openstack.org/#/c/142432/8/zaqar/tests/unit/transport/wsgi/v2_0/test_subscriptions.py | 22:23 |
| flwang | about the db uri | 22:23 |
| flaper87 | flwang: sup ? | 22:26 |
| flwang | now messages test is used the same way like I did | 22:26 |
| flwang | and there is any other test using pool for test(pls correct me) | 22:26 |
| flwang | and I also tried use sqlite memory db, but it will fail | 22:27 |
| *** mpanetta has quit IRC | 22:27 | |
| flwang | I haven't find the root cause | 22:27 |
| flaper87 | np, what I meant to say there is that we should probably have a generic way to determine what db should be loaded | 22:28 |
| flaper87 | That test exists just in test_mongo_impl | 22:28 |
| flaper87 | I mean, it's subclassed there | 22:28 |
| *** kgriffs is now known as kgriffs|afk | 22:34 | |
| flwang | flaper87: https://github.com/openstack/zaqar/blob/master/zaqar/tests/unit/transport/wsgi/v2_0/test_messages.py#L42 | 22:44 |
| flwang | I think you're talking about two points. #1 is we shouldn't allow duplicated db uri for pools | 22:45 |
| flaper87 | yeah, forget about the duplicate uris for now | 22:45 |
| flwang | #2 is remove the mongodb specific stuff | 22:46 |
| flaper87 | mmh, but that test isn't a base test | 22:46 |
| flaper87 | mhh, I'd like to remove mongodb specific things | 22:46 |
| flaper87 | if we can move that to base.V2Base | 22:46 |
| flaper87 | and make loading a db uri dynamic, it'd be better | 22:47 |
| flaper87 | can you file a bug for that, at least ? | 22:47 |
| flaper87 | and put a comment there | 22:47 |
| flwang | flaper87: sure | 22:47 |
| flwang | flaper87: i just want to point out the db uri can't be replace with sqlite memory db at here | 22:48 |
| flwang | flaper87: we may need another patch to address that | 22:48 |
| flaper87 | yeah, we'll fix this in a follow up patch | 22:49 |
| flaper87 | vkmc: will do it | 22:49 |
| flaper87 | :P | 22:49 |
| flwang | flaper87: haha | 22:49 |
| flwang | flaper87: I will file 2 bugs to address above issues separately | 22:50 |
| *** openstackgerrit has quit IRC | 22:51 | |
| *** openstackgerrit has joined #openstack-zaqar | 22:51 | |
| flwang | flaper87: as for the notification sending, last time you suggested add a post_message method in the subscriber driver, right? | 22:53 |
| flaper87 | flwang: yeah, well. it should have the same name and signature as the message controller `post` method | 22:54 |
| flwang | flaper87: but I'm still not very clear when to trigger the notification sending | 22:54 |
| flwang | flaper87: i mean | 22:54 |
| flwang | when user post a message to a queue, then transport layer will get the message at https://github.com/openstack/zaqar/blob/master/zaqar/transport/wsgi/v2_0/messages.py#L148 | 22:55 |
| flwang | then the it will call the post method of message controller | 22:56 |
| flaper87 | flwang: right but, what the transport actually does is call the *pipeline* which will the call the message_controller | 22:57 |
| flaper87 | we want to put the subscribers controller in there so we can catch the `post_message` call and send the notification | 22:58 |
| flaper87 | does that make sense? | 22:58 |
| flwang | flaper87: so, you mean we need to add subscriber to the pipeline of message controller, is it? | 22:59 |
| flaper87 | yes | 22:59 |
| flaper87 | :) | 22:59 |
| flaper87 | somehow, in a nice way :D | 22:59 |
| flwang | ok, it may works, but just a little big ugly from the semantics view :D | 23:00 |
| flwang | thanks for your comment, Fla | 23:03 |
| flaper87 | flwang: np :) | 23:03 |
| *** cpallares has quit IRC | 23:18 | |
| *** echevemaster has joined #openstack-zaqar | 23:30 | |
| *** exploreshaifali has quit IRC | 23:37 | |
| *** ametts has quit IRC | 23:39 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!