| *** wanghao has joined #openstack-zaqar | 00:15 | |
| openstackgerrit | Merged openstack/zaqar-ui master: Imported Translations from Zanata https://review.openstack.org/449459 | 01:02 |
|---|---|---|
| *** yanyanhu has joined #openstack-zaqar | 01:26 | |
| *** zhurong has joined #openstack-zaqar | 02:04 | |
| *** mpanetta has quit IRC | 02:05 | |
| *** mpanetta has joined #openstack-zaqar | 02:06 | |
| *** wanghao has quit IRC | 03:02 | |
| *** wanghao has joined #openstack-zaqar | 03:03 | |
| openstackgerrit | Feilong Wang proposed openstack/zaqar master: Using generic instead of v3 for identity https://review.openstack.org/451154 | 03:08 |
| flwang | wxy_: ping | 03:08 |
| wxy_ | flwang, here | 03:09 |
| flwang | wxy_: see this https://review.openstack.org/451090 ? | 03:09 |
| *** wanghao_ has joined #openstack-zaqar | 03:10 | |
| wxy_ | flwang, yeah, I'm fixing the test now. | 03:10 |
| flwang | wxy_: cool | 03:12 |
| flwang | I checked keystone config | 03:12 |
| flwang | keystone doesn't use the 'enabled' as well | 03:12 |
| *** wanghao has quit IRC | 03:13 | |
| flwang | btw, i'm testing the performance of swift | 03:14 |
| flwang | and i found it's very slow than mongo | 03:15 |
| wxy_ | flwang, have you found some reason? | 03:25 |
| flwang | you mean the swift driver? | 03:27 |
| flwang | i think it just because the communication is http :( it's slow | 03:27 |
| flwang | wxy_: have you found the reason why https://review.openstack.org/#/c/451090/ failed ? | 03:47 |
| wxy_ | yeah | 03:48 |
| flwang | cool | 03:49 |
| wxy_ | flwang, But I'm not suer it's related to the falcon or cors. | 03:51 |
| flwang | oh, really? | 03:51 |
| wxy_ | flwang, fixing now. | 03:51 |
| flwang | interesting | 03:51 |
| flwang | wxy_: feel free to upload patch set on my patch | 03:51 |
| wxy_ | flwang, ok | 03:52 |
| wxy_ | flwang, such as zaqar.tests.unit.transport.wsgi.v2_0.test_messages L390 | 03:52 |
| wxy_ | flwang, the value of "self.srmock.headers_dict['location']" is changed. So the test failed. | 03:53 |
| wxy_ | flwang, before is "v2/xxxx", now is "http://www.xx.org/v2/xxxx" | 03:53 |
| flwang | ah, because the cors is impacting the location? | 03:54 |
| wxy_ | flwang, Perhaps, or falcon's test frame | 03:56 |
| wxy_ | flwang, I'll try to fix it today. | 03:56 |
| flwang | ok, thanks | 03:57 |
| *** flwang1 has quit IRC | 04:05 | |
| openstackgerrit | wangxiyuan proposed openstack/zaqar master: Remove enabled option for CORS https://review.openstack.org/451090 | 04:14 |
| *** XueFeng has joined #openstack-zaqar | 04:52 | |
| *** flwang1 has joined #openstack-zaqar | 05:21 | |
| *** nadya has joined #openstack-zaqar | 05:38 | |
| *** rcernin has joined #openstack-zaqar | 05:39 | |
| *** nadya has quit IRC | 05:41 | |
| *** XueFeng has quit IRC | 05:44 | |
| openstackgerrit | wanghao proposed openstack/zaqar-specs master: Service Queue in Zaqar https://review.openstack.org/449433 | 05:53 |
| *** zhurong has quit IRC | 05:54 | |
| *** pcaruana has joined #openstack-zaqar | 06:24 | |
| openstackgerrit | wanghao proposed openstack/zaqar-specs master: Service Queue in Zaqar https://review.openstack.org/449433 | 06:46 |
| *** XueFeng has joined #openstack-zaqar | 06:50 | |
| *** tesseract has joined #openstack-zaqar | 07:14 | |
| *** amoralej|off is now known as amoralej | 07:24 | |
| openstackgerrit | wangxiyuan proposed openstack/zaqar master: Remove enabled option for CORS https://review.openstack.org/451090 | 07:34 |
| *** zhurong has joined #openstack-zaqar | 07:34 | |
| *** flwang1 has quit IRC | 07:35 | |
| *** nadya has joined #openstack-zaqar | 08:22 | |
| *** wanghao has joined #openstack-zaqar | 08:25 | |
| *** wanghao_ has quit IRC | 08:27 | |
| therve | wanghao, Let me know if you want to chat about the service queue spec | 08:29 |
| wanghao | therve: sure | 08:30 |
| therve | Don't hesitate to write stuff in that etherpad, too | 08:31 |
| wanghao | therve: okay | 08:32 |
| wanghao | the use case can be updated to BP in launchpad. | 08:33 |
| wanghao | I'll do it. | 08:33 |
| therve | Is there something in launchpad? | 08:33 |
| therve | I couldn't find it | 08:33 |
| wanghao | here is bp link: https://blueprints.launchpad.net/zaqar/+spec/service-queue-in-zaqar | 08:37 |
| wanghao | I saw your etherpad too, that's a good place we can talk about this feature. | 08:37 |
| wanghao | I'm reading it now, some stuff will be there later. | 08:42 |
| therve | wanghao, Your spec should use that name in the "Implements", FWIW | 08:57 |
| wanghao | therve: yeah, I changed the name today, must forget to update it. | 08:58 |
| therve | OK | 08:58 |
| wanghao | therve: some thoughts be the etherpad, btw. | 08:58 |
| therve | wanghao, I understand you last thought, but it needs to be much more details on how it works | 08:59 |
| therve | "user query queues with his own token" That only works for queue list. What happens when you get messages from the service queue? | 09:00 |
| wanghao | therve: en, since an arg is the queue_name when getting messages, Zaqar will find the queue is not belong to user's project but the queue is a service queue. that will allow zaqar to return messages to user. | 09:03 |
| wanghao | therve: do you think it works? | 09:04 |
| therve | wanghao, There is an order issue here. Authentication happens first. We only get to the queue controller *after* | 09:06 |
| therve | I don't understand how " Zaqar will find the queue is not belong to user's project" works | 09:06 |
| therve | It's possible it's simple, but you need to explain it to me :) | 09:08 |
| wanghao | therve: it like this: user with his own token, and get it wrong queue name, that'll pass the authentication, right. | 09:14 |
| wanghao | therve: Zaqar will return HTTP not found now. | 09:15 |
| therve | wanghao, Yeah | 09:15 |
| therve | I think I see where you're going, and I don't like it :) | 09:15 |
| therve | You mean every time we don't find a queue, we look in the service project to find if the queue exists? | 09:16 |
| wanghao | therve: em, we find the service queue with the name if it didn't belong to user. | 09:17 |
| therve | wanghao, What if you have a queue with the same name in your project? | 09:17 |
| wanghao | therve: yes, that is a problem... | 09:19 |
| therve | OK, so I'm glad we talked about it, because I don't think it's the proper solution :) | 09:19 |
| therve | wanghao, We should remove the ambiguity, and have the user says "I want queue X in project Y", and manage the authorization ourselves, I think | 09:20 |
| wanghao | therve: yes, I'll think it more. | 09:23 |
| therve | wanghao, It's a little bit tricky, because I believe if we reuse X-PROJECT-ID, the keystone middleware will reject the auth | 09:24 |
| therve | wanghao, Maybe use a new header for that purpose | 09:24 |
| wanghao | therve: emm, you mean user bring service project id in new header? | 09:29 |
| therve | wanghao, Yeah, or simply a header saying "I want a service queue" | 09:30 |
| therve | We can manage the translation server side | 09:30 |
| wanghao | therve: or we need a new api, just need project own by user and project own by service. | 09:30 |
| therve | Yeah that's an option. There are pros and cons to both approaches | 09:31 |
| wanghao | therve: yes, do you think we need to know user want which project's service queue. | 09:33 |
| therve | wanghao, I think so, unless omitting it has a benefit that I don't see | 09:35 |
| wanghao | therve: okay, that's great to discuss this with you, I'll update our ideas to spec. | 09:36 |
| therve | Thanks! | 09:36 |
| wanghao | therve: thank you too :) | 09:42 |
| openstackgerrit | Shu Muto proposed openstack/zaqar-ui master: Use common Angular view for Queues panel https://review.openstack.org/433429 | 09:45 |
| *** yanyanhu has quit IRC | 09:51 | |
| *** yanyanhu has joined #openstack-zaqar | 09:52 | |
| *** XueFeng has quit IRC | 10:17 | |
| *** yanyanhu has quit IRC | 10:23 | |
| *** stanchan has quit IRC | 10:48 | |
| *** nadya has quit IRC | 11:00 | |
| *** zhurong has quit IRC | 11:34 | |
| *** nadya has joined #openstack-zaqar | 11:52 | |
| *** amoralej is now known as amoralej|lunch | 11:56 | |
| *** wanghao has quit IRC | 12:21 | |
| *** wanghao has joined #openstack-zaqar | 12:21 | |
| *** wanghao has quit IRC | 12:21 | |
| *** wanghao has joined #openstack-zaqar | 12:22 | |
| *** wanghao has quit IRC | 12:22 | |
| *** wanghao has joined #openstack-zaqar | 12:22 | |
| *** wanghao has quit IRC | 12:23 | |
| *** wanghao has joined #openstack-zaqar | 12:24 | |
| *** wanghao has quit IRC | 12:24 | |
| *** wanghao has joined #openstack-zaqar | 12:24 | |
| *** wanghao has quit IRC | 12:24 | |
| *** wanghao has joined #openstack-zaqar | 12:25 | |
| *** wanghao has quit IRC | 12:25 | |
| *** wanghao has joined #openstack-zaqar | 12:26 | |
| *** wanghao has quit IRC | 12:26 | |
| *** wanghao has joined #openstack-zaqar | 12:26 | |
| *** wanghao has quit IRC | 12:27 | |
| *** wanghao has joined #openstack-zaqar | 12:27 | |
| *** wanghao has quit IRC | 12:28 | |
| *** wanghao has joined #openstack-zaqar | 12:28 | |
| *** wanghao has quit IRC | 12:28 | |
| *** nadya has quit IRC | 12:38 | |
| *** wanghao has joined #openstack-zaqar | 12:59 | |
| *** amoralej|lunch is now known as amoralej | 13:17 | |
| *** zhurong has joined #openstack-zaqar | 13:21 | |
| *** mpanetta has quit IRC | 13:33 | |
| *** mpanetta has joined #openstack-zaqar | 13:33 | |
| *** zhurong_ has joined #openstack-zaqar | 14:06 | |
| *** zhurong has quit IRC | 14:07 | |
| *** nadya has joined #openstack-zaqar | 14:09 | |
| *** zhurong_ has quit IRC | 14:18 | |
| *** wanghao has quit IRC | 14:49 | |
| *** rcernin has quit IRC | 15:05 | |
| *** wxy| has joined #openstack-zaqar | 15:09 | |
| *** eliqiao has quit IRC | 15:33 | |
| *** eliqiao has joined #openstack-zaqar | 15:37 | |
| *** eliqiao has quit IRC | 15:46 | |
| *** eliqiao has joined #openstack-zaqar | 15:53 | |
| *** tesseract has quit IRC | 15:58 | |
| *** nadya has quit IRC | 16:04 | |
| *** wxy_ has quit IRC | 16:24 | |
| *** wxy| has quit IRC | 16:37 | |
| *** nadya has joined #openstack-zaqar | 16:42 | |
| *** nadya has quit IRC | 16:46 | |
| *** nadya has joined #openstack-zaqar | 16:46 | |
| *** eliqiao has quit IRC | 17:01 | |
| *** eliqiao has joined #openstack-zaqar | 17:02 | |
| *** mpanetta has quit IRC | 17:12 | |
| *** mpanetta has joined #openstack-zaqar | 17:13 | |
| *** nadya has quit IRC | 17:18 | |
| *** flwang1 has joined #openstack-zaqar | 17:47 | |
| *** eliqiao has quit IRC | 17:49 | |
| *** eliqiao has joined #openstack-zaqar | 17:50 | |
| *** pcaruana has quit IRC | 17:52 | |
| *** amoralej is now known as amoralej|off | 18:07 | |
| *** nadya has joined #openstack-zaqar | 18:07 | |
| *** rbrady is now known as rbrady-afk | 18:12 | |
| *** nadya has quit IRC | 18:13 | |
| *** openstackstatus has joined #openstack-zaqar | 18:45 | |
| *** ChanServ sets mode: +v openstackstatus | 18:45 | |
| *** flwang1 has quit IRC | 19:26 | |
| *** eliqiao has quit IRC | 19:28 | |
| *** openstackgerrit has quit IRC | 19:33 | |
| *** eliqiao has joined #openstack-zaqar | 19:33 | |
| *** flwang1 has joined #openstack-zaqar | 20:24 | |
| *** Guest6666 has quit IRC | 20:28 | |
| *** Guest6666 has joined #openstack-zaqar | 20:29 | |
| flwang | therve: around? | 21:23 |
| flwang | therve: are you available for a talk about swift driver? | 21:23 |
| therve | flwang, Yep | 21:34 |
| flwang | cool | 21:35 |
| flwang | i just done some tests with the swift driver | 21:35 |
| flwang | and i got some numbers you may be interested in | 21:35 |
| flwang | generally the mongo driver will take about 50-80ms for a post request | 21:36 |
| flwang | swift driver needs to take 280-320ms for a same request | 21:36 |
| *** rbrady-afk is now known as rbrady | 21:36 | |
| flwang | i know it's a very rough test with osprofiler | 21:36 |
| flwang | but it just gives me some ideas | 21:37 |
| therve | OK, that's not too bad | 21:37 |
| flwang | so do you think if there is anything we can do to improve the performance of swift driver | 21:37 |
| therve | I'm not sure there is much we can do | 21:38 |
| therve | I'd have to check with the swift guys | 21:38 |
| therve | I don't think it supports many advanced features like pipelining | 21:39 |
| therve | Possibly if we can make auth faster that would speed things up | 21:39 |
| flwang | right | 21:42 |
| flwang | and with the swift driver, could i use the zaqar user and service tenant in the uri? | 21:43 |
| flwang | in my test, seems it doesn't work well | 21:44 |
| therve | Yeah? That's what devstack uses IIRC | 21:45 |
| therve | You're probably just missing a role | 21:45 |
| flwang | the reseller role? | 21:45 |
| flwang | ok, i will double check it | 21:46 |
| flwang | and another thing i found is, after the TTL, the object/message has gone, but you can still see it when list the queue, are you aware of it? | 21:46 |
| therve | Yeah I think it's handled in the code | 21:47 |
| therve | You mean when you do message list? Maybe I missed the check somewhere | 21:48 |
| flwang | with openstack object show you got 404, it's right | 21:48 |
| flwang | but you can see the object by openstack object list | 21:49 |
| flwang | i will test it again and open a bug if it's reproducable | 21:49 |
| flwang | therve: and as for this, https://etherpad.openstack.org/p/zaqar-service-queue do you think overall it's a useful, it's trying to do the similar thing like Nabu | 21:50 |
| therve | Yeah sure it's not like nabu got any traction :) | 21:51 |
| flwang | hah | 21:52 |
| therve | The spec needs way more details though | 21:52 |
| flwang | if we can get it in zaqar, then we can easily write a publisher in ceilometer or searchlight to forward notifications | 21:52 |
| therve | Not sure what you mean. Get nabu in zaqar? | 21:54 |
| flwang | i mean service queue | 21:55 |
| flwang | sorry | 21:55 |
| therve | OK sure | 21:57 |
| therve | Yeah that's fine, we need to flash out how that works | 21:57 |
| therve | It's still not clear to me how that would look like | 21:57 |
| flwang | in my mind, it means service user can create queue in tenant's project to forward messages | 21:59 |
| flwang | we just need to sort the auth problem out | 21:59 |
| therve | Yeah that's not what I'm thinking, so you need to convince me :) | 22:00 |
| therve | Just name collision in the user project is tricky | 22:01 |
| therve | (IE IMO it's much easier to allow user to do reads in the service project, than to allow services to do writes in the user project) | 22:02 |
| flwang | ha, but put all the queues in service tenant, we may have to handle the quota, billing stuff | 22:06 |
| flwang | though it's not a real problem for zaqar | 22:06 |
| flwang | we just hand off the problem to the others | 22:06 |
| flwang | but i can see the pros of put queues in service tennat | 22:07 |
| therve | Billing uses for infrastructure is kind of fishy | 22:07 |
| flwang | so basically, you like the idea, but you just want to keep the queues in service tenant and let tenant user can access them, right? | 22:07 |
| therve | You can still bill them for Zaqar API usage | 22:07 |
| flwang | right | 22:08 |
| therve | That seems easier to achieve to me | 22:08 |
| flwang | and that's most of the messaging services are doing | 22:08 |
| flwang | I don't really mind doing that in service tenant | 22:08 |
| therve | In any case, we need a precise plan :) | 22:10 |
| flwang | agree | 22:10 |
| flwang | i'm really happy to see you're interested in this spec | 22:10 |
| therve | I mean that's a use case I'd like to see in OpenStack since forever | 22:12 |
| flwang | hah | 22:13 |
| flwang | if so, i will do my best to get it in | 22:13 |
| *** rbrady is now known as rbrady-afk | 22:44 | |
| *** mpanetta has quit IRC | 23:27 | |
| *** stanchan has joined #openstack-zaqar | 23:30 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!