| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 00:04 |
|---|---|---|
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 00:19 |
| flwang | Eva-i: ping | 00:20 |
| Eva-i | flwang: pong | 00:20 |
| flwang | Eva-i: some questions about https://review.openstack.org/#/c/275030/8/zaqar/notification/notifier.py | 00:20 |
| Eva-i | flwang: yes? | 00:21 |
| flwang | can you remind me why do we need the change from line #43 - #50? | 00:21 |
| Eva-i | flwang: Reasons: 1. We do deepcopy to not modify original messages which will be sent to storage controller. Though it would not harm, if we modify original list, because our storage controllers parse messages correctly and will just omit the new key that we added to each message's dict. If you want, we can not do deepcopy, which affects performance. | 00:27 |
| Eva-i | 2. If there are no subscribers, let's not do any work (deepcopy and adding "queue_name" key to each message. For this reason we check the number of subscribers. | 00:29 |
| *** flwang has quit IRC | 00:29 | |
| *** flwang has joined #openstack-zaqar | 00:43 | |
| *** achanda has joined #openstack-zaqar | 00:46 | |
| flwang | Eva-i: did you see my comments? | 00:51 |
| Eva-i | flwang: have you received my IRC messages with reasons? | 00:51 |
| flwang | i saw that, but i can't be convenienced | 00:51 |
| flwang | convinced | 00:52 |
| flwang | i can't see any benefit to do the change at notifier level | 00:52 |
| Eva-i | flwang: did you see my answer in gerrit also? | 00:53 |
| Eva-i | flwang: the gerrit answer tells why I did it in notifier level. | 00:53 |
| Eva-i | *at | 00:54 |
| flwang | i can see the point, let me explain it a bit | 00:54 |
| flwang | maybe i'm wrong | 00:55 |
| Eva-i | flwang: oki | 00:56 |
| flwang | generally, we assume the messages are million level or more, then it's 1/10 million level, that means there are many times loop to add the queue name | 00:56 |
| flwang | if we add the queue_name at the task driver level, firstly, we don't need any loop | 00:56 |
| flwang | we can just add a new line after https://github.com/openstack/zaqar/blob/master/zaqar/notification/task/mailto.py#L43 for example | 00:57 |
| flwang | and as you know, we can benefit the power of sub thread instead of bugging the main thread | 00:57 |
| flwang | thoughts? | 00:58 |
| * Eva-i thinking | 00:58 | |
| *** achanda has quit IRC | 00:59 | |
| Eva-i | flwang: first of all, let's decide can we modify original list of messages or we must do deepcopy? | 01:02 |
| flwang | Eva-i: i don't think we can touch the original messages | 01:05 |
| flwang | since it's not the responsibility of notifications, even zaqar | 01:05 |
| flwang | that's my opinion | 01:06 |
| Eva-i | flwang: the messages would be parsed and delivered to the database normally | 01:06 |
| Eva-i | flwang: I mean the messages will be stored in databases without the additional key we have added. | 01:07 |
| openstackgerrit | Thai Tran proposed openstack/zaqar-ui: WIP Queues create wizard https://review.openstack.org/277590 | 01:07 |
| flwang | Eva-i: yep | 01:07 |
| Eva-i | flwang: and the user will not notice any change in messages, when it will retrieve the messages normally from Zaqar. | 01:08 |
| Eva-i | *he will | 01:08 |
| flwang | sorry, i can't follow | 01:08 |
| openstackgerrit | Thai Tran proposed openstack/zaqar-ui: Queues create wizard https://review.openstack.org/277590 | 01:09 |
| Eva-i | flwang: I mean even if the storage controller will receive messages(dicts) each one with additional key, they will store the messages normally in the database. Our added "queue_name" key will not be stored in any way in databases. | 01:11 |
| Eva-i | flwang: so maybe it's safe to modify it. | 01:11 |
| *** amitgandhinz has joined #openstack-zaqar | 01:13 | |
| *** amitgandhinz has quit IRC | 01:13 | |
| Eva-i | flwang: okay. If we decide to use deepcopy, where we will do it: in task driver level or in notifier level? | 01:13 |
| flwang | Eva-i: yep, i can see your point now | 01:14 |
| flwang | i don't think we need the deepcopy either way | 01:14 |
| Eva-i | flwang: okay, without deepcopy things are easier. I'll benchmark tasks with adding new key and without. ;) | 01:18 |
| flwang | awesome, i just want to keep things simple and no performance impact | 01:19 |
| openstackgerrit | wanghao proposed openstack/zaqar: Fix improperly LOG using in Zaqar https://review.openstack.org/277812 | 01:20 |
| Eva-i | flwang: things would be even easier if we were allowed to change notifications format in non-backward compatible way. | 01:20 |
| Eva-i | especially easier on CPU | 01:21 |
| flwang | what do you mean non-backword compatible? | 01:21 |
| *** pt_15 has quit IRC | 01:25 | |
| Eva-i | I mean to change our notifications format from this: {'body': 'message1', 'ttl': 60}. | 01:25 |
| Eva-i | To this: {'queue_name': 'somequeue', 'messages' : [{'body': 'message1', 'ttl': 60}, {'body': 'message2', 'ttl': 60}]} | 01:25 |
| Eva-i | flwang: ^^^ | 01:25 |
| *** itisha has quit IRC | 01:26 | |
| Eva-i | For example, with second format, we can notify the subscriber with more than one message in single request. And we don't need to append "queue_name" to each message dict. | 01:27 |
| flwang | interesting | 01:30 |
| flwang | personally, i like the idea | 01:34 |
| Eva-i | Perhaps it's possible to do it in backward compatible way. We just need to add an option either to Zaqar or to subscription options. So the user would be able to choose in which format he would like to receive notifications. | 01:35 |
| Eva-i | And if there will be such option, there is no need in my patch https://review.openstack.org/#/c/275030 | 01:39 |
| *** shakamunyi has quit IRC | 01:40 | |
| flwang | hmm... i don't want to introduce an option, but i would like to know the others' opinion | 01:40 |
| Eva-i | flwang: would you like to discuss it on the next meeting? | 01:41 |
| Eva-i | flwang: maybe for now we should abandon my patch | 01:41 |
| flwang | you don't have to abandon it | 01:43 |
| flwang | yep, we can discuss it on next meeting | 01:44 |
| Eva-i | flwang: so, now I make these changes to my patch: 1. stop using deepcopy 2. Move to the task level adding "queue_name" key to each message dict. | 01:46 |
| *** kgriffs is now known as kgriffs|afk | 01:46 | |
| flwang | Eva-i: good for me | 02:00 |
| *** achanda has joined #openstack-zaqar | 02:00 | |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 02:10 |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 02:13 |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 02:15 |
| openstackgerrit | Eva Balycheva proposed openstack/python-zaqarclient: Do not catch ConflictError on subscription create https://review.openstack.org/277195 | 02:59 |
| flwang | Eva-i: around? | 03:03 |
| Eva-i | flwang: yes | 03:03 |
| flwang | did you ever use the queue set metadata command? | 03:03 |
| Eva-i | flwang: in our client? or via curl? | 03:05 |
| flwang | client, i'm trying to figure out how to set the metadata after creating queue | 03:06 |
| flwang | for v2 | 03:06 |
| Eva-i | flwang: do we need it? Isn't queue metadata feature for v1 only? | 03:07 |
| flwang | we need it | 03:08 |
| flwang | v1.1 and v2 supports metadata as well | 03:08 |
| flwang | there is a bug unfortunately | 03:16 |
| Eva-i | flwang: when I fix API version number to float conversion (which our patch will fix soon), I'm able to set metadata like this: openstack --os-queues-api-version=1 queue set metadata my-lovely-queue '{"oh": "yeah"}' | 03:19 |
| flwang | i see. but for v1.1 and v2, after you created the queue, there is no way to update its metadata | 03:20 |
| flwang | you can try | 03:20 |
| Eva-i | flwang: alright, you can fix this in your patch https://review.openstack.org/#/c/267923/. | 03:23 |
| flwang | i can't | 03:24 |
| flwang | since it's a server side bug | 03:24 |
| Eva-i | flwang: aha, I see. Zaqar returns 404 error on these operations. | 03:28 |
| flwang | 404? | 03:30 |
| flwang | do you mean 204? | 03:30 |
| flwang | it returns 204 but didn't update the metadata | 03:30 |
| Eva-i | When I use curls to set metadata: 1. on API v1: 204 response code (and I see update in the database). 2. On API v1.1: 404 response code. 3. On API v2: 404 response code | 03:33 |
| Eva-i | flwang: ^^^ | 03:34 |
| flwang | Eva-i: that's correct | 03:35 |
| flwang | since we don't support metadata endpoints for v1.1 and v2 | 03:35 |
| flwang | and unfortunately, we didn't provide any other way to let the user update metadata | 03:36 |
| flwang | that's the problem | 03:36 |
| *** wanghao_ has joined #openstack-zaqar | 03:36 | |
| Eva-i | flwang: "since we don't support metadata endpoints for v1.1 and v2", but should we? Maybe it's unsupported intentionally? | 03:37 |
| flwang | it's designed intentionally, but i can't remember the reason | 03:40 |
| flwang | oh, i think i can | 03:40 |
| flwang | IIRC, we was going to remove the metadata for queue | 03:41 |
| flwang | and after that we decide to keep them | 03:41 |
| wanghao_ | flwang: hi, happy Chinese year! I have committed the client patch(https://review.openstack.org/#/c/276603/) for bug ' | 03:41 |
| wanghao_ | "pool_group" word should be used instead "pool"', if you have time, please check it, thanks. | 03:41 |
| flwang | wanghao_: happy chinese year | 03:42 |
| flwang | clicking... | 03:42 |
| flwang | wanghao_: interesting... there is no gate failure for the patch | 03:45 |
| flwang | i expect there is a failure since you're posting pool_group now instead of pool | 03:45 |
| wanghao_ | flwang: yeah, it's also shocking me. | 03:46 |
| flwang | so obviously, we need a check at here https://github.com/openstack/python-zaqarclient/blob/master/zaqarclient/queues/v1/api.py#L204 | 03:46 |
| Eva-i | flwang: wanghao: yes, happy Chinese year too you =) | 03:46 |
| flwang | i think on the server side, the pool_group is just skipped silently | 03:47 |
| flwang | Eva-i: thank you:) | 03:47 |
| flwang | wanghao_: but i prefer to do that later | 03:47 |
| wanghao_ | Eva-i: thanks Eva | 03:47 |
| flwang | wanghao_: for now, you can try to let your server patch depends on the client patch, and see if the gate will be happy | 03:47 |
| flwang | if all of them are good, then we can merge them and add the api param check for client | 03:48 |
| Eva-i | flwang: I found blueprint, but can't find a spec to read https://blueprints.launchpad.net/zaqar/+spec/api-v1.1-remove-queue-metadata | 03:48 |
| wanghao_ | flwang: yes, I added dependent-on in server patch and will check the gate. | 03:49 |
| Eva-i | flwang: seems like Zaqarians decided to remove queue metadata, because it's not much compatible with the idea of lazy queues | 03:52 |
| Eva-i | flwang: I noticed a strange thing In our API reference in wiki I read: "If auth is not enabled, clients must provide the X-Project-Id header themselves.". But in my Zaqar database I see many created queues without xprojectid, so seems like it's possible to create resources without xprojectid. Is it a bug? | 04:01 |
| *** flwang1 has quit IRC | 04:02 | |
| *** wanghao has quit IRC | 04:46 | |
| *** wanghao_ is now known as wanghao | 04:46 | |
| *** shu-mutou-AFK is now known as shu-mutou | 04:57 | |
| *** rcernin has joined #openstack-zaqar | 07:13 | |
| *** openstackgerrit has quit IRC | 07:47 | |
| *** openstackgerrit has joined #openstack-zaqar | 07:47 | |
| *** jtomasek_ has joined #openstack-zaqar | 07:50 | |
| *** rcernin has quit IRC | 07:52 | |
| *** pt_15 has joined #openstack-zaqar | 08:03 | |
| *** itisha has joined #openstack-zaqar | 08:03 | |
| *** mdnadeem has quit IRC | 08:13 | |
| *** mdnadeem has joined #openstack-zaqar | 08:13 | |
| *** achanda has quit IRC | 08:13 | |
| *** rcernin has joined #openstack-zaqar | 08:15 | |
| *** dynarro has joined #openstack-zaqar | 08:27 | |
| *** achanda has joined #openstack-zaqar | 08:37 | |
| *** mdnadeem has quit IRC | 08:43 | |
| *** mdnadeem has joined #openstack-zaqar | 08:44 | |
| *** shu-mutou is now known as shu-mutou-AFK | 09:56 | |
| *** achanda has quit IRC | 10:06 | |
| *** achanda has joined #openstack-zaqar | 10:08 | |
| *** mdnadeem has quit IRC | 10:20 | |
| *** mdnadeem has joined #openstack-zaqar | 10:20 | |
| *** achanda has quit IRC | 10:45 | |
| *** dynarro has quit IRC | 11:14 | |
| *** achanda has joined #openstack-zaqar | 11:45 | |
| *** achanda has quit IRC | 11:50 | |
| openstackgerrit | wanghao proposed openstack/python-zaqarclient: "pool_group" word should be used instead "pool" https://review.openstack.org/276603 | 12:43 |
| *** achanda has joined #openstack-zaqar | 12:48 | |
| *** achanda has quit IRC | 12:54 | |
| *** pt_15 has quit IRC | 13:15 | |
| *** itisha has quit IRC | 13:16 | |
| *** amitgandhinz has joined #openstack-zaqar | 13:33 | |
| *** amitgandhinz has quit IRC | 13:47 | |
| *** amitgandhinz has joined #openstack-zaqar | 13:47 | |
| *** sriram has joined #openstack-zaqar | 14:11 | |
| *** shu-mutou-AFK has quit IRC | 14:34 | |
| *** dynarro has joined #openstack-zaqar | 14:37 | |
| openstackgerrit | Merged openstack/zaqar: Add the missing Client-ID check in wsgi v2 https://review.openstack.org/274972 | 14:38 |
| flaper87 | flwang: ryansb vkmc What do you think about moving from a release model based on cycles and milestones to an independent one? | 14:39 |
| ryansb | I wouldn't mind having shorter freezes :) | 14:40 |
| flaper87 | I think Zaqar would benefit from smaller, isolated and more frequent releases | 14:40 |
| vkmc | me too | 14:40 |
| therve | flaper87, How so? | 14:42 |
| ryansb | Also how would associating with cycles work? | 14:43 |
| ryansb | e.g. if someone deploys a Liberty (or Mitaka or N) cloud, how would they pick a zaqar version? | 14:43 |
| flaper87 | therve: well, the project is still young and small, the 6month cycle is too long for Zaqar. There are plans and features that could be implemented and released faster | 14:44 |
| flaper87 | ryansb: No need to associate it if there's backwards compatibility, right? | 14:44 |
| therve | flaper87, Really? | 14:44 |
| flaper87 | A newer Zaqar (Mitaka) won't/shouldn't break Heat Liberty | 14:44 |
| flaper87 | therve: you seem to know something I don't so please, share | 14:45 |
| therve | flaper87, I mean, which feature in Mitaka would have benefited from an earlier release? | 14:45 |
| ryansb | there could be users that want, say, a new feature that doesn't exist yet in Zaqar 5, released at the start of Mitaka, but does in Zaqar 6, released in the middle of mitaka | 14:45 |
| ryansb | but if they're that aware, they can make that choice themselves | 14:45 |
| flaper87 | therve: I'm not saying I have one specific feature in mind. I'm saying more often releases might help Zaqar moving forward. This affects planning, of course. | 14:46 |
| flaper87 | I can tell you releasing pre-signed urls earlier would have helped in liberty, for instance | 14:46 |
| flaper87 | ryansb: they can always upgrade to Zaqar 6, I don't see the issue | 14:46 |
| ryansb | yeah, I answered my own question while writing it - nm | 14:47 |
| therve | flaper87, I'm not against the idea in principle, I don't see it making it a big difference in practice | 14:48 |
| therve | Also releases take some effort, so more releases == more effort | 14:48 |
| flaper87 | There's also 6month releases w/ intermediary releases | 14:48 |
| therve | Even if you can streamline it a bit | 14:48 |
| flaper87 | Sure but they don't in Zaqar, let's be honest. Releasing zaqar would take the same amount of effort as cutting a milestone | 14:49 |
| therve | That said, I won't work on the releases, so maybe I should just shut up | 14:49 |
| flaper87 | I guess my main point is that I don't think Zaqar needs to align with other projects for releases and it could use some extra flexibility on that side | 14:49 |
| flaper87 | I don't like the lack of prediction in not having a scheduled release | 14:50 |
| *** achanda has joined #openstack-zaqar | 14:52 | |
| openstackgerrit | Merged openstack/zaqar: Make error response better for patching flavors https://review.openstack.org/274495 | 14:52 |
| openstackgerrit | Merged openstack/zaqar: Don't return links if subscriptions are empty https://review.openstack.org/274304 | 14:52 |
| therve | flaper87, I don't get that last point. There is a scheduled release? Every six months? | 14:55 |
| flaper87 | therve: yeah, I meant to say. What I don't like about the independent release model is that there's no schedule | 14:56 |
| therve | Oh I see you're criticizing your own idea :) | 14:56 |
| flaper87 | therve: yeah :D I'm known for contradicting myself | 14:56 |
| flaper87 | :D | 14:56 |
| *** achanda has quit IRC | 14:57 | |
| *** ametts has joined #openstack-zaqar | 14:58 | |
| *** vkmc is now known as vkmc|midcycle | 15:01 | |
| *** mpanetta has joined #openstack-zaqar | 15:13 | |
| *** mdnadeem has quit IRC | 15:14 | |
| *** mdnadeem has joined #openstack-zaqar | 15:14 | |
| *** kgriffs|afk is now known as kgriffs | 15:28 | |
| *** rcernin has quit IRC | 15:33 | |
| *** wanghao has quit IRC | 15:37 | |
| *** malini has joined #openstack-zaqar | 15:42 | |
| *** malini1 has joined #openstack-zaqar | 15:44 | |
| *** malini has quit IRC | 15:47 | |
| *** rcernin has joined #openstack-zaqar | 16:00 | |
| *** csoukup has joined #openstack-zaqar | 16:18 | |
| *** rcernin has quit IRC | 16:24 | |
| *** boris-42 has quit IRC | 16:33 | |
| *** jtomasek_ has quit IRC | 16:42 | |
| *** achanda has joined #openstack-zaqar | 16:55 | |
| *** achanda has quit IRC | 16:59 | |
| *** mpanetta has quit IRC | 16:59 | |
| therve | I think zaqar devstack plugin is broken | 17:23 |
| therve | Because of the client id check | 17:23 |
| therve | If someone can have a look, otherwise I'll do it tomorrow | 17:23 |
| ryansb | therve: I'll put it on my list & ping you if I get it solved | 17:27 |
| *** davideagnello has quit IRC | 17:35 | |
| *** davideagnello has joined #openstack-zaqar | 17:36 | |
| *** boris-42 has joined #openstack-zaqar | 17:39 | |
| therve | ryansb, I believe it block Heat gate, though we don't really know because people are rechecking without looking | 17:43 |
| *** jtomasek has joined #openstack-zaqar | 17:48 | |
| *** rcernin has joined #openstack-zaqar | 17:52 | |
| *** itisha has joined #openstack-zaqar | 17:55 | |
| *** achanda has joined #openstack-zaqar | 17:57 | |
| *** dynarro has quit IRC | 17:58 | |
| *** achanda has quit IRC | 18:01 | |
| *** mpanetta has joined #openstack-zaqar | 18:04 | |
| *** achanda has joined #openstack-zaqar | 18:12 | |
| *** achanda has quit IRC | 18:15 | |
| *** malini1 has quit IRC | 18:23 | |
| *** achanda has joined #openstack-zaqar | 19:00 | |
| *** malini has joined #openstack-zaqar | 19:03 | |
| *** malini1 has joined #openstack-zaqar | 19:05 | |
| ryansb | therve: yeah, looks like it's broken (just spun up a devstack) | 19:06 |
| *** malini has quit IRC | 19:07 | |
| *** c_soukup has joined #openstack-zaqar | 19:09 | |
| *** csoukup has quit IRC | 19:13 | |
| *** c_soukup has quit IRC | 19:19 | |
| *** kgriffs is now known as kgriffs|afk | 19:20 | |
| * Eva-i looking at devstack issue too | 19:24 | |
| *** csoukup has joined #openstack-zaqar | 19:26 | |
| therve | ryansb, Eva-i: I belive the wget is simply missing the client-id | 19:27 |
| *** kgriffs|afk is now known as kgriffs | 19:27 | |
| Eva-i | therve: yes | 19:27 |
| ryansb | yep, I'm doing a patch | 19:28 |
| Eva-i | ryansb: I made patch too, but go on =) | 19:30 |
| openstackgerrit | Ryan Brown proposed openstack/zaqar: Add Client-ID to devstack health check https://review.openstack.org/278577 | 19:33 |
| * ryansb waits for the gate... | 19:36 | |
| *** achanda has quit IRC | 19:36 | |
| ryansb | ah good, only a billion changes ahead in the queue | 19:37 |
| *** achanda has joined #openstack-zaqar | 19:45 | |
| *** pt_15 has joined #openstack-zaqar | 19:48 | |
| *** eliqiao_ has quit IRC | 19:59 | |
| *** eliqiao_ has joined #openstack-zaqar | 20:02 | |
| *** jtomasek has quit IRC | 20:10 | |
| *** achanda has quit IRC | 20:16 | |
| *** achanda has joined #openstack-zaqar | 20:23 | |
| flwang | ryansb: therve: Eva-i: thanks for the quick fix | 20:32 |
| Eva-i | flwang: good morning | 20:33 |
| ryansb | ahh, so nice | 20:34 |
| ryansb | gerrit lets you edit commit message in-browser now | 20:34 |
| openstackgerrit | Ryan Brown proposed openstack/zaqar: Use Client-ID in devstack ping request to Zaqar https://review.openstack.org/278577 | 20:34 |
| ryansb | so neat | 20:34 |
| Eva-i | ryansb: gerrit even lets edit python files in-browser | 20:36 |
| ryansb | yeah, but then I can't run tests | 20:36 |
| ryansb | commit messages don't have test coverage ;) | 20:37 |
| flwang | anybody know Neerja Narayan ? | 20:37 |
| flwang | she/he registered a bp in zaqar for tempest test | 20:37 |
| *** flwang1 has joined #openstack-zaqar | 20:40 | |
| flwang | Eva-i: as for your patch https://review.openstack.org/#/c/275030 | 20:42 |
| flwang | as we discussed, we don't need the change in notifier.py to pass in the queue_name | 20:43 |
| flwang | since we can easily get it from the subscription | 20:43 |
| Eva-i | flwang: I'm sorry I forgot about it, gonna fix it | 20:44 |
| flwang | Eva-i: cool, thanks | 20:44 |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar-ui: Queues create wizard https://review.openstack.org/277590 | 20:45 |
| ryansb | flwang: if you wouldn't mind ninja-approving something, it'd be great if you could peek at https://review.openstack.org/278577 once the CI passes | 20:50 |
| ryansb | ;) | 20:50 |
| openstackgerrit | Eva Balycheva proposed openstack/zaqar: Add queue name to notifications https://review.openstack.org/275030 | 21:02 |
| *** c_soukup has joined #openstack-zaqar | 21:03 | |
| Eva-i | flwang: seems like Zaqarians decided to remove queue metadata in the past, because it's not much compatible with the idea of lazy queues | 21:04 |
| *** csoukup has quit IRC | 21:06 | |
| flwang | Eva-i: yes | 21:07 |
| flwang | Eva-i: but | 21:07 |
| flwang | i'm not a fan of lazy queues, especially now the subscriptions depends on that | 21:08 |
| openstackgerrit | Fei Long Wang proposed openstack/zaqar-ui: Queues create wizard https://review.openstack.org/277590 | 21:08 |
| Eva-i | flwang: maybe we can redesign subscriptions to make them not depend on queue existence, like messages do not depend. | 21:15 |
| flwang | Eva-i: it sounds like subscribe a non-existing object(queue)? | 21:18 |
| Eva-i | flwang: right | 21:18 |
| flwang | it's weird for me :) | 21:18 |
| Eva-i | flwang: it's like subscribing to tag | 21:19 |
| flwang | we can discuss it in Newton | 21:19 |
| Eva-i | flwang: what I think: we can't treat queues as lazy resources in one situations, and as non-lazy in other situations. We have to choose. | 21:22 |
| flwang | for now, i don't think it's a yes or no problem | 21:23 |
| flwang | user can post message without queue create | 21:23 |
| flwang | user can't create subscription without queue create | 21:23 |
| flwang | for me, it's not too bad | 21:23 |
| flwang | and i prefer to let user poke me instead of deciding by the end user | 21:24 |
| Eva-i | flwang: we have strange API =) | 21:25 |
| flwang | just like you called it unstable API, right? | 21:25 |
| flwang | you will see more strange thing in openstack | 21:25 |
| flwang | i call it 'surprise' | 21:26 |
| flwang | software is a product of compromise | 21:27 |
| Eva-i | flwang: okay, that's sad to hear | 21:27 |
| Eva-i | Is X-PROJECT-ID header required in requests to Zaqar, when noauth is used? | 21:28 |
| flwang | Eva-i: i'm sorry to show you the real world :) | 21:28 |
| flwang | Eva-i: yes, IIRC | 21:30 |
| flwang | Eva-i: anything wrong? | 21:30 |
| Eva-i | flwang: I feel I can make this world better. And in lazy resources problem I don't see where compromise is. | 21:30 |
| Eva-i | flwang: in my mongodb database I noticed a strange thing. Some of queues records don't contain project id. And is it okay that one of the queues have project id "{}", do we allow such symbols?. See http://paste.openstack.org/show/k9udhl7qqS4GOt2flTKM/ | 21:33 |
| flwang | Eva-i: hah, glad to hear you can make the world better, though i'm trying either | 21:35 |
| flwang | Eva-i: i think you need to figure out why there is no project id | 21:36 |
| flwang | based on your paste, i have no idea | 21:36 |
| *** csoukup_ has joined #openstack-zaqar | 21:39 | |
| *** malini1 has quit IRC | 21:39 | |
| Eva-i | flwang: because Zaqar allows it. I can create resources in Zaqar via curl without passing X-PROJECT-ID header. | 21:39 |
| Eva-i | flwang: if it allows it, it's a bug, right? | 21:40 |
| *** achanda_ has joined #openstack-zaqar | 21:40 | |
| *** njohnston_ has joined #openstack-zaqar | 21:41 | |
| flwang | Eva-i: i need to check, wait a moment | 21:41 |
| flwang | Eva-i: i think it only happens when you using noauth | 21:43 |
| flwang | right? | 21:43 |
| Eva-i | flwang: yes, as I said | 21:43 |
| flwang | Eva-i: i think it can be improved | 21:44 |
| Eva-i | Wiki API v1 reference tells: "If auth is not enabled, clients must provide the X-Project-Id header themselves." | 21:45 |
| *** c_soukup has quit IRC | 21:48 | |
| *** achanda has quit IRC | 21:48 | |
| *** njohnston has quit IRC | 21:48 | |
| *** sriram has quit IRC | 21:48 | |
| flwang | Eva-i: if noauth enabled, can the request work without project id? | 21:54 |
| Eva-i | flwang: yes, I tried few types of requests to Zaqar without project id. They all seem to be working. | 22:00 |
| flwang | Eva-i: if so, we may need to fix it | 22:05 |
| *** rcernin has quit IRC | 22:07 | |
| Eva-i | flwang: okay, I'm filling a bug report, so we will not forget about this problem | 22:11 |
| *** mpanetta has quit IRC | 22:15 | |
| flwang | Eva-i: thanks :) | 22:15 |
| *** boris-42_ has joined #openstack-zaqar | 22:17 | |
| *** boris-42 has quit IRC | 22:22 | |
| *** boris-42_ is now known as boris-42 | 22:24 | |
| *** ametts has quit IRC | 22:29 | |
| therve | flwang, Hey, we really need https://review.openstack.org/278577 it breaks heat gate | 22:36 |
| flwang | therve: yep, i see. i will nijia approve it | 22:38 |
| therve | Thanks! | 22:38 |
| flwang | therve: did you test it? | 22:39 |
| *** malini has joined #openstack-zaqar | 22:40 | |
| therve | No hopefully jenkins will | 22:46 |
| *** achanda_ has quit IRC | 22:49 | |
| flwang | therve: is it possible to let a trival patch of heat depends on https://review.openstack.org/#/c/278577/ ? | 22:52 |
| flwang | though i have approved it | 22:52 |
| flwang | therve: anyway, let's see | 22:52 |
| therve | flwang, Not sure the depends would pick up the devstack config | 22:53 |
| openstackgerrit | Merged openstack/zaqar: Use Client-ID in devstack ping request to Zaqar https://review.openstack.org/278577 | 22:57 |
| flwang | ok, meged | 23:01 |
| flwang | therve: can you recheck the heat patch? | 23:01 |
| therve | flwang, Yep thanks! | 23:01 |
| flwang | therve: welcome :) | 23:02 |
| *** malini has quit IRC | 23:17 | |
| *** achanda has joined #openstack-zaqar | 23:31 | |
| *** csoukup_ has quit IRC | 23:50 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!