*** betherly has quit IRC | 00:32 | |
*** jjardon has quit IRC | 00:32 | |
*** jjardon has joined #storyboard | 00:36 | |
*** betherly has joined #storyboard | 01:20 | |
*** coolsvap has quit IRC | 01:23 | |
*** coolsvap has joined #storyboard | 01:25 | |
*** jtomasek has joined #storyboard | 07:21 | |
pedroalvarez | morning! | 09:13 |
---|---|---|
pedroalvarez | I had a debugging session yesterday and I found useful things for the task '"task_assignee_changed" event is not being used when changing assignee ' of the story https://storyboard.openstack.org/#!/story/2000435 | 09:14 |
pedroalvarez | I added this debugging information: https://github.com/palvarez89/storyboard/commit/1b3e531f463822b1a2a8e302adac35ec03209a68 | 09:14 |
pedroalvarez | and got the result: http://paste.openstack.org/show/481123/ | 09:14 |
pedroalvarez | I don't know yet how to solve it, that's why I'm sharing this here. | 09:16 |
Zara | I've discovered a bug in pagination. I can already hear the sound of SotK quietly weeping. | 10:09 |
Zara | setting limit doesn't work when you're logged out, so you can't see more than 10 entries per page | 10:09 |
Zara | unless you set limit, then browse to another page, then back to the first page | 10:10 |
Zara | so it updates it, but not on the page you're looking at | 10:10 |
Zara | you can test this by going between projects and stories. | 10:10 |
Zara | (that log is a mystery to me atm, but thanks for sharing it) | 10:12 |
* SotK quietly weeps | 10:25 | |
Zara | =D | 10:26 |
SotK | pedroalvarez: I'm sorry for my nonsensical reviews yesterday :) | 10:43 |
pedroalvarez | hah, no worries, they made sense to me until I tried to fix my patches | 10:44 |
SotK | I thought "oh what about that pop-up", went and looked at it, decided the boxes weren't there, looked at the code, still didn't notice them already being there | 10:45 |
* SotK was very sleepy yesterday | 10:45 | |
pedroalvarez | I learnt the word thanks to that | 10:48 |
pedroalvarez | the word "cog"* | 10:48 |
Zara | SotK: happy to +1 https://review.openstack.org/#/c/251566/ ? I haven't merged it yet since it was updated after your last +1, but I don't think you oppose it. | 10:50 |
SotK | done :) | 10:51 |
Zara | hehe | 10:54 |
openstackgerrit | Merged openstack-infra/storyboard: Add story_title to tag events. https://review.openstack.org/251566 | 10:59 |
* pedroalvarez dances | 11:01 | |
Zara | :P | 11:01 |
* SotK wonders if Zara will +1 https://review.openstack.org/#/c/251605/ ? :P | 11:02 | |
pedroalvarez | first database migration script in my life :) | 11:02 |
*** openstackgerrit has quit IRC | 11:02 | |
*** openstackgerrit has joined #storyboard | 11:02 | |
Zara | once I've tested it :P | 11:05 |
Zara | I want to get event notificatons set up on my instance at some point; might be worth doing that first | 11:06 |
Zara | now there are some docs | 11:06 |
* SotK expects that setting up event notifications will be useful for testing a patch to fix the event notifications :) | 11:15 | |
Zara | yup | 11:16 |
Zara | I just got distracted by a rogue storyboard instance, but I'm back again | 11:16 |
pedroalvarez | hehe, note, after getting them working you might find more bugs :) | 11:18 |
Zara | that's my super-power! :D | 11:19 |
SotK | ugh, that pagination bug was a regression caused by the fix for default preferences | 11:21 |
Zara | gah | 11:21 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard-webclient: Fix setting preferences when logged out https://review.openstack.org/254702 | 11:33 |
Zara | recent events seem to be working for me now \o/ | 11:37 |
SotK | \o/ | 11:37 |
pedroalvarez | quick quick, tag a story and un-tag it :) | 11:37 |
Zara | yup, I can now see the blank event. now to look at the patches... | 11:40 |
openstackgerrit | Merged openstack-infra/storyboard: Document manually setting up notifications https://review.openstack.org/253591 | 11:42 |
Zara | while looking at tags patches, have found that subscriptions sometimes seem to be removed when recentevents list changes. trying to work out what's going on. I don't think it's related to the tags patches. | 12:05 |
Zara | removing a tag can remove a subscription | 12:07 |
Zara | whyyyyyyyyyyyyyyyyy | 12:07 |
pedroalvarez | that is something I haven't hit.. | 12:08 |
pedroalvarez | I wonder if this is something you can reproduce elsewhere, (openstack's or sotk's instances) | 12:09 |
Zara | okay, removing a tag stops a subscription from displaying (that is, it shows an empty star). but tag removal still comes up in the list of recent events, suggesting that either the user is still subscribed, or recentevents isn't paying attention to a user's subscriptions | 12:10 |
Zara | this happens on master branch, so am confident it's not to do with the tags template patch, anyway | 12:12 |
Zara | idk why it would be but am now certain :P | 12:12 |
pedroalvarez | failed to reproduce it so far | 12:13 |
Zara | yeah, doesn't seem to be an issue on stories list on storyboard.openstack.org. going to see if it's just a problem with the dashboard, otherwise it might just be my instance (though still not sure why) | 12:15 |
Zara | hm, happens on my instance on the story list, so suggests it's different to storyboard.openstack.org | 12:16 |
Zara | my instance is running master... | 12:17 |
Zara | so either some strangeness in my db, or possibly an issue from a patch merged recently that hasn't made it to production yet. | 12:19 |
SotK | o.O | 12:31 |
SotK | you always find the weirdest things | 12:31 |
Zara | hm. on my instance, adding a tag PUTs the tag, then GETs events and the relevant user. deleting the tag deletes the tag and GETs the events. On storyboard.openstack.org, it's the same, but it doesn't GET the user. | 12:31 |
Zara | or maybe it does, but in a different order | 12:32 |
SotK | my test instance doesn't GET the user | 12:33 |
Zara | http://paste.openstack.org/show/481153/ | 12:34 |
Zara | I have user-related errors when I search for things, since I don't think I added fulltext stuff properly for users | 12:34 |
Zara | so maybe it's related? | 12:34 |
SotK | oh hm, it did lose the subscription star though | 12:34 |
Zara | hm, permanently? sometimes it's just slow to load | 12:35 |
Zara | but if you refresh and it's gone, then :/ | 12:35 |
SotK | yeah, permanently | 12:36 |
Zara | oh, look, it's that time of the day where I make tea and hide. | 12:36 |
SotK | but I'm still subscribed to it | 12:37 |
SotK | (at least, I'm still getting notified of changes/new comments) | 12:38 |
SotK | or am I? | 12:38 |
pedroalvarez | I'm still failing to reproduce it :/ | 12:38 |
SotK | aha, I got the notification because I'm subscribed to the project too | 12:39 |
SotK | pedroalvarez: I just added a tag to a story I was subscribed to, deleted the tag, then refreshed and the star next to the title was empty | 12:39 |
Zara | ah, maybe that's why I was getting notifications | 12:42 |
Zara | I definitely don't get the issue on storyboard.openstack.org. if it starts, we'll know it's from a recently-merged patch. | 12:44 |
Zara | if not, yet another mystery? | 12:44 |
pedroalvarez | still can't reproduce :S | 12:45 |
pedroalvarez | we could try to reproduce the error without that patch merged | 12:45 |
* SotK has reproduced it without pedroalvarez's patch | 12:45 | |
pedroalvarez | phew | 12:46 |
Zara | \o/ though is that also rolling back the db? I don't know why it would be an issue but I'm already confused. | 12:46 |
SotK | yep | 12:48 |
Zara | gah. can we see what patch production is currently running from? | 12:49 |
Zara | maybe we have to go back further? | 12:49 |
Zara | otherwise maybe we've both done something strange with our databases | 12:49 |
* SotK tries again and this time it does GET the user | 12:52 | |
SotK | and doing it again doesn't | 12:53 |
* SotK shrugs | 12:53 | |
Zara | \o/ | 12:53 |
Zara | I can't say that it definitely does it every time for me | 12:53 |
Zara | it was an immediate difference I spotted, might not be relevant at all | 12:54 |
Zara | looks like it's GETting the same things on my instance whether or not there's a tag. maybe we should investigate what goes on when a tag is deleted... | 13:01 |
* SotK is doing so | 13:01 | |
Zara | :P | 13:01 |
SotK | I think I've discovered the cause of the issue | 13:15 |
* SotK double checks | 13:16 | |
Zara | :0 | 13:17 |
SotK | indeed I have | 13:17 |
Zara | oh my! | 13:18 |
SotK | the worker is deleting the subscription because it sees that the http method is DELETE | 13:20 |
Zara | omg | 13:20 |
SotK | I'm not sure why it isn't happening for pedroalvarez or in production, maybe we are doing something weird when running it manually | 13:21 |
pedroalvarez | maybe I need to restart the workers...? | 13:21 |
pedroalvarez | not sure.. | 13:21 |
Zara | and here I was going to joke that it was the docs patch that caused it. | 13:21 |
pedroalvarez | nope, not happening here | 13:22 |
*** tlbr has quit IRC | 13:22 | |
* SotK investigates some more | 13:23 | |
Zara | deleting a task doesn't seem to affect the subscription | 13:24 |
Zara | oh, actually, hang on, maybe it would for the related project | 13:25 |
Zara | no, still seems fine | 13:26 |
SotK | nope, that would do the right thing looking at it (delete the subscription to the task, if one exists) | 13:26 |
Zara | hm, my worker is reporting errors | 13:27 |
SotK | can you paste them somewhere please? | 13:27 |
Zara | http://paste.openstack.org/show/481160/ | 13:27 |
SotK | hmm, I've never seen that | 13:28 |
SotK | what did you do that caused it? | 13:28 |
Zara | I'm not positive, might have been from adding and deleting a task in a subscribed resource, going by the timestamp | 13:28 |
Zara | will try to narrow it down | 13:29 |
* SotK just did that without issue... weird | 13:29 | |
Zara | yeah, seemed to work fine | 13:31 |
Zara | and now I can't reproduce the error | 13:31 |
Zara | managed to get it again; think it's something to do with navigating to a page before another page has loaded properly (maybe) | 13:34 |
Zara | I didn't get it when refreshing each page bfore doing anything | 13:34 |
SotK | So the deletion problem is that the notifications hook intercepts the DELETE request and creates an event with method = 'DELETE', which is fine, except that the path in the request for deleting tags is "/v1/stories/1/tags". | 13:36 |
SotK | That gets parsed such that the affected resource is a "story" with id 1, and the affected sub_resource is a "tag" with id None... | 13:37 |
SotK | and the check for handling deletions (which is done so you don't end up with subscriptions to non-existent things) is `if method == 'DELETE' and not sub_resource_id:` | 13:38 |
Zara | haha | 13:38 |
SotK | tl;dr: the worker is taking `DELETE /v1/stories/1/tags` to mean the story was deleted, and unsubscribing from it | 13:38 |
Zara | I wondered if deleting worked like that but didn't get why it would only be the case there, since I was missing the notification-hook-intercepting piece of the puzzle | 13:39 |
SotK | me too, since the creation of the "tag deleted" event is handled by a different thing | 13:40 |
SotK | which creates an event with method='POST' | 13:40 |
Zara | (btw, I think the weird error I got might involve a race-condition, though I'm not sure. I only seem to be able to reproduce it by accident) | 13:41 |
pedroalvarez | I wonder if workers are not being upgraded on my instance | 13:42 |
SotK | I guess I should send a patch to fix it | 13:55 |
Zara | probably wise. | 13:55 |
Zara | I got thoroughly distracted from testing the tag templates patch | 13:55 |
Zara | well, tags have now had a thorough testing. | 13:56 |
SotK | \o/ | 13:57 |
SotK | hah, I've seen that error now | 13:59 |
Zara | :D | 14:01 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Don't delete subscription if sub_resource is set https://review.openstack.org/254770 | 14:06 |
* pedroalvarez destroys his insance in another attempt to reproduce the error | 14:08 | |
Zara | patch doesn't seem to be fixing the problem for me. Should I restart the worker? | 14:10 |
SotK | yes :) | 14:10 |
* Zara will try that anyway | 14:10 | |
Zara | hahah | 14:10 |
Zara | ah, all seems well | 14:10 |
Zara | thanks | 14:10 |
openstackgerrit | Merged openstack-infra/storyboard-webclient: Add templates for tags added/deleted subscription events https://review.openstack.org/251605 | 14:11 |
Zara | \o/ | 14:13 |
*** tlbr has joined #storyboard | 14:14 | |
SotK | Zara: I've never had to refresh the story page when changing the preferences, unless you mean they won't auto-update if you change them on the preferences screen in a different window | 14:31 |
Zara | yeah, sorry, that's what I meant | 14:32 |
Zara | I had it open in a different tab | 14:32 |
Zara | just mentioned it because I spent a brief moment wondering why it wasn't working. | 14:33 |
openstackgerrit | Merged openstack-infra/storyboard-webclient: Add checkboxes for tag events on preferences page https://review.openstack.org/254162 | 14:45 |
pedroalvarez | triple dance today | 14:49 |
openstackgerrit | Zara proposed openstack-infra/storyboard-webclient: Align comment timestamps on the right on dashboard https://review.openstack.org/254802 | 14:52 |
Zara | I have patches from a couple of weeks back, though not many :P https://review.openstack.org/#/q/owner:zara.zaimeche%2540codethink.co.uk+status:open,n,z | 14:54 |
pedroalvarez | damn, fresh deployment, and I still can't reproduce the removing-subscriptions error | 14:58 |
SotK | :/ | 15:02 |
Zara | are the rabbitmq settings different? gah | 15:02 |
SotK | pedroalvarez: can you add some debugging code to the handle function in storyboard/plugin/subscription/base.py, to print out the method, path, and resource of each request? | 15:03 |
pedroalvarez | SotK: http://paste.openstack.org/show/481178/ | 15:10 |
SotK | hmm, I wonder if its because of the "Attempted to send payload with no destination resource." thing | 15:12 |
SotK | does the regex on line 126 of storyboard/notifications/notification_hook.py match the path properly? | 15:14 |
pedroalvarez | you nailed it | 15:16 |
pedroalvarez | so, I found a solution some days ago for the "Attempted to send payload with no destination resource." | 15:16 |
pedroalvarez | but given that I don't know if that is really a solution, I didn't send a patch for that | 15:17 |
pedroalvarez | if I apply the solution, the bug happens | 15:17 |
SotK | I wonder if that issue is happening in production too | 15:18 |
pedroalvarez | here the log: http://paste.openstack.org/show/481179/ | 15:18 |
SotK | since the api is at https://storyboard.openstack.org/api/v1/ there too | 15:19 |
SotK | that log is the same as I see on my test instance | 15:19 |
pedroalvarez | and here the fix: http://paste.openstack.org/show/481181/ | 15:20 |
pedroalvarez | but I guess that the fix will break your development environments | 15:20 |
SotK | indeed it will | 15:20 |
pedroalvarez | I also wonder, what is that for :) | 15:21 |
pedroalvarez | given that everything works nice without it | 15:21 |
pedroalvarez | heh | 15:21 |
pedroalvarez | and if it's needed maybe we can: | 15:22 |
pedroalvarez | url_pattern = re.match("^(\/api)?\/v1\/([a-z_]+)\/?([0-9]+)?" | 15:22 |
pedroalvarez | not sure if that is an expression here, tbh | 15:22 |
SotK | pedroalvarez: what is what for? | 15:25 |
pedroalvarez | I.. I don't know how to even call it. I mean, the code that is broken in production environments, and working in development environment | 15:27 |
pedroalvarez | this notification_hook | 15:27 |
SotK | it gets called each time there is a request to the API, and if its a PUT, POST, or DELETE request it publishes a message to rabbitmq describing the request, which the event worker can then handle however it wants to (eg. creating a subscription_event, or sending an email) | 15:30 |
SotK | the notifications work in production despite that not working because I think all of the timeline event changes also publish a message with information about the event, and that is what actually gets used to create subscription_events AIUI | 15:31 |
* SotK isn't sure how much sense that made | 15:32 | |
pedroalvarez | makes some sense :) thanks for the explanation | 15:32 |
*** tlbr has quit IRC | 15:41 | |
pedroalvarez | I can say now that your patch works | 15:43 |
Zara | \o/ | 15:48 |
* pedroalvarez fails to fix the regex | 15:56 | |
pedroalvarez | :) | 15:56 |
SotK | what went wrong? | 15:56 |
pedroalvarez | my ability of doing regexes | 15:57 |
pedroalvarez | my suggestion didnt work | 15:58 |
pedroalvarez | url_pattern = re.match("^(\/api)?\/v1\/([a-z_]+)\/?([0-9]+)?" | 15:58 |
pedroalvarez | also, I'm not sure if it makes sense | 15:58 |
openstackgerrit | Merged openstack-infra/storyboard: Don't delete subscription if sub_resource is set https://review.openstack.org/254770 | 15:58 |
*** tlbr has joined #storyboard | 16:05 | |
*** tlbr has quit IRC | 16:10 | |
*** tlbr has joined #storyboard | 16:13 | |
SotK | pedroalvarez: I managed to get it working with that regex, but I had to increase all the indices used later in the function by 1 | 16:21 |
pedroalvarez | let me know if I can help | 16:23 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Parse path correctly when it begins with /api https://review.openstack.org/254859 | 16:46 |
*** openstackgerrit has quit IRC | 16:47 | |
*** openstackgerrit has joined #storyboard | 16:47 | |
pedroalvarez | oh yes, now I understand | 16:49 |
*** yarkot has joined #storyboard | 17:26 | |
*** yarkot has quit IRC | 17:58 | |
*** openstackgerrit has quit IRC | 19:02 | |
*** openstackgerrit has joined #storyboard | 19:03 | |
*** yarkot has joined #storyboard | 20:18 | |
*** yarkot has quit IRC | 20:49 | |
*** jtomasek has quit IRC | 21:03 | |
*** yarkot has joined #storyboard | 21:24 | |
*** yarkot has quit IRC | 22:39 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!