mozilla :: #mozreview

20 Mar 2017
04:23pulsebotCheck-in: - byron jones - vcssync: add "no bug" to rust revendoring commit message (bug 1348714)
08:40zalunDoes that mean MozReview has done all what was needed to post to autoland?
11:17freddybError publishing: Bugzilla error: You can't ask u540588@disabled.tld because that account is disabled.
11:17freddyb^--- I get that when trying to publish a review request draft that was submitted via mozreview
11:18freddybIt was neither me, nor ckerschb (who I had as reviewers), because I removed them individually. I have now removed _all_ reviewers and it works, but I have no idea who that user id was pointing to.
11:44AutomatedTesterare we any closer to understanding the 500 errors when pushing to mozreview
12:23zalunAutomatedTester: I've got no info for you :(
13:24globAutomatedTester: i've been working with webops and devservices about it. it's so weird
13:25AutomatedTesterglob: weird in what way?
13:26AutomatedTesterfor me its 2/3 pushes result in a 500
13:26globAutomatedTester: 500s without errors in the logs, accompanied by 500s in the access log that were sent over the wire as 200
13:26glob(for two different requests)
13:26AutomatedTesterok... that is weird
13:26globthat it's intermittent makes it all the more confusing
13:27AutomatedTesterare we using our own infra for this or is it EC2?
13:27AutomatedTesteris it the same nodes (or do we only have 1)
13:27globone node only
13:28glob(and no staging, making it all the more fun to poke at)
13:28* AutomatedTester is wondering if we have memory chips/HDD with bad bits
13:28globit's a VM
13:28globso that's possible, but unlikely
18:05nalexanderAnd again. My network seems to be flaky, but this is frustrating.
18:05nalexanderBecause the UI is unusable after this happens -- all the buttons are disabled.
18:05mcotenalexander: you're referring to the 500s on push?
18:06nalexandermcote: no, "Http error 0" using the reviewboard UI.
18:06mcoteI think I might have missed your earlier message
18:06nalexandermcote: seems to be XHR problems due to network intermittent drops? But the UI requires a hard-refresh to recover, which is un-nerving.
18:06nalexanderAnd lost my message at least onece.
18:07nalexanderFuck, now I get this infinite spinner again.
18:07mcoteI haven't heard of this problem before
18:07nalexanderI hope this is just network badness in the YVR office.
18:08nalexanderIt's taking 10+ seconds of "Saving" spinner to succeed or fail.
18:08mcotethat's really weird
18:08mcotebugzilla issues maybe?
18:08mcotebut then other people would be reporting this
18:09nalexanderFuck, it times out and hten the UI is disabled!
18:09nalexanderSo I can't try again without reloading, which always makes me worry about data loss.
18:09mcoteyeah if there's a serious error a hard refresh is usually required
18:09mcotesomething weird in Review Board exception handling
18:09mcoteis this when you are publishing a review?
18:10nalexandermcote: yes, when publishing a review. It just silently fails and disables the UI (so I can't try to resubmit again).
18:10mcoteokay so data shouldn't be lost then
18:10mcoteif it didn't finish, it'll still be in draft form
18:12nalexanderIt seems to have preserved my data all abut the first time, when my connection hard dropped.
18:12nalexanderSo that's mostly good!
18:12mcoteI see a log enry for you pushing review 124044
18:13mcotenalexander: seems to have worked, if this is the review you're talking about:
18:13nalexanderI was eventually able to post them both, but it took several tries for each. I think it was client-side network blips at the root.
18:15mcotehm yeah I guess so. doesn't look like those requests made it to the server, at least, I see no related logs
18:35mcotelooks like reviewbot has been wedged for some time
18:35mcotegps: ^
18:35mcotejust fyi, doesn't really matter from pulse's perspective
18:38pulsebotCheck-in: - Piotr Zalewa - mozreview: skip all Selenium tests (bug 1348725) r=mcote
18:47aswanarg, i was in the midst of responding to a bunch of review comments and i clicked over to the diff, then i came back to the "Reviews" page and all my comments are gone
18:48mcotehm if you clicked "ok" they should save as drafts
18:49mcotethey may not autosave, though; we had to hack that into the publish-review dialog ourselves
18:58aswanthen shouldn't it have the unbeforeunload warning before i navigate away?
19:08mcoteyeah that's weird
20:24dklmcote: while pushing a 2nd set of changes up for review on reviewboard it is hanging after "(published review request 121484)" any ideas?
20:25erahmIs there some sort of "suggest reviewers" feature in mozreview?
20:33mcoteerahm: no, there's a bug on file, but we're not entirely sure the best route to do it
20:33mcotewe'll figure something out for the new conduit stuff
20:33erahmhmmm and there's no autofill from bugzilla names I guess
20:34mcotethere should be
20:34mcotedkl: hm did it finish?
20:34dklmcote: it did but it was 'remote: Connection to closed by remote host.'
20:34mcoteerahm: there's autofill in the UI
20:35mcotedkl: that's interesting
20:35dklmcote: when doing reviews, do you have a separate repo checkout that you use for reviews directly from
20:35erahmmcote: I mean in the maps to a users bugzilla email address sense, for instance if I type in ehsan it's not clear which one is the one I want
20:35dklor do you just have the main checkout and then pull down the review commits from into your current work dir
20:37mcoteerahm: ah, so you can either use his email, or his full name. also anyone who uses the :ircnick convention in their full name in Bugzilla will have that set as their username in Review Board
20:37mcoteerahm: so you can actually just type "ehsan" and hit enter
20:38mcotedkl: no, I just pull them down as extra heads and bookmark them appropriately
20:38dklok so all from a single checkout
20:39dklany special wording you prefer such as review-1234 or similar?
20:39mcoteit's entirely up to you; it's only on your local side
20:39dkljust curious what others workflows look like
20:41mcoteI don't do much reviewing so I'm probably not a good person to ask :)
20:41mcotesmacleod, mars?
20:43marsreading back
20:44marsdkl, I keep a separate checkout of mainline for pulling reviews just to keep my main dev tree focused on my own work (and tools).
20:44dklmars: make sense
20:45smacleoddkl: I pull down into my normal checkout, unless I have a reason not to. I know that is super nebulous though
20:45smacleodI don't name them with bookmarks, I just keep an unnamed head
20:46smacleoddkl: if you're not using "hg wip" you should be, and if you don't understand phases you should read " hg help phases"
20:46smacleodIt will really help with workflows
20:46smacleod(and is important for conduit work)
20:46dklsmacleod: i am using wip
20:47smacleoddkl: making sure it's in color helps too
21 Mar 2017
No messages
Last message: 5 days and 14 hours ago