mozilla :: #conduit

18 Apr 2017
09:59janxsmacleod: Hello! I hope you're well, and just had a quick question: in your static-analysis-review-bot design, you had "analysis" step and "processing-and-feedback-posting" step in two different TaskCluster tasks. What was the reason again for not doing all steps in once task? Wouldn't the latter be simpler to implement?
09:59janx(actually "feedback-processing-and-posting" would be a better name for the second task)
14:37smacleodjanx: The processing and posting was actually going to be a service, not a task in my design
14:38smacleodjanx: but the reason I thought it should be separate is so the standard task cluster tasks used for try and m-c builds could be re-used and run sooner
14:41janxsmacleod: ok, thanks! what was the benefit of making it a service instead of a task? and did you have an idea of what the analysis task could be re-used for? (maybe for various try task-sets that developers might want to run on treeherder?)
14:42smacleodjanx: just what I was used to, and it would mean if the place it needed to post was down it could wait until up then start processing again, instead of a ton of taskcluster jobs just retrying over and over until it was up
14:43smacleodjanx: I was starting out with tasks that were already defined, mainly the lint ones, and figured any task for in review should be runnable on try as well
14:49smacleoddavidwalsh: did you not put your bmoauth.php update in your gist?\
14:49janxsmacleod: I agree, thanks a lot for your replies! We ended up choosing a straight (non-try) task, to make it faster and easier to implement, but I agree that in the future we should work on integrating try jobs with the review tools, allowing manual triggers and also automated + mandatory try jobs to ensure sanity of patches before they land
14:50janx(I'd like to say "when things settle down a bit", but I guess they never really settle)
15:10davidwalshsmacleod: https://gist.github.com/darkwing/dba72ee1308b30ac886f666c6bb74c33
15:10davidwalshJust needed to make an update
16:02imaduemeMy current schedule:
16:02imadueme12p - 1p: Brown bag all hands meeting
16:02imadueme1p - 2p: Standup phabricator docker image, review new diagrams
16:02imadueme2p+: Story meeting
16:11pulsebotCheck-in: https://hg.mozilla.org/automation/conduit/rev/786a160a02ad - Mris Fogels - conduitdemo: add phabricator demo (bug 1355527)
16:16circleci-botSuccess: davehouse's build (#75; push) in mozilla-conduit/conduit (master) -- https://circleci.com/gh/mozilla-conduit/conduit/75?utm_campaign=chatroom-integration&utm_medium=referral&utm_source=irc
16:23marssmacleod, davidwalsh, dkl, imadueme, ^ that's the first revision of the demo container. The commit message explains the /etc/hosts file. arc diff works from the demo container, and IIRC I had arc diff running from my workstation, too.
16:24marsimadueme: on first run you need to access the web interface to create an admin user. after you create an admin user you can run 'arc diff'.
16:24imaduemecool, I'll check it out and let you know how it goes
16:27smacleodmars: nit: you should end your files with newlines
16:39marssmacleod: ? I'm surprised the editor didn't
16:42ckolosI think I know the answer to this question, but do any of you mind if I tear down the existing conduit-autoland dev instance running in AWS or will someone cry if I do so?
16:42marsturns out it's a per-project setting I had missed
16:43marsckolos: I'm not aware of any plans for it. smacleod, ^ do you have any plans?
16:44ckoloskk
16:44ckolossilence assumes acceptance ;)
16:44smacleodckolos: autoland in the new world will be that same image, and function similarily, but we're not using it right now
16:44smacleodso don't delete the stuff to bring it up heh
16:45ckolosoh, no, just the existing running instances.
16:45smacleod
16:45* ckolos wishes his irssi fonts supported emoji :)
17:43armenzgwhen's mars back?
17:47marsback now
17:51imaduemephabricator_1 | Set 'mysql.user' in local configuration.
17:51imaduemephabricator_1 | + set +x
17:51imaduemephabricator_1 | Set 'mysql.pass' in local configuration.
17:51imaduemephabricator_1 | + ./bin/config set phabricator.base-uri http://phabricator.dev:7788/
17:51imaduemephabricator_1 | Set 'phabricator.base-uri' in local configuration.
17:51imaduemephabricator_1 | + ./bin/storage upgrade --force
17:51imaduemephabricator_1 | MySQL Credentials Not Configured
17:51imaduemephabricator_1 |
17:51imaduemephabricator_1 | Unable to connect to MySQL using the configured credentials. You must
17:51imaduemephabricator_1 | configure standard credentials before you can upgrade storage. Run these
17:51imaduemephabricator_1 | commands to set up credentials:
17:51imaduemephabricator_1 |
17:51imaduemephabricator_1 | phabricator/ $ ./bin/config set mysql.host __host__
17:51imaduemephabricator_1 | phabricator/ $ ./bin/config set mysql.user __username__
17:51imaduemephabricator_1 | phabricator/ $ ./bin/config set mysql.pass __password__
17:51imaduemephabricator_1 |
17:51imaduemephabricator_1 | These standard credentials are separate from any administrative credentials
17:51imaduemephabricator_1 | provided to this command with __--user__ or __--password__, and must be
17:51imaduemephabricator_1 | configured correctly before you can proceed.
17:51imaduemephabricator_1 |
17:51imaduemephabricator_1 | Raw MySQL Error: Attempt to connect to root@phabdb failed with error #2002:
17:51imaduemephabricator_1 | Connection refused.
17:51imaduemeconduitdemo_phabricator_1 exited with code 1
17:51imaduemesorry!
17:51mars:)
17:51imaduememeant to be a snippet
17:51imaduemehttps://irccloud.mozilla.com/pastebin/ov7t3S62/
17:51imaduememars: looks like phab can't connect to the db?
17:52marsimadueme: could be a timing issue, mysql container startup coordination it troublesome
17:52marsimadueme: try running 'up -d' on just the mysql container first, so it has time to start. Then 'up' the rest.
17:53imaduememars: that worked, thanks!
17:53marscool
17:54marsI was hoping to add a 'mysql ready?' busy-wait loop at some point to the scripts.
17:57imaduememars: works like a charm :)
17:57marselegant answer to that problem here: http://stackoverflow.com/a/29793382
18:00smacleodmcote: fine to join?
18:00marsdavidwalsh, smacleod: do we need a repo-level project for building and running phabricator extensions? I added the app to the demo, but we probably don't want to use the demo for extension development.
18:01smacleod"the app"?
18:01marsphabricator
18:01* smacleod is unsure
18:02davidwalshsmacleod: I'm starting to think we need to inherit from PhutilOAuthAuthAdapter; looks like it has a bunch of stuff we probably need and PhutilAuthAdapter seems too simplistic
18:02zalunI think extensions might be outside of the app
18:02smacleoddavidwalsh: like what?
18:02zalunit just need to be connected
18:02smacleodwe aren't using OAuth
18:02mcotedkl: /win 39
18:02mcoteer
18:02davidwalshI suppose we could just rip out a bunch of it
18:02mcotedkl: meeting?
18:02smacleoddavidwalsh: can you give me specific examples?
18:48ckolosThe lesson here is that you should never ask ckolos anything. Apologies all around if I bikeshedded too mcuh
19:03zalunIt looks like all of the fields in Differential are created in this way
19:04smacleodzalun: I'm not saying fields should be created differently, I'm wondering if a field is even needed or the right way to do this
19:04smacleodI understand it is *a* way
19:06zalunI haven't found any general field which we could use
19:07imaduemeI have mars' phab image running and the smacleod's diagrams make sense. I'm going for lunch now, but if there is any area where I need to get caught up let me know. I think I can be really valuable on the autoland stuff, both ui and api. Back in like 45
19:08zalunthere is a diff info, but that would require (?) modifying arc
19:09smacleodhav eyou looked at anything else built into phabricator? setting diff info with herald or something?
19:10ckolosfrom the phabricator security policy: "We receive many reports (significantly more than 50%) from researchers who do not read these rules. To prove that you've read and understood these rules, please include the word "mongoose" somewhere in your report. If you do not, your report will be closed as "Not Applicable"."
19:10ckolosso be forewarned, the codeword is mongoose.
19:13davidwalshWith a callback url of "http://bugzilla.mozilla.org/auth.cgi?callback=http://mydocker.com:62080/auth&description=MozillaPhabricator", I'm logging into Bugzilla but then seeing "The callback URI uses an illegal protocol: http. Only http and https are allowed."
19:14davidwalshdkl: smacleod: Does that look familiar for a basic request like this?
19:17smacleoddavidwalsh: bugzilla forces https for the auth protocol :(
19:17smacleoddavidwalsh: you should speak to dylan on how to deal with that
19:28smacleoddkl: thoughts on my reply to your etherpad?
19:42dklsmacleod: sorry. commented again in the etherpad
19:43dklmars: is it just me, or was it odd that i needed to manually run 'bin/phd start' in the demo docker container?
19:43* dkl looks at startup scripts
20:23zalunsmacleod: I'll take a look into herald tomorrow, thanks for the idea
21:33marsdkl: I don't know. That does sound odd.
21:34marsI didn't dig into php-cli except to get the arc script working. I didn't look into how the php interpreter is actually wired up to run the cli.
21:38dklmars: these changes gets me down to 0 setup warnings https://reviewboard.mozilla.org/r/131392/
21:39dklmars: also noticed we are not opening up a ssh port on the phab container for arc to connect to. If we did we would not need to shell into the container first.
21:39dkli can look into that later tonight
21:40dkllet me know what you think. gotta run
22:06marsdkl|afk: I didn't need to use ssh to upload patches with arc diff from the demo container to the phabricator container.
22:11marsdkl|afk: is there some part of the workflow that I'm missing that needs ssh?
19 Apr 2017
No messages
   
Last message: 7 days and 2 hours ago