16:30:47 <djmitche> #startmeeting weekly
16:30:47 <bb-supy> Meeting started Tue May 31 16:30:47 2016 UTC and is due to finish in 60 minutes.  The chair is djmitche. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:30:47 <bb-supy> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:30:47 <bb-supy> The meeting name has been set to 'weekly'
16:30:54 <djmitche> #topic Introduction
16:30:58 <djmitche> https://titanpad.com/buildbot-agenda
16:30:59 <rutsky> Huh, I'm interested to see presentation about those
16:31:07 <djmitche> he just briefly mentioned them
16:31:13 <verm__> i'm here hi everyone!
16:31:19 <djmitche> I assume there are PEPs
16:31:23 <djmitche> :)
16:31:41 <djmitche> #topic Development Week in Review
16:32:42 <djmitche> looks like PRs this week are mostly slave -> worker changes?
16:32:56 <djmitche> am I missing other important changes this week?
16:32:57 <rutsky> I think so
16:33:19 <djmitche> ok
16:33:27 <djmitche> so let's talk about that :)
16:33:29 <djmitche> #topic Bug 2340 update
16:34:01 <rutsky> master/worker protocol changes fully implemented - there are PRs awaiting review (plus need few more tests)
16:34:17 <rutsky> also I updated docs accordingly
16:34:35 <djmitche> https://circle-artifacts.com/gh/rutsky/buildbot/213/artifacts/0/home/ubuntu/buildbot/master/docs/_build/html/developer/master-worker.html
16:34:42 <djmitche> ^^ new docs for the master/worker protocol
16:34:53 <rutsky> so after all PRs will be reviewed/fixed/merged most significant part of bug 2340 will be complete
16:35:21 <rutsky> master/worker protocol hasn't changed significantly, actually
16:35:32 <rutsky> at least due to bug 2340
16:35:37 <djmitche> should the remote_getWorkerInfo docs mention a version key?
16:35:49 <rutsky> but introduction of local worker make that doc a bit obsolete --- I tried to fix that
16:36:08 <rutsky> djmitche: good point!
16:36:13 <djmitche> :)
16:36:21 <rutsky> version and list of supported commands should be included
16:36:38 <djmitche> #info after all PRs will be reviewed/fixed/merged most significant part of bug 2340 will be complete
16:36:45 <djmitche> so what will be left after that point?
16:37:22 <bb-github> [13buildbot] 15rutsky commented on issue #2246: TODO for myself from IRC meeting:... 02https://git.io/vrdbg
16:37:46 <rutsky> in codebase: tiny unfixed strings will left, such as some docstrings, manpages etc
16:38:04 <djmitche> ok, wow :)
16:38:07 <rutsky> I need to carefully review left "slave" strings (there are about 2.5K of them)
16:38:28 <djmitche> #info after that all that remains are removals of minor uses of the word "slave" (docstrings, comments, manpages, etc.)
16:38:30 <rutsky> but bug2340 also includes non-codebase changes, such as Trac/Wiki, metabbot, etc
16:38:56 <rutsky> I haven't touched those parts yet, but I don't expect a lot of work there
16:39:01 <djmitche> right
16:39:22 <djmitche> that's awesome, so we're very close :)
16:39:29 <bdbaddog> Great work!
16:39:54 <rutsky> yes, after merging protocol changes bug 2340 should not block 0.9 release anymore
16:40:05 <djmitche> oh, very important point, thanks
16:40:19 <rutsky> sure we need to fix all docs too, but this can be don't even after 0.9 release
16:40:23 <djmitche> #info once protocol changes (current PRs) are merged, bug 2340 no longer blocks an 0.9.0 release
16:40:32 <djmitche> right
16:40:37 <djmitche> the remaining changes won't be compatibility issues
16:40:53 <djmitche> ok, let's talk trac
16:41:03 <djmitche> #topic Trac rejecting logins
16:41:04 <rutsky> ok
16:41:17 <verm__> logins or account creation?
16:41:28 <verm__> as of now there have been two cases that i know of where users couldn't make accounts
16:41:29 <djmitche> By my read of the channel, we've had about one person per week unable to connect to trac
16:41:38 <tomprince> I know twisted recently switched to using github oauth login.
16:41:38 <verm__> and that's because their names matched banned words on two different spamlits
16:41:42 <djmitche> yeah, some making accounts, but for cmouse it was posting
16:41:52 <tomprince> But, we've gotten spam after switching.
16:41:54 <djmitche> tomprince: was that trac_github?
16:42:07 <verm__> cmouse is one of the two i listed above they should have been in the NOTSPAM group i fixed that
16:42:12 <tomprince> I think so. Looking.
16:42:13 <verm__> another one i fixed the other day
16:42:16 <djmitche> ok
16:42:26 <verm__> i haven't heard of this other issue
16:42:32 <djmitche> so that's generally a good solution to people who let us know they can't post
16:42:44 <verm__> one person per week unable to connect... do you remember who any of them were or what the problem was?  is it signup or logging in?
16:42:51 <djmitche> well, that's what we're hearing
16:43:04 <djmitche> there are probably more :)
16:43:19 <djmitche> I'm not sure -- cmouse were you unable to make an account, or unable to post?
16:43:30 <verm__> i belive you but i have only heard of / helped two people
16:43:46 <verm__> djmitche: cmouse couldn't make an account before because 'cmouse' is in two seperate spamlists (the word)
16:43:47 <tomprince> djmitche: Yes https://github.com/twisted-infra/braid/blob/master/services/trac/fabfile.py#L64
16:43:53 <djmitche> right, I just mean probably many people are just giving up b/c they don't use IRC
16:43:55 <verm__> and then they couldn't create a ticket for that reason
16:44:01 <verm__> djmitche: oh, i see
16:44:24 <verm__> well, the only thing i could do is have more visual captchas for login
16:44:28 <verm__> and disable all the automated stuff
16:44:34 <djmitche> maybe
16:44:37 <verm__> we could also have a notice on the webpage
16:44:42 <djmitche> yeah, that would be helpful
16:44:45 <verm__> letting users know they can email sysadmin@ to get it fixed
16:44:47 <djmitche> maybe with a link to webirc
16:44:48 <djmitche> right
16:44:51 <djmitche> that's good too
16:44:54 <verm__> link to webirc is good
16:44:57 <verm__> i say we do that first
16:45:01 <verm__> to see how big of a problem it is
16:45:08 <verm__> if it's big i'll figure out a real solution
16:45:12 <djmitche> ok
16:45:12 <rutsky> what about oauth login from other services like GitHub, GMail, etc?
16:45:25 <verm__> rutsky: that doesn't negate they can sign up to trac directly
16:45:28 <verm__> i wouldn't disable that
16:45:40 <tardyp> verm__: why not?
16:45:42 <verm__> i've tried it with other projects there are a *lot* of opensource devs that like their privacy
16:45:45 <djmitche> or even if we did disable that, it's not hard to set up a github account
16:45:58 <djmitche> and tomprince indicated there's still spam
16:46:05 <tardyp> we require a github login in order to contribute already
16:46:13 <verm__> not to create a ticket
16:46:16 <verm__> or edit a wiki page
16:46:29 <djmitche> well, nobody edits wiki pages :)
16:46:36 <bdbaddog> people who care.. realy really care about this.
16:46:51 <verm__> yes, it because a terrible set of discussions
16:46:51 <tardyp> do we care about those people?
16:46:55 <verm__> yes
16:46:57 <tardyp> if they dont contribute to the code
16:47:10 <djmitche> if they categorically WILL NOT contirbute to the code
16:47:17 <tomprince> verm__: Are there really a lot, and is the amount that would not get github larger than the ones that are turned off by having to register on yet another site?
16:47:21 <djmitche> they are likely to also be filing heavily redacted bugs
16:47:31 <rutsky> I don't suggest to limit accounts to accounts of specific service (which is bad), but if we could cover 70-80% of accounts by supporting popular services, it would be nice (if only there is not lots of spam accounts on such services, of course)
16:47:41 <tomprince> Yes, it will exclude some people. But it might include more.
16:47:57 <verm__> tomprince: depends on the project but i have yet to see it stick perminantly (disabling private logins)
16:48:02 <tardyp> myself, I prefer to use github openid, rather rhan creating another account
16:48:06 <tardyp> with another pw
16:48:07 <verm__> as bdbaddog said the ones who care care a *lot*
16:48:08 <djmitche> fwiw, switching devel to github saw a 2x increase in contirbutions within the first few months
16:48:10 <djmitche> and continued growth since then
16:48:23 <verm__> you end up with tons of discussions about privacy and all other tangent comments
16:48:30 <verm__> can we at least try to see how big of a problem this is?
16:48:38 <tardyp> and I would expect a lot of people are like me. much more than the github FUD people
16:48:39 <verm__> i can create a custom email alias
16:48:43 <verm__> and let people email me directly
16:49:01 <djmitche> that would be good -- at least for new accounts, I think you're the only one who can help
16:49:02 <verm__> that will let me see what is catching them
16:49:10 <bdbaddog> so is this an either/or issue.   github OA only  vs standalone login only?
16:49:11 <verm__> for current accounts
16:49:17 <verm__> anyone with USER_ADMIN or PERMISSION_ADMIN (i think)
16:49:29 <verm__> can give the user a 'NOTSPAM' permission
16:49:35 <verm__> that will buypass 100% of all spam checks everywhere
16:49:42 <verm__> bypass, too
16:49:44 <djmitche> except new accounts, right?
16:49:46 <tardyp> seriously. We did not had any such discussion when we moved the code to github!
16:50:01 <tardyp> lets move the trac auth to github
16:50:17 <verm__> i would be against it because it doesn't solve the problem
16:50:27 <verm__> lots of users are getting caught when creating tickets, too
16:50:39 <verm__> i want to solve all the issues centrally and leave things as they are it's less to maintain
16:50:53 <verm__> if it means removing botcoute and fspamlist then we'll do that and find some other options
16:51:00 <verm__> botscoute
16:51:17 <tomprince> I would vote to move login to github.
16:51:17 <djmitche> I think that's enough to agree on this week
16:51:24 <djmitche> we can re-visit the oauth next week
16:51:25 <rutsky> is Trac provides some convenient tools for moderation?
16:51:28 <verm__> ok
16:51:34 <djmitche> rutsky: no, not really
16:51:48 <verm__> well, that's not quite true
16:51:52 <verm__> there is the spam moderation log
16:51:58 <verm__> but we get so many spam attempts it becomes useless
16:52:00 <tardyp> I would like to have a vote on oauth as well
16:52:03 <verm__> all the false positives get washed out
16:52:06 <tardyp> at least on the mailing list
16:52:24 <verm__> tardyp: it is a pain to maintain
16:52:31 <djmitche> #agree will add an email alias for trac spam/login problems, put that alias on the account-creation and bug-creation pages, and see how many issues we find
16:52:33 <verm__> unless we wipe out everyones passwords
16:52:36 <rutsky> verm__: do you mean this page http://trac.buildbot.net/admin/spamfilter/monitor with 146118 unread notifications? :)
16:52:38 <verm__> and force everyone to use github
16:52:41 <djmitche> verm__: so is spam :)
16:52:52 <tardyp> yes. we force everyone to github
16:52:53 <verm__> rutsky: yes that page :)
16:53:08 <verm__> and then deal with all the github oauth timeouts with trac
16:53:25 <djmitche> tardyp: I'm open to putting it out to the mailing list
16:53:48 <djmitche> and we can take some of Twisted's experience to heart there
16:54:03 <djmitche> tardyp: can you make that post?
16:54:07 <verm__> moving to oauth with github isn't going to solve this problem and introduces a set of new ones at least give me a chance to solve this problem
16:54:16 <djmitche> right, I think they're separate issues
16:54:17 <verm__> rather than adding another layer of issues to deal with, please
16:54:26 <rutsky> djmitche: last time I tried to post bug on Twisted's bug tracker... I was filtered as spam :)
16:54:27 <tardyp> ok then
16:54:46 <verm__> if the spam problem gets fixed then github login shouldn't be required
16:55:08 <djmitche> verm__: I think github login is still useful -- I certainly almost never report a bug to a project that doesn't use github issues
16:55:15 <djmitche> it's just not worth it
16:55:16 <tomprince> verm__: I think github login is probably a good idea anyway.
16:55:20 <djmitche> much less when I get a spam error
16:56:11 <verm__> trachelp@buildbot.net goes to me now
16:56:22 <djmitche> #action tardyp to propose github login as an option on the mailing list
16:56:24 <verm__> it may be but you guys are assuming the trac github plugin actually works well
16:56:25 <djmitche> awesome :)
16:56:28 <verm__> it doesn't
16:56:35 <verm__> the oauth plugin
16:56:36 <djmitche> well, that's what we can talk to Twisted about
16:57:00 <djmitche> none of the Trac plugins work well, so there's a balance point to find
16:57:23 <verm__> we'll also need to ask them what happens with usernames the collide
16:57:51 <tomprince> I think twisted just ignored the issue.
16:58:06 <verm__> i'm not a fan of yanking peoples usernames
16:58:10 <tomprince> And manually verified the names of those people with admin priveleges.
16:58:54 <djmitche> this sounds like good info to see in the mailing list thread
16:59:19 <djmitche> verm__: you're on the list, right? will you reply?
16:59:27 <verm__> which? devel?
16:59:41 <verm__> i'm no both as far as i know
16:59:54 <djmitche> I tihnk this would be good for users@
16:59:56 <djmitche> up to tardyp though
16:59:58 <verm__> oh nope just devel
17:00:14 <tardyp> ok
17:00:18 <tardyp> I'll make both
17:00:21 <djmitche> cool :)
17:00:23 <djmitche> ok
17:00:24 <verm__> i'll get on users, too
17:00:28 <djmitche> #topic Add "trivial" tag to GitHub for trivial PR (typos, formatting fixes, etc).
17:00:45 <djmitche> ^^ I bet we can agree this is OK?
17:01:03 <verm__> oh i'm already subscribed heh where are the emails going
17:01:03 <rutsky> some of my PRs are super trivial, and I think it's better to handle them faster
17:01:07 <verm__> trivial sounds like a great idea
17:01:39 <djmitche> #agreed will create "trivial" tag that can be merged much more quickly (if the reviewer agrees they are trivial)
17:01:43 <rutsky> trivial is just like "mergeme", but haven't reviewed yet
17:02:12 <djmitche> right
17:02:23 <rutsky> afk...
17:02:52 <djmitche> label created :)
17:02:57 <djmitche> #topic http://buildbot.buildbot.net/ is down
17:03:19 <djmitche> Does anyone know what's wrong?  I haven't looked
17:03:25 <djmitche> although when I set up uptimerobot.com, I noticed this :)
17:03:28 <rutsky> djmitche: thanks!
17:03:36 <verm__> who maintains that?
17:03:44 <djmitche> sysadmins in generla
17:03:49 <tomprince> djmitche: I'm curious if there is anything in the logs?
17:04:02 <verm__> it's not running
17:04:12 <verm__> supervisord is running
17:04:32 <tomprince> I know at one point it would have had a bad config.
17:04:38 <tomprince> Did it never recover from that?
17:04:44 <djmitche> maybe not
17:04:48 <djmitche> http://trac.buildbot.net/ticket/3545 would help us discover such things at least
17:05:05 <verm__> This could either be a previously started instance of your application or a
17:05:05 <verm__> different application entirely. To start a new one, either run it in some other
17:05:05 <verm__> directory, or use the --pidfile and --logfile parameters to avoid clashes.
17:05:05 <verm__> Another twistd server is running, PID 50302
17:05:08 <verm__> just that forever
17:05:16 <verm__> that pid doesn't exist
17:05:53 <verm__> it fataled out on 2016-05-24
17:06:13 <skelly> exceptions.KeyError: 'CHECK_CONFIG'
17:06:29 <skelly> is what I see, same date
17:06:33 <verm__> yeah
17:06:36 <tomprince> Gah.
17:06:37 <verm__> that is the problem
17:06:42 * tomprince *headdesk*
17:06:43 <verm__> just tried starting it manually
17:07:20 <verm__> how do we fix that
17:07:51 <tomprince> PR incomming.
17:08:07 <djmitche> tomprince: would it be helpful if you had host access?
17:08:35 <bb-trac> [trac] #3545/enhancement (new) updated by rutsky (Dustin, is uptimerobot.com provides some kind of status page? Or only API to query ...) http://trac.buildbot.net/ticket/3545
17:09:10 <bb-trac> [trac] #3545/enhancement (new) updated by dustin (No, it doesn't -- I was thinking of building a static page on S3 that pulls from the ...) http://trac.buildbot.net/ticket/3545
17:09:11 <tomprince> https://github.com/buildbot/metabbotcfg/pull/43
17:09:39 <skelly> ^- LGTM
17:09:44 <djmitche> sheep it
17:09:57 <tomprince> Have to wait for travis. :)
17:10:02 <djmitche> heh
17:10:03 <djmitche> ok
17:10:19 <djmitche> tomprince: lmk if you want host access (potentially in the form of a PR to buildbot-infra)
17:10:23 <djmitche> #topic MOSS Funding status update
17:10:39 <tomprince> I added checks to make sure the config wasn't busted. But missed that the check for checking configs was busted when not checking configs.
17:10:49 <djmitche> #info Conservancy has the funding from Mozilla, and we are prepared to pay out the bounties we have already posted
17:10:56 <djmitche> hehe
17:11:05 <tomprince> djmitche: I probably won't unless something else comes up.
17:11:08 <djmitche> ok
17:11:30 <rutsky> djmitche: nice to hear that :)
17:11:30 <djmitche> ^^ is that enough information re MOSS?
17:11:34 <djmitche> ok
17:11:58 <djmitche> any other business?
17:12:07 <rutsky> djmitche: do I need to do something in terms of contract, etc?
17:12:24 <verm__> nope
17:12:33 <djmitche> rutsky: I'll double-check.  I think we had an agreement already?
17:12:49 <rutsky> djmitche: I haven't singed anything AFAIK :)
17:12:53 <verm__> tardyp: was it you that turned me onto pbr for the console client?
17:13:01 <djmitche> #action djmitche to verify p's and q's for rutsky's bounty
17:13:02 <rutsky> djmitche: I saw agreement example on Wiki I think...
17:13:08 <djmitche> ok
17:13:12 <djmitche> I'll be in touch by email
17:13:32 <verm__> i need to get this onto github so others can contribute
17:13:34 <rutsky> djmitche: ok, thanks!
17:13:37 <verm__> i got the websocket problem sorted
17:13:47 <verm__> so now i have a 'tail' mode working for watching a master
17:13:51 <djmitche> verm__: that'd be great :D
17:14:00 <verm__> works good i use it as my main method of monitoring now
17:14:09 <verm__> i'm working on a multi-master mode
17:14:13 <verm__> so i can watch all of them at once :)
17:14:20 <verm__> from different projects
17:14:48 <djmitche> sweet
17:15:04 <djmitche> ok, I'm going to go back to listening to the keynote :)
17:15:26 <djmitche> Parisa Tabriz now
17:15:30 <djmitche> #endmeeting