16:01:55 <tardyp> #startmeeting weekly
16:01:55 <bb-supy> Meeting started Tue May  2 16:01:55 2017 UTC and is due to finish in 60 minutes.  The chair is tardyp. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:55 <bb-supy> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:55 <bb-supy> The meeting name has been set to 'weekly'
16:02:05 <tardyp> #topic introduction
16:02:09 <tardyp> anybody here?
16:02:20 <bdbaddog> I'm here.
16:02:22 <tardyp> djmitche is at a conference and cannot chair
16:02:51 <tardyp> agenda: https://titanpad.com/buildbot-agenda
16:03:03 <rjarry> I'm here
16:03:11 <tardyp> hi rjarry , bdbaddog
16:03:20 <rjarry> o/
16:04:06 <tardyp> looks like this will be a short meeting.
16:04:14 <tardyp> #topic week in review
16:04:46 <tardyp> #info Work ongoing on logHorizon. ready to be merged
16:05:14 <rjarry> tardyp: about logHorizon, do you intend to include builders, builds cleanup in the same PR?
16:05:16 <tardyp> #info kubernetes support is being working on
16:05:29 <tardyp> rjarry: no, that would be on subsequent PR
16:05:37 <tardyp> this one is already big enough
16:05:46 <rjarry> ok, and how about the amount of disk space saved after cleanup ?
16:06:06 <tardyp> It is also for a subsequent feature.
16:06:10 <rjarry> ok
16:06:19 <tardyp> actually we can only do that for sqlite.
16:06:41 <tardyp> and we need a step in the end doing the vacuum command, and then figure out the size of the file
16:06:49 <skelly> could get the size of all to-be-deleted chunks and use that as an estimate
16:06:53 <tardyp> for other db, I am not sure how we can query db size
16:07:23 <rjarry> there must be a "standard" SQL way of getting the size of a field
16:07:32 <bdbaddog> len(fieldname)
16:07:41 <bdbaddog> pretty sure..
16:07:58 <tardyp> well there is also the size in the indices, and so on. so I am not sure how correct this data would be
16:07:59 <rjarry> bdbaddog: does that work with aggrgates ?
16:08:07 <bdbaddog> select len(lgo_field_name) from TABLE where GONNA_DELETE_YOU_CONDITION ?
16:08:15 <skelly> sum(len(x))
16:08:18 <rjarry> select sum(len())
16:08:19 <rjarry> yep
16:08:34 <skelly> anyway, it should be in a subsequent PR
16:08:50 <rjarry> agreed
16:08:55 <tardyp> why not. There is already a before and after num of logchunk, which is logged in the step
16:09:16 <tardyp> I guess sum(len()) could be doable to
16:09:23 <tardyp> again, for subsequent PR
16:09:32 <skelly> I view it as more of an estimate rather than a precise number
16:09:37 <rjarry> sure
16:10:51 <tardyp> #info interresting discussion on how to estimate janitor db size gains
16:11:04 <rjarry> another subject: I wanted to talk about making workers/builders "hideable" from the UI
16:11:24 <rjarry> meaning, they would be accessible directly, but not listed by default
16:11:41 <rjarry> this would be convenient for janitor, for example
16:11:43 <tardyp> please add it to the buildbot agenda
16:11:58 <rjarry> but also other features which are not part of the "build" workflow
16:12:15 <tardyp> well, we don't have anything else do discuss actually. so..
16:12:20 <rjarry> in the titanpad thing ?
16:12:25 <tardyp> yes
16:12:39 <tardyp> #topic hiding builders/workers from the UI
16:12:45 <bdbaddog> still working on getting my bitbucket patch reasonable for pull request. maybe this weekend.
16:12:51 <tardyp> #undo
16:12:51 <bb-supy> Removing item from minutes: <ircmeeting.items.Topic object at 0x806961350>
16:12:59 <tardyp> yes, I forgot about that
16:13:16 <rjarry> tardyp: should I add this under "Topics"?
16:13:24 <tardyp> #info better bitbucket support discussion is ongoing
16:13:28 <tardyp> rjarry: yes
16:13:59 <tardyp> bdbaddog: did you see the email thread about gitpoller, and bitbutcket adding fake refs in ls-remote?
16:15:08 <rjarry> agenda updated
16:15:24 <bdbaddog> this; https://lists.buildbot.net/pipermail/users/2017-May/001261.html
16:15:32 <bdbaddog> btw. that link fails under chrome..
16:15:42 <bdbaddog> "An error occurred during a connection to lists.buildbot.net. The OCSP server suggests trying again later. Error code: SEC_ERROR_OCSP_TRY_SERVER_LATER"
16:15:57 <tardyp> works for me
16:15:58 <skelly> works for me in Chrome
16:16:33 <tardyp> anyway. maybe something to take in account in your dev
16:16:42 <cmouse> skelly: fails with firefox
16:16:47 <bdbaddog> sorry.. fails on firefox.
16:16:54 <tardyp> ah
16:17:03 <bdbaddog> missed that thread. I'll take a look at it later..
16:17:12 <cmouse> OCSP Response Data:
16:17:12 <cmouse> OCSP Response Status: trylater (0x3)
16:17:33 <cmouse> maybe restart web sewer?
16:18:15 <tardyp> looks like it comes from another server than list
16:18:21 <cmouse> tardyp: it comes frm lists.
16:18:30 <bdbaddog> The Online Certificate Status Protocol (OCSP) is an Internet protocol used for obtaining the revocation status of an X.509 digital certificate.
16:18:42 <cmouse> http://p.ip.fi/K9q_
16:18:48 <tardyp> anyway. time passes. we need to move on.
16:18:57 <cmouse> yes, the OCSP response comes from someone else but your server is responsible acquiring it
16:19:01 <tardyp> skelly: could you try and restart the webserver on lists?
16:19:21 <tardyp> #topic  hiding builders/workers from the UI
16:19:45 <tardyp> #info rjarry would like to hide Janitor builders and workers from UI
16:20:06 <rjarry> and this can go for schedulers as well, forgot to mention that
16:20:16 <tardyp> We could attach a special tag to the builders
16:20:32 <tardyp> we already have UI to filter by tags
16:20:37 <tardyp> which is not perfect
16:20:55 <rjarry> I fear this will interfere with user defined tags
16:20:56 <tardyp> there are tags info on workers to I think (not sure) but no UI for filtering (this is for sure)
16:21:10 <rjarry> and also, by default, all tags are displayed
16:21:21 <tardyp> we can create a __internal__ tag
16:21:30 <tardyp> we would change that default
16:21:35 <rjarry> ok
16:21:42 <tardyp> to filter out tags starting with __
16:22:02 <tardyp> we can also filter out stuff starting with __, which is actually much simpler
16:22:15 <rjarry> directly from the name then
16:22:19 <rjarry> no need for tags
16:22:23 <tardyp> yeah
16:22:37 <rjarry> then, we need an option to force display hidden stuff
16:22:42 <tardyp> would you like to submit an issue? I think that would be kind of easy
16:22:49 <tardyp> yes, like the show-old-builders option
16:22:58 <rjarry> for example yes
16:23:00 <rjarry> ok
16:23:08 <rjarry> I'll submit the issue
16:23:51 <tardyp> ok, there is a release topic, I did not start
16:23:55 <tardyp> #topic releases
16:24:53 <tardyp> I merged a fix this week https://github.com/buildbot/buildbot/pull/3145
16:25:11 <tardyp> this is about login issue with basic auth.
16:25:22 <tardyp> is there anybody here using basic auth?
16:26:02 <tardyp> I've got some report from people who were not able to make it work, and I think there may be some new CORS restriction coming from browsers which buildbot hit.
16:26:07 <skelly> blah, restarted the web server and now it won't even load
16:26:12 <bdbaddog> I'm using it..
16:26:16 <skelly> well, works now
16:26:20 <tardyp> I've suddenly been able to reproduce it
16:26:53 <tardyp> I am using chrome Version 58.0.3029.81 (64-bit)
16:27:07 <tardyp> and I could not login anymore with 0.9.6 and basic auth
16:27:16 <tardyp> bdbaddog: does it still work for you?
16:27:27 <cmouse> lists is still saying trylater.
16:27:35 <bdbaddog> yup. 0.9.6
16:27:42 <bdbaddog> just logged in.
16:27:58 <bdbaddog> lemm try crhome
16:28:08 <tardyp> I think this could be a reason for a emergency release
16:28:17 <bdbaddog> yup. works for me too.
16:28:20 <bdbaddog> on chrome.
16:28:24 <tardyp> which version of chrome?
16:28:42 <bdbaddog> 57.0.2987.133 (64-bit) on mac
16:28:58 <bdbaddog> wow.. thats older than ur verison.
16:29:01 <tardyp> ah.. so this could be chrome 58 which will kill it
16:30:19 <bdbaddog> updated: Version 58.0.3029.81 (64-bit) still seems to work.
16:30:30 <tardyp> mhhh...
16:30:37 <bdbaddog> URL is http: and not https: if that matters.
16:30:48 <tardyp> I tested with localhost:8010
16:30:52 <bdbaddog> (yes I know.. bad bad bdbaddog)
16:31:37 <tardyp> so I don't understand how to reproduce it exactly. anyway. this is now fixed in buildbot.
16:31:55 <tardyp> the result is that the login password is no more in the UI, but rather provided by the browser
16:32:00 <bdbaddog> should that only show up if using py3?
16:32:01 <tardyp> do you think it is a big issue?
16:32:25 <tardyp> bdbaddog: I though so, but I could reproduce it with 2.7 as well
16:33:04 <tardyp> and this is really a pure browser issue.
16:33:09 <bdbaddog> you're talking about using somehting like this: auth=util.UserPasswordAuth ?
16:33:31 <tardyp> https://github.com/buildbot/buildbot/issues/3163
16:33:37 <tardyp> yes
16:34:44 <tardyp> maybe this is a problem only for localhost or something.
16:34:52 <tardyp> anyway.
16:34:53 <bdbaddog> I can give that a try later.
16:34:57 <tardyp> we are running out of time.
16:35:04 <tardyp> any other topic?
16:35:06 <cmouse> are you setting Access-Control-Allowed-Origin?
16:35:15 <tardyp> cmouse: yes
16:35:16 <cmouse> ok
16:36:21 <tardyp> #endmeeting