#archlinux32 | Logs for 2018-07-17

[00:27:13] <buildmaster> haskell-binary-conduit is broken (says tyzoid-srv0-bs0). - I rescheduled: haskell-conduit.
[02:02:19] -!- NoobAlice1 has joined #archlinux32
[02:04:05] -!- NoobAlice has quit [Ping timeout: 240 seconds]
[03:12:21] -!- buildmaster has quit [Ping timeout: 240 seconds]
[03:31:14] -!- buildmaster has joined #archlinux32
[04:38:42] * buildmaster resumes sanity.
[06:07:34] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-07-17T05:58:30.ckSgk3"?.
[06:32:07] -!- deep42thought has joined #archlinux32
[06:33:56] <deep42thought> tyzoid what's that liberapay? I thought, we wanted to abandon github?
[06:34:21] <deep42thought> the buildmaster has "load average: 19.59, 19.03, 19.68"
[06:34:22] <deep42thought> :-/
[06:41:52] -!- deep42thought has quit [Quit: Leaving.]
[06:47:30] <buildmaster> Hi deep42thought!
[07:10:49] -!- isacdaavid has quit [Quit: Leaving.]
[08:00:12] -!- deep42thought has joined #archlinux32
[08:00:13] <buildmaster> Hi deep42thought!
[08:00:20] <deep42thought> Hi buildmaster, what's up?
[08:00:20] <buildmaster> up? I'm up for 5 days, 10 hours, 20 minutes, load average: 0.60, 0.62, 1.71 ... and I'm insane :-D
[08:00:54] -!- abaumann has joined #archlinux32
[08:00:54] <buildmaster> Hi abaumann!
[08:00:58] <deep42thought> Hi abaumann!
[08:01:17] <abaumann> Hi.
[08:01:31] <abaumann> git seems to be under heavy load..
[08:01:49] <deep42thought> maybe that killed the buildmaster, too?
[08:02:08] <abaumann> both running on the same physical machine?
[08:02:11] <deep42thought> no
[08:02:20] <deep42thought> but the buildmaster heavily relies on git
[08:02:25] <deep42thought> e.g. pulls from there
[08:02:27] <abaumann> but maybe the buildmaster got troubled because it cannot access git.
[08:02:32] <abaumann> but then it should only hang.
[08:02:35] <deep42thought> and I'm afraid, I didn't implement error handling :-/
[08:02:48] <deep42thought> yeah, exactly
[08:03:00] <abaumann> which is sort of error handling ;-)
[08:03:43] <deep42thought> thats one optimistic way to interpret this :-)
[08:05:03] <abaumann> well, at least it doesn't continue building not using the updated git repo..
[08:05:06] <abaumann> https://bbs.archlinux32.org
[08:05:06] <deep42thought> buildmaster issue seems to have a different cause: some network error
[08:05:07] <phrik> Title:x86_64 Extra firefox 61.0.1-1 Standalone web browser from mozilla.org / Installation / Arch Linux 32 Forums (at bbs.archlinux32.org)
[08:05:26] <abaumann> I think, this could be a valid option, to start using binary packages for things we cannot build.
[08:05:32] * buildmaster resumes sanity.
[08:05:52] * deep42thought dislikes this
[08:05:59] <deep42thought> but yeah, you're right
[08:06:17] <abaumann> * abaumann also dislikes it
[08:10:05] <deep42thought> huh? python2 failed again, but when I patched it and built manually, it succeeded O.o
[08:13:14] <deep42thought> ah, typo
[08:15:37] <abaumann> uff
[08:15:41] <abaumann> :-)
[08:15:42] <deep42thought> hmm?
[08:15:56] <abaumann> only a typo. I was referring to that.
[08:16:02] <deep42thought> ah, ok
[08:16:10] -!- NoobAlice1 has quit [Quit: Leaving.]
[08:16:27] <deep42thought> the sed expression applied to the PKGBUILD, but not to what "declare -f" printed :-D
[08:17:10] * deep42thought wondered why "git push" takes so long
[08:19:20] <deep42thought> maybe we should host a mirror of git somewhere
[08:19:30] <deep42thought> so the build system does not go down if git does
[08:29:34] <deep42thought> load average: 40.05, 39.77, 36.06
[08:29:35] <deep42thought> :-D
[08:44:49] -!- eduardoeae has quit [Ping timeout: 248 seconds]
[08:49:55] <abaumann> that's a good idea, git is sort of a central service not using a lot of resources anyway.
[08:50:30] <deep42thought> I would skip all the fancy web frontend parts and simply provide git over ssh or a git daemon - what do you think?
[08:56:24] <abaumann> yes. exacly. the frontends can still live virtually on the existing machine.
[08:56:30] <abaumann> *exactly
[08:56:39] <abaumann> but.
[08:56:44] <abaumann> there is user management.
[08:56:51] <abaumann> repo management.
[08:57:09] <deep42thought> hmmmmmmmmm
[08:57:18] <abaumann> We should ask tyzoid, because he knows the details, how things are connected..
[08:58:07] <abaumann> maybe it's easier to find out, what is eating up resources and do some optimizations there?
[08:58:23] <deep42thought> I think, redundancy is a good idea anyways
[08:58:46] <abaumann> that's true. but it also opens up a nasty maintainance problem..
[08:58:50] <deep42thought> we could add the user recognition into git, too :-D
[08:58:58] <deep42thought> then it would be naturally duplicated
[08:58:58] <abaumann> :-)
[08:59:22] <deep42thought> obviously, not every packager should have write access to _that_ repository
[08:59:38] <abaumann> Having a read-only copy of the git repos at least allows continuation of builds.
[08:59:42] <deep42thought> and also probably allowing everyone to read from it is a bad idea
[08:59:45] <abaumann> but we cannot work on the build system nor on packages.
[09:00:17] <deep42thought> we would also need to make sure, that the repos stay synchronized
[09:00:39] <deep42thought> e.g. if main goes down, changes are applied to backup, then main comes up again, it needs to pull the changes from the backup
[09:00:53] <abaumann> ok. this we could arrange with some branches.
[09:01:00] <deep42thought> yeah, right
[09:01:00] <abaumann> which get pushed in the "right" way.
[09:01:23] * deep42thought is setting up gitolite (and probably cgit) now
[09:01:57] <abaumann> the problem I see is: if you for instance have a backup branch in your backup repo, you have to almost automatically sync it with let's say master from the main repo, to be up-to-date.
[09:02:27] <abaumann> but there is also the builder scripts which now suddenly have to hop from one repo branch to another repo/branch.
[09:02:50] <deep42thought> I would add all repos as remotes and decrease the pull timeout drastically
[09:03:07] <deep42thought> on the buildmaster
[09:03:15] <abaumann> that's an option.
[09:07:16] <abaumann> gogs, aka gitea never had a "replicate to other server" feature?
[09:08:43] <deep42thought> no idea
[09:09:13] <abaumann> would be a real killer feature for them to implement.
[09:19:00] <abaumann> the buildmaster builds again? so, which git repo is it using?
[09:19:43] <deep42thought> the slaves time out on the "git pull" and then use, what they already have in their local git repository
[09:19:54] <deep42thought> this works as long as there are no newer commits to be built
[09:20:11] <deep42thought> and we cannot push new patches
[09:20:20] <deep42thought> to the packages and to the build scripts
[09:20:45] <abaumann> ah. but at least unchanged packages get built and we can empty the queue a bit.
[09:20:53] <deep42thought> yes
[09:20:55] <deep42thought> exactly
[09:21:07] <deep42thought> but I cannot push my fixed python2 patch :-(
[09:21:18] <abaumann> which is terrible ;-)
[09:28:19] -!- abaumann has quit [Quit: Page closed]
[09:40:56] <buildmaster> linux is broken (says nlopc46).
[09:49:40] <buildmaster> linux-zen is broken (says nlopc46).
[09:57:48] <buildmaster> nvidia-340xx-utils is broken (says nlopc46).
[10:15:00] <buildmaster> sonic-pi is broken (says nlopc46).
[10:32:14] <deep42thought> https://git.eckner.net
[10:32:15] <phrik> Title:Git repository browser (at git.eckner.net)
[10:36:29] <deep42thought> please don't overload that site - it's hosted on my bananapi and my upstream bandwidth is quite low (~2MBit/s) :-)
[10:36:38] -!- deep42thought has quit [Remote host closed the connection]
[10:39:52] -!- deep42thought has joined #archlinux32
[10:39:52] <buildmaster> Hi deep42thought!
[11:21:27] -!- Alina-malina has quit [Ping timeout: 240 seconds]
[11:40:10] -!- deep42thought has quit [Quit: Leaving.]
[13:33:06] <buildmaster> pandoc is broken (says nlopc46).
[14:14:37] -!- oaken-source has joined #archlinux32
[14:24:18] -!- eduardoeae has joined #archlinux32
[14:35:06] -!- deep42thought has joined #archlinux32
[14:35:07] <buildmaster> Hi deep42thought!
[15:09:05] <tyzoid> Let's see if this helps. I increased allocated ram from 2Gb to 3Gb, removed the 2Gb allocation of swap, and added a vcore
[15:09:11] <tyzoid> deep42thought ^
[15:09:24] <tyzoid> I'm hoping now if it hits issues, it'll oom
[15:11:51] <deep42thought> so you say, gitea itself was the problem?
[15:12:12] <tyzoid> oh yeah, definately
[15:12:28] <tyzoid> something weird is going on with it
[15:12:31] <tyzoid> just no idea what or why
[15:12:34] <deep42thought> :-/
[15:14:06] <tyzoid> btw, I started a liberapay page if people wanted to help support the other web services I run
[15:14:16] <tyzoid> I'm hoping to get some more dedicated resources for projects
[15:14:35] <deep42thought> ah, that was the request you sent on github?
[15:14:41] <tyzoid> so hopefully if I get donations on that, it'll free up my servers
[15:14:41] <tyzoid> yeah
[15:15:18] <tyzoid> Links here, if anyone was interested in seeing what my plan is: https://liberapay.com
[15:15:19] <phrik> Title:tyzoid's profile - Liberapay (at liberapay.com)
[15:26:32] -!- abaumann has joined #archlinux32
[15:26:33] <buildmaster> Hi abaumann!
[15:26:49] <deep42thought> Hi abaumann!
[15:26:51] <abaumann> Hi.
[15:26:58] <abaumann> The git-knot is solved, I see.
[15:27:02] <deep42thought> yeah
[15:27:10] <deep42thought> and the backup did nothing to it :-D
[15:27:30] -!- abaumann has quit [Client Quit]
[15:27:37] -!- Alina-malina has joined #archlinux32
[15:27:50] <deep42thought> ^swiss brevity
[15:28:03] -!- abaumann has joined #archlinux32
[15:28:04] <buildmaster> Hi abaumann!
[15:28:26] <deep42thought> no offense
[15:28:48] <abaumann> no, this time it was pressing Ctrl-W in the wrong window. :-)
[15:28:54] <deep42thought> hehe
[15:29:02] * deep42thought knows that
[15:29:21] <deep42thought> or rather: has done the same before
[15:55:13] <tyzoid> wb abaumann
[16:02:09] <abaumann> hi tyzoid. :-)
[16:20:25] -!- thePiGrepper has joined #archlinux32
[16:22:38] <thePiGrepper> hi, I've been having issues with xorg-server since the upgrade to 1.20 on my netbook. has anyone else experienced a similar issue?
[16:24:04] <abaumann> deep42thought:On testing I have a mirrorlist from 64-bit :-)
[16:24:17] <abaumann> thePiGrepper: yeah, xorg 1.20 broke a lot.
[16:24:23] <deep42thought> oh, huh?
[16:24:36] <abaumann> What kind of drivers where you using and what is you model of the GPU?
[16:24:39] * deep42thought is not using the mirrorlist
[16:24:57] <abaumann> * abaumann thinks this could be a good idea, to use one :-)
[16:25:15] <deep42thought> I'm to lazy to always merge it
[16:25:25] <abaumann> let me see on staging, if we have the same issue there.
[16:25:26] <deep42thought> so I made my own package providing pacman-mirrorlist
[16:25:31] <deep42thought> yeah, we have
[16:25:36] <deep42thought> PKGBUILD is borked
[16:25:41] <deep42thought> more precisely:
[16:25:50] * deep42thought broke core/pacman-mirrolist's PKGBUILD
[16:26:14] <abaumann> good we noticed. :-)
[16:27:11] <deep42thought> ah, no, wait
[16:27:12] <deep42thought> gmmm
[16:27:24] <abaumann> huh?
[16:27:33] <deep42thought> actually, the source looks good
[16:27:36] <deep42thought> strange
[16:27:58] <deep42thought> maybe if git.archlinux32.org is unavailable, our modification is not being applied?
[16:28:00] <deep42thought> :-/
[16:28:23] <abaumann> that was exactly my thought too.
[16:28:28] <deep42thought> :-/
[16:28:40] <deep42thought> actually, it should fail if it cannot find the commit ...
[16:28:46] <abaumann> ok. this would also explain the linux, linux-zen failures?
[16:28:54] <deep42thought> yeah, probably
[16:29:45] <deep42thought> I'll add a TODO and look at it later
[16:29:52] <deep42thought> for now, you should reschedule pacman-mirrorlist
[16:29:56] <deep42thought> but I have to go now ...
[16:30:00] -!- deep42thought has quit [Quit: Leaving.]
[16:30:08] <abaumann> ok. will to that.
[16:30:31] <abaumann> but also later. I'm at work and I don't have mutt with the proper keys here.
[16:31:25] <abaumann> thePiGrepper: did you check the forums at https://bbs.archlinux32.org there were quite some people complaining and offering solutions over there..
[16:32:54] -!- abaumann has quit [Quit: Page closed]
[17:07:04] -!- isacdaavid has joined #archlinux32
[17:15:17] <thePiGrepper> I've been one of the ppl writing threads regarding xorg issues. I've been having this problem for the last 3 months maybe. And I'm pretty sure is an upstream issue, because exactly the same issue is happening to me using void. I've been bisecting the xserver git repo and it's a mess, most of the commits are not even building, because it's plagued with make/meson "fixes" that break the build.
[17:16:44] -!- thePiGrepper has quit [Quit: leaving]
[17:17:51] -!- thePiGrepper has joined #archlinux32
[17:17:58] <tyzoid> weird
[17:18:28] <tyzoid> thePiGrepper: Can you document the hardware you've got, if none of the existing solutions work for you?
[17:18:49] <tyzoid> If you post on the forum, it'll give us somewhere central to refer to when we send the bug up
[17:19:01] <tyzoid> or you could post on the bugtracke
[17:19:04] <tyzoid> bugtracker*
[17:33:38] -!- davor has quit [Quit: Bye]
[17:34:50] -!- davor has joined #archlinux32
[17:41:37] -!- rcf has quit [Remote host closed the connection]
[17:53:41] <thePiGrepper> here's the thread link: https://bbs.archlinux32.org . I have here the Xorg dump. https://pastebin.com
[17:53:42] <phrik> Title:Xorg stopped working after updating yesterday / Pacman / Pacman Upgrades / Arch Linux 32 Forums (at bbs.archlinux32.org)
[18:00:29] <tyzoid> Ah, right.
[18:01:03] <thePiGrepper> well, the part which actually is bothering me is not the bug itself. I can deal with the bug, but the debugging process was/is really annoying. I downloaded the xserver repo and my only thought is: how can the xorg developers work with a repo which half of the commit can't even build? wouldnt that be like the minimum requirement to commit a patch?
[18:08:53] <thePiGrepper> another curious thing is that all 1.19 branch, up to the last commit builds just fine and works on my system. however only up to the 49th commit from 1.19.0 builds and works. apparently some sha1 lib dependency broke the build up to 1.19.0+80, where it builds but no longer works on my system.
[18:09:01] <tyzoid> I would generally say so, but (without knowledge of their repo or structure), do they have release branches?
[18:09:16] <tyzoid> huh
[18:11:02] <thePiGrepper> they development rules are quite simple. for example, they have a development branch based on the last stable version(ie 1.19.0), in parallel they have an stable version which starts from there(1.19.1, 1.19.2, and so on)
[18:11:38] <thePiGrepper> in their dev branch, when they they want to tag the first rc, they tag it as 1.19.99
[18:12:13] <thePiGrepper> I think it is like 1.19.99-1, 1.19.99-2 something like that.(dont have the repo at hand).
[18:13:01] <thePiGrepper> some security patches from dev, as well as bug fixes, go to stable and create the patch versions.(ie: 1.19.6)
[18:13:17] <thePiGrepper> my issue is that their dev branch is really terrible for bisecting
[18:13:56] <thePiGrepper> I dont have issues with the dev-branch having bugs, that's fine. but not even building is ... something else.
[18:15:10] <thePiGrepper> one more thing is interesting though. void and arch32 have this issue with xorg-server 1.20, however debian doesnt.
[18:15:40] <thePiGrepper> I'm even thinking about checking their .deb and try to spot the differences
[18:42:34] -!- thePiGrepper has parted #archlinux32
[18:43:23] -!- thePiGrepper has joined #archlinux32
[18:43:37] -!- deep42thought has joined #archlinux32
[18:43:38] <buildmaster> Hi deep42thought!
[18:43:54] <deep42thought> thePiGrepper: looking at the patches in debian might not be the worst idea
[18:50:24] -!- bill-auger has quit [Read error: Connection reset by peer]
[18:53:09] -!- bill-auger has joined #archlinux32
[18:59:58] <deep42thought> abaumann: I rescheduled pacman-mirrorlist
[19:00:04] <deep42thought> (and found a bug in the buildmaster thereby)
[19:24:08] -!- rcf has joined #archlinux32
[19:32:35] -!- lukeshu has quit [Ping timeout: 276 seconds]
[19:51:01] -!- lukeshu has joined #archlinux32
[19:58:43] <thePiGrepper> one question, how is the package testing process? I've been trying to find a thread explaining what's the process needed to send packages from testing to core.
[19:58:58] <deep42thought> the "is_tested" bit needs to be set in the database :-)
[20:00:20] <thePiGrepper> hehe, that I didnt know either, however I was asking more about what are the requirements for a package to be declared as 'tested'?
[20:00:38] <deep42thought> someone needs to report it to the buildmaster as such
[20:00:41] <deep42thought> ... formally
[20:01:01] <deep42thought> e.g. send the package name + sha512sum in an encrypted, signed, stamped email
[20:01:16] <deep42thought> or from time to time, I set all packages older than x as "tested"
[20:01:18] <deep42thought> :-/
[20:02:58] <thePiGrepper> older than x? how so?
[20:03:27] <deep42thought> update binary_packages set is_tested=1 where build_date < 2018-07-10
[20:04:11] <deep42thought> ah: to be actually moved, a package must not have any open bug reports on the bug tracker
[20:04:27] <deep42thought> e.g. "is_tested" is ignored if there are unresolved bug reports about the package
[20:04:52] <thePiGrepper> oh, I see now. that thing regarding the database is useful info as well.
[20:04:53] <deep42thought> https://bbs.archlinux32.org
[20:04:54] <phrik> Title:automatic testing via manual installation(s) / Testing / Arch Linux 32 Forums (at bbs.archlinux32.org)
[20:05:43] <deep42thought> if you want to participate in testing, I need to put your gpg key into the buildmaster's database
[20:05:45] <thePiGrepper> ok, that last part is interesting. that was what I was asking. then, if a package on testing doesnt have unresolved bug reports after X time, it goes to core, right?
[20:05:51] <deep42thought> then you should be able to report pacakges
[20:06:05] <deep42thought> yes
[20:06:13] <deep42thought> but currently, this is a manual process
[20:06:27] <deep42thought> e.g. when I get annoyed by the number of untested packages, I mark them manually
[20:06:58] <thePiGrepper> as a matter of fact, I think I do want to do testing. that's why I'm trying to familiarize with the process
[20:07:32] <deep42thought> the forum thread has a lot of information on it
[20:08:56] <thePiGrepper> I have one question regarding unresolved bugs. you say no unresolved bugs is a requirement. but what about no bugs at all? couldnt that mean that it wasnt tested enough as well? or is there a fixed rotation time?
[20:09:28] <deep42thought> no bugs => no issues
[20:09:36] <deep42thought> that's the assumption
[20:09:53] <deep42thought> and as a matter of fact, there is too less testing :-)
[20:11:10] <thePiGrepper> ok, that's good to know. and how much time do you wait for this testing->core refresh. I've observed that it's around half a month, maybe more. is that a fixed period or it depends or something else?
[20:11:17] <thePiGrepper> *of
[20:11:33] <deep42thought> it depends on how annoyed I get by the large number of packages
[20:11:44] <deep42thought> but I should/will implement some fixed interval some time
[20:11:49] <deep42thought> probably something around 2 weeks
[20:12:00] <thePiGrepper> haha. right.
[20:30:54] -!- Cthulu201 has quit [Ping timeout: 260 seconds]
[20:38:38] <thePiGrepper> I'm reading the thread link you posted, report-installed-packages and sendemailadvanced, those packages no longer exist apparently. do I still need them?
[20:38:46] <deep42thought> yes
[20:38:48] <deep42thought> they were moved
[20:38:55] <deep42thought> they're now in releng
[20:39:10] <deep42thought> ah
[20:39:10] <thePiGrepper> thanks
[20:39:11] <deep42thought> no
[20:39:23] <tyzoid> deep42thought: Monitoring git.archlinux32.org, seems like it's slowly increasing in memory consumption
[20:39:34] <deep42thought> :-/
[20:39:36] <tyzoid> started ~400MB pretty soon after reboot, now up to ~700MB ~5hrs later
[20:39:39] <tyzoid> it's slow
[20:39:42] <tyzoid> but it's something
[20:39:43] <deep42thought> cronjob: systemctl restart gitea
[20:39:52] <tyzoid> Yeah, I'll have to see if that might fix it
[20:39:58] <deep42thought> "fix"
[20:40:04] <tyzoid> imo, I just need to get cacti running again
[20:40:13] <tyzoid> had it back when I had vps's with a vps provider
[20:40:20] <tyzoid> never set it up when I switched to dedis
[20:40:21] <deep42thought> thePiGrepper: they were probably just updated
[20:41:16] <deep42thought> have a look here: http://arch.eckner.net
[20:41:18] <phrik> Title:Index of /archlinuxewe/os/i686/ (at arch.eckner.net)
[20:43:20] <tyzoid> deep42thought: Just looked at the graphs
[20:43:20] <tyzoid> https://i.imgur.com
[20:43:24] <tyzoid> looks like a classic memory leak
[20:43:37] <deep42thought> :-)
[20:43:40] <deep42thought> bugreport!
[20:49:27] -!- Cthulu201 has joined #archlinux32
[20:49:29] <tyzoid> restarting gitea doesn't seem to reduce memory consumption any
[20:49:34] <tyzoid> still at 720M
[20:49:38] <tyzoid> most seem to be from mysql
[20:50:41] <deep42thought> memory leak in mysql seems unlikely
[20:50:52] <tyzoid> Yeah, that's just it doing caching
[20:51:15] <tyzoid> It could've been the fact that I had swap available to it, and it was just freezing up with swap on slow disks
[20:51:29] <tyzoid> when I increased the ram limit from 2-3, I removed swap
[20:54:23] -!- oaken-source has quit [Quit: (brb)]
[21:09:17] -!- oaken-source has joined #archlinux32
[21:11:29] <deep42thought> should we put tools necessary for reporting tested packages into some i686/releng repository?
[21:25:10] -!- bill-auger has quit [Remote host closed the connection]
[21:25:28] <tyzoid> deep42thought: I have no issues with that
[21:27:54] -!- bill-auger has joined #archlinux32
[21:34:54] -!- bill-auger has quit [Remote host closed the connection]
[22:07:02] -!- bill-auger has joined #archlinux32
[22:07:16] -!- bill-auger has quit [Remote host closed the connection]
[22:10:35] -!- bill-auger has joined #archlinux32
[22:22:05] -!- deep42thought has quit [Quit: Leaving.]
[22:57:38] -!- thePiGrepper has parted #archlinux32