IRC logs for #baserock for Monday, 2015-04-20

*** zoli__ has quit IRC00:02
*** zoli__ has joined #baserock01:47
*** zoli__ has quit IRC01:51
*** Kinnison has quit IRC03:08
*** Kinnison has joined #baserock03:08
*** zoli__ has joined #baserock03:17
*** zoli__ has quit IRC03:22
*** zoli__ has joined #baserock04:11
*** mike has joined #baserock06:09
*** mike is now known as Guest672306:10
*** gary_perkins has joined #baserock06:19
*** zoli__ has quit IRC06:26
*** paulw has joined #baserock06:33
*** gary_perkins has quit IRC06:42
*** gary_perkins has joined #baserock06:42
*** zoli__ has joined #baserock06:48
*** rdale has joined #baserock07:13
*** sambishop has quit IRC07:14
*** gary_perkins has quit IRC07:17
*** Albert has joined #baserock07:20
*** bashrc_ has joined #baserock08:10
*** jonathanmaw has joined #baserock08:10
*** sambishop has joined #baserock08:14
*** mariaderidder has joined #baserock08:29
*** tiagogomes has joined #baserock08:32
SotKtlsa: Could you maybe look at the review comments on the XFCE patches I sent to Gerrit at some point if you have time please?08:41
*** CTtpollard has joined #baserock08:42
tlsaSotK: sure08:42
SotKthanks :)08:42
SotKthere's no rush08:43
*** gary_perkins has joined #baserock08:45
*** ssam2 has joined #baserock08:46
*** ChanServ sets mode: +v ssam208:46
KinnisonMerp, erlang in build-system?08:59
Kinnisonwas that part of the openstack merge?09:00
*** marcdunford has quit IRC09:02
*** a1exhughe5 has joined #baserock09:03
*** fay_ has quit IRC09:03
*** fay_ has joined #baserock09:04
*** edcragg has joined #baserock09:06
pedroalvarezKinnison: hmm.. that looks wrong09:08
Kinnisonand django is in the build system now09:09
Kinnisonand devtools09:10
Kinnisonwhich seems wrong09:10
pedroalvarezKinnison: are you sure that you aren't building the openstack system?09:10
Kinnisonpython /src/ybd/ systems/build-system-x86_64.morph x86_6409:10
KinnisonI'll admit I'm not 100% sure ybd won't be being odd, but it seems unlikely09:10
Kinnisonxstatic-bootstrap-datepicker now09:10
pedroalvarez`git grep erlang |grep system` doesn't show the build system09:11
Kinnisonmight be a build-dep09:11
SotKdjango only looks to be used in openstack stuff09:11
SotKopenstack-services build-depends on erlang too09:12
pedroalvarezerlang is a build-dep in strata/openstack-services.morph only09:12
SotKbut that is only in the openstack system09:12
SotKand isn't a build dep for any of the other openstack stuff afaict09:12
KinnisonI'll le tyou know when ybd tells me :)09:13
KinnisonIt's on coreutils build right now, so I imagine it'll be a while before it reveals its secrets09:14
pedroalvarezKinnison: if ybd builds it, something is wrong :)09:14
KinnisonI wouldn't be surprised :)09:14
* Kinnison has been doing evil things to ybd09:15
pedroalvarezI don't expect you to add evil code to ybd to build random packages09:15
*** franred has joined #baserock09:16
KinnisonI wouldn't *intend* to09:16
pedroalvarezSotK: that was clever09:17
pedroalvarezdoes ybd have a command like that?09:17
KinnisonSotK: I expect ybd has done something "interesting" then09:17
Kinnisonybd doesn't have subcommands09:17
Kinnisonyou give it something and it does it -- build, or later when my hacking is done, deploy09:18
pedroalvarez[19:43:39] <paulsherwood> in other news, i timed building openstack-system: ybd 02:58:13, morph 02:58:0809:18
pedroalvarezthis is good :)09:18
KinnisonCoo, I notice we're now at version 1 of definitions format09:20
Kinnisonis there now a full description of what that means?09:20
ssam2work in progress09:21
KinnisonWIP :(09:21
*** Krin has joined #baserock09:21
*** rdale has quit IRC09:23
*** lachlanmackenzie has joined #baserock09:25
Kinnisonpaulsherwood: Any idea why ybd would calculate cache keys for things not in the build-graph of the goal ?09:25
ssam2I wonder if this 'files being overwritten' issue is due to me building stuff with the OSTree branch, and deploying it using rsync ...09:27
ssam2or, deploying stuff with the OSTree branch09:28
ssam2I'm suspicious about this, although I'm not quite sure what it means09:28
ssam2 stat -c '%h' usr/lib/python2.7/site-packages/Crypto/Random/Fortuna/__init__.py09:28
ssam2'%h' shows 'number of hard links' for the file... which suggests to me that a lot of files in the system are actually hardlinked to the same place09:29
Kinnisonssam2: merp09:29
SotKhmm, maybe look at the deployment tempdir and see what it looks like in there09:30
* straycat wonders why ubuntu suggest mounting xfs partitions used on swift storage nodes with nobarrier09:51
straycatSotK, it would be useful to get those xfce patches merged09:54
straycati'd like to be able to use dwm in baserock and it probably makes more sense to run it under x than weston, even though in theory you can run it under weston09:54
SotKstraycat: it would, but I don't know the answer to any of the questions that were asked, as most of the work wasn't mine :)09:56
* straycat nods09:58
radiofreessam2: is the drive in your fstab?09:58
radiofreeregarding auto-mounting in master09:58
ssam2radiofree: yes09:58
radiofreewhat message to you get from systemd?09:58
ssam2the one where it gives a timeout like [13s/1m30]09:59
ssam2then it goes into emergency mode09:59
straycati mean i guess they suggest it for performance, but that doesn't exactly seem safe09:59
radiofreei thought i added nofail to the fstab09:59
ssam2I don't see why it should need 'nofail'. Once in emergency mode, the disk can be mounted10:00
radiofreethat sounds broken then, do you get something like "sd 0:0:0:0: [sda] Attached SCSI disk" in dmesg before that happens10:00
ssam2not sure, i don't have time to look at it now, sorry10:00
radiofreessam2: no fail would mean you're not waiting 1m30s10:00
ssam2I don't have 'nofail' in there10:00
SotKwhere is the baserock: repo-alias defined?10:06 maybe ?10:08
KinnisonSotK: initially in I'd imagine10:09
ssam2oh. morphlib.util.combine_aliases() in fact10:09
KinnisonSotK: later via a number of routes including combine_aliases10:09
SotKaha! thanks!10:09
KinnisonIt's a bit "special" due to troves10:10
SotKits special in just the right way :)10:11
Kinnisonpedroalvarez: well, for whatever reason it had erlang -- it's now building it10:28
* Kinnison looks forward to working out why10:28
*** rdale has joined #baserock10:33
Kinnisoninterestingly ybd seems to think it's part of the build-deps for things like python-markdown10:45
SotKoh, python-markdown is in openstack-services as well as morph-utils10:46
SotKso I guess ybd is using the wrong build deps10:46
Kinnisonprobably part of its unifying things with the same name stuff then10:46
persiaDuplicate things should either be differentiated, or broken out.10:47
* SotK wonders why the intltool lorry file has no url12:21
ssam2is it a bazaar lp: launchpad url ?12:51
SotKnope, it has "branches": { "trunk": ... } instead12:52
*** paulwaters_ has joined #baserock12:54
franredSotK, that looks wrong... that page says `Get this branch: bzr branch lp:intltool`12:54
*** paulwaters_ has quit IRC12:55
ssam2SotK: that might be useful13:03
*** zoli__ has quit IRC13:03
SotKssam2: thanks!13:04
franredssam2, cheers :)13:06
* pedroalvarez sends 13:13
* pedroalvarez fixes a couple of errors, and resends quickly13:16
*** zoli__ has joined #baserock13:27
* pedroalvarez starts increasing disk to 300G13:57
*** tiagogomes has quit IRC13:58
*** tiagogomes_ has joined #baserock13:58
persia \o/14:00
franredpedroalvarez, thanks :)14:00
franredummm, my deploy is failing because install-files template require jinja2... but not sure how I can fix this...?
SotKinstall jinja2 I guess14:07
franredSotK, where?14:07
franredin my devel?14:07
jonathanmawsomething about my baserock VM is broken - my systemd journal is empty.14:07
franredSotK, that works, thanks!!14:09
SotKfranred: no problem :)14:09
SotKwe should maybe make morph have jinja2 in its build dependencies14:10
ssam2oh, yeah14:11
doffmCould someone review a lorry addition for me? Its for google compute engine instance configuration software.14:12
pedroalvarezjonathanmaw: hm.. I've seen that in the past, but I never managed to make enough noise to attract attention  from someone with more knowledge14:13
pedroalvarezjonathanmaw: `systemctl restart systemd-journald` might fix it14:13
jonathanmawooh, things have appeared!14:14
jonathanmawta pedroalvarez!14:14
pedroalvarezI think that /var is being  mounted after systemd starts writing on the journal14:15
*** mariaderidder has quit IRC14:15
pedroalvarezs/on the journal/in the journal filesin \/var/14:16
franreddoffm, the patch looks fine, but it would be nice if you can submitted via gerrit --> (
franreds/submitted/submit it/14:16
doffmfranred: Ok, will do.14:17
*** mariaderidder has joined #baserock14:18
franreddoffm, thanks :)14:20
*** mariaderidder has quit IRC14:23
*** mariaderidder has joined #baserock14:23
*** paulsherwood has quit IRC14:34
*** paulsherwood has joined #baserock14:35
ssam2seems something is going wrong with the systems I've deployed because they are missing some system users that systemd expects14:44
ssam2'systemd-network' and 'systemd-timesync'14:44
ssam2could that be more fallout from the upgrade to an unstable version of systemd?14:44
KinnisonPresumably systemd needs system-integration scripts to add those users14:45
KinnisonIf it doesn't have them in its own nss14:45
pedroalvarezI've been using that version for weeks, and I didn't see any error14:45
ssam2I get this in the journal:
pedroalvarezI think franred has a system with this version of systemd runing14:46
pedroalvarezfranred: can you check your journal to see if you have similar messages?14:47
jjardonssam2:  I think those user/groups are created by systemd at boot time14:47
franredpedroalvarez, sure14:47
ssam2jjardon: it hasn't created anything in /etc/passwd or /etc/group14:48
pedroalvarezfranred: you can also check if you have those users in /etc/passwd14:48
ssam2maybe this problem is because I have to use Linux 3.15.10 on this system? although I'm not sure exactly how that could break user creation14:49
ssam2Nov 08 23:17:50 hawk2 systemd-sysusers[101]: Failed to write files: File exists14:49
ssam2so I guess it's failing to create the users14:50
franredpedroalvarez, ssam2, systemd-journal is a group and systemd-network and systemd-timesync are users as systemd-resolve14:51
ssam2ok, thanks14:51
franredthey are created fine in my Openstack system14:51
ssam2would be useful if the systemd-sysusers error message was more detailed :)14:51
ssam2I can run it manually14:51
ssam2this may be because I got /etc into a bit of an inconsistent state by copying it from a system that had already booted. I'll try again with a clean /etc15:01
ssam2works with a clean /etc. I guess it had got in some partial state that systemd-sysusers didn't deal with very well15:07
franredthat are good news15:07
pedroalvarezstop hating systemd master! :)15:08
pedroalvarezwell, there is still the "mounting external disks" bug15:08
jjardonwe should aim to be able to boot with a empty /etc . Some people upstream is working on it but I do not know what is the current status15:09
jjardonssam2: by a clean /etc you meant a empty one?15:11
ssam2I mean one that hasn't been used in a system that was booted yet15:11
ssam2empty etc wouldn't be possible here: /etc/distbuild/*, /etc/hostname, /etc/resolv.conf all need to be present at least15:12
persiaThat indicates unfortunate things about the idempotency of the first-boot stuff15:12
persia/etc/resolv.conf shouldn't be required: systemd-networkd should take care of that.15:12
jonathanmawhrm, after moving linux-pam into core, it's started failing to configure with './configure: ./configure.linelo: line 14909: syntax error: unexpected word(expecting ")")'15:12
jonathanmawI suspect an unmet dependency15:12
ssam2missing m4 macros for autoconf ?15:12
jonathanmawssam2: it build-depends on autoconf-tarball15:13
jjardonssam2: I think the idea is to have a "default/system vendor" "/etc" in "/usr/share/etc"15:13
ssam2that error suggests that it's missing a dep that would provide an M4 macro that is used in configure.ac15:13
ssam2jjardon: that would be good.15:13
jjardonjonathanmaw: check you have at least autoconf, automake and libtool as deps15:14
jonathanmawjjardon: yep, gettext too.15:15
jjardonjonathanmaw: sorry, automake doesnt seems to be needed15:15
jjardonsorry, Im blind, its needed: AM_INIT_AUTOMAKE("Linux-PAM", 1.1.8)15:16
jjardonjonathanmaw: ^15:16
jjardonBTW, maybe some of you have time to take a look to ?15:17
jonathanmawit's configured now! I was missing pkg-config.15:23
jonathanmawand it's failing to build, undefined reference to yywrap.15:23
jonathanmawgoogling suggests flex.15:24
jjardonjonathanmaw: you migth want to cherry-pick this:
jonathanmawjjardon: flex was readily available15:34
jonathanmawcurrently, I need to work out why acl has decided that /lib64/ should be a symlink to a nonexistent
*** marcdunford has joined #baserock15:57
*** a1exhughe5 has quit IRC16:08
*** jonathanmaw has quit IRC16:14
*** zoli__ has quit IRC16:30
*** gary_perkins has quit IRC16:31
*** Albert has quit IRC16:32
*** Albert has joined #baserock16:34
*** marcdunford has quit IRC16:35
*** Albert has quit IRC16:46
*** Albert has joined #baserock16:48
ssam2guide for people who are maintaining Calxeda distbuild networks:
ssam2i know the target audience is limited for this, but it needs to be written down !16:49
ssam2I'm going to do a final test of and the two related changes, then merge them unless anyone has strong objections16:51
*** Albert has quit IRC16:57
straycatyes please don't merge that17:01
straycati've spotted a couple of issues with it17:01
straycatssam2, ^17:01
*** bashrc_ has quit IRC17:02
straycatit's also a little sad that we're losing the ability to serialise artifacts17:03
straycati'd have suggested a separate function for creating your artifact references, it may be useful in future to be able to serialise artifacts, and it's not completely trivial to write17:04
*** Krin has quit IRC17:04
ssam2straycat: why do we need to serialise artifacts? It's much faster to just calculate the build graph from scratch, because a serialised artifact is a huge amount of data17:08
straycatthere's a lot about this change i don't understand as well17:08
straycatssam2, maybe we'll want to write our artifacts to disk one day for some reason, i don't know?17:08
straycatthere's no terribly good reason to delete that code17:09
ssam2yes there is. It'll be there in the Git history if we still need it.17:09
straycatsorry but no, we'd previously discussed moving serialisation into morphlib proper17:09
straycati'm not up for deleting it17:09
straycati'm up for your speed improvements however17:09
*** mariaderidder has quit IRC17:10
*** ssam2 has quit IRC17:13
straycatok bye...17:15
*** CTtpollard has quit IRC17:22
*** Guest6723 has quit IRC17:23
*** CTtpollard has joined #baserock17:24
*** edcragg has quit IRC17:24
jjardonHi, is there something special about the build-system-* systems? I updated a component (syslinux) and while the resulting base-system-x86_64-generic has the correct version (6.03), using exactly the same changes but building a build-system-x86_64 , it shows a syslinux version 4.x in the resulting system. Any idea what is going on?17:57
*** zoli__ has joined #baserock18:12
*** lachlanmackenzie has quit IRC18:29
pedroalvarezjjardon: odd18:50
pedroalvarezjjardon: looks like they have syslinux duplicated18:52
jjardonah!  strata/bsp-x86_both-tools!18:53
pedroalvarezWas the upgrade painful?18:53
pedroalvarezjjardon: yes, I think it was created for the chroot systems18:54
jjardonI guess this is a good moment to request a review to :)18:54
pedroalvarezWe should move the common chunks out of the generic bsps18:54
jjardonpedroalvarez: not much, syslinux now supports efi apart from bios, so you have to specify that in the build commands18:54
pedroalvarezjjardon: we have been avoiding it18:55
jjardonpedroalvarez: you mean efi?18:56
pedroalvarezUpgrading syslinux19:10
pedroalvarezOh, I can't log in in Gerrit from here to vote..19:11
jjardonah! yeah the version we are currently using is quite old. I also think some btrfs features have to been disabled because they were not supported in our version as well19:12
jjardonI built and run a build and a base systems without problems. Is that enough or should I build a system with the built system? Not sure if syslinux is used when building the images19:18
*** zoli__ has quit IRC19:34
paulsherwoodKinnison: master ybd on build-system does *not* calculate cache-key for erlang on build-system, for me?19:40
*** zoli__ has joined #baserock19:51
paulsherwood(master ybd, master definitions)19:52
pedroalvarezjjardon: you should test that the built system boots,  I guess20:03
jjardon pedroalvarez: yes, I built a -build and a -devel systems, both boot and run20:10
Kinnisonpaulsherwood: perhaps you have made changes since the last time I pulled (Fridayish)20:35
*** zoli__ has quit IRC21:23
jjardonmason is red: I can read "cc: internal compiler error: Killed (program cc1)" in the logs: I think the ci machine needs more RAM22:51
rjekCI machines always need more RAM22:53

Generated by 2.15.3 by Marius Gedminas - find it at!