IRC logs for #baserock for Friday, 2014-09-12

*** dabukalam [] has joined #baserock00:04
*** dabukalam [] has quit [Ping timeout: 272 seconds]01:34
*** dabukalam [] has joined #baserock01:50
*** dabukalam [] has quit [Ping timeout: 245 seconds]03:30
*** dabukalam [] has joined #baserock03:45
*** dabukalam [] has quit [Ping timeout: 260 seconds]05:27
*** dabukalam [] has joined #baserock05:42
*** dabukalam [] has quit [Ping timeout: 245 seconds]07:07
*** inara [~inara@] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]07:28
*** inara [~inara@] has joined #baserock07:31
*** franred [] has joined #baserock07:53
pedroalvarezgood morning!08:06
*** tiagogomes [~tiagogome@] has joined #baserock08:16
*** ssam2 [] has joined #baserock08:31
*** jonathanmaw [] has joined #baserock08:31
*** violeta [] has joined #baserock08:44
* ssam2 notices the 'diff options' feature of cgit for the first time08:54
Kinnisonssam2: :-)08:55
franredssam2, cheers10:13
richard_mawssam2, franred: `sys.version_info < (2, 7, 3)`10:23
richard_mawstrictly, if we're going to monkeypatch fix this bug, we'd have to also check if the version were less than 3.3, but we don't have any versions of morph running that10:28
ssam2radiofree: branch-from-image is for 'I have this system image and nothing else'10:28
richard_mawand if we were, we'd probably jump to 3.410:28
franredrichard_maw, if the version is less than 2.7.3 does not includes also 3.3 and 3.4?10:29
* franred thinks that maybe he is missing something10:29
richard_mawssam2: it's not 100% useful for that currently, as we'd have to do some complicated logic to reconstruct the build graph from the contents of /baserock10:30
richard_mawbut if the branch still exists, it can pre-petrify refs10:30
richard_mawthough currently its sole remaining use is to find the branch for you10:30
* radiofree is still confused10:30
radiofreewhy would i use that over checkout, or branch/10:31
ssam2radiofree: if you've forgotten the name of your branch, perhaps :)10:31
ssam2I think it was more use back when the 'ref' field pointed at named refs10:32
ssam2so you'd have your release branch but it'd just say ref: 'master' or ref: 'stable'10:32
ssam2and you wouldn't know what actual commits those refs pointed to at the time10:32
radiofreemorph branch-from-image baserock:baserock/morphs baserock/release/baserock-9 - morph checkout baserock:baserock/morphs baserock/release/baserock-910:32
radiofreewhat's the difference between those commands?10:33
ssam2one won't work :)10:33
ssam2well, it will10:33
ssam2branch-from-image by default will recreate the branch /for the system you are running it in/10:33
pedroalvarezfranred: maybe the thing is that in python 3.2 the bug is also there?10:34
ssam2think of it as 'morph branch-from-image --metadata-dir=/baserock branch-for-the-system-im-running10:34
* radiofree is still confused10:35
ssam2:( the command clearly has issues if it doesn't even make sense to people10:35
* radiofree is still confused10:35
radiofreeso if i'm running a baserock system, branch-from-image will give me all the things i need to rebuild that exact image?10:36
radiofreeah ok, that's pretty useful!10:36
radiofreebut doesn't work anymore?10:36
ssam2richard_maw has fixed it10:36
ssam2might not be in master yet10:37
pedroalvarezssam2: I see in the lorries log that you merged the glibc lorry11:28
pedroalvarezI can't see it in the trove :(11:28
pedroalvarezalso, we modified some days ago the gusb lorry, and is still not there11:30
ssam2this is all true11:37
ssam2I understand there was an upgrade of which couldn't be done to fix an issue with lorrying11:37
ssam2perhaps that is related11:37
richard_mawI'll see what I can dig up after I've gotten this branch-from-image patch merged11:41
pedroalvarezrichard_maw: thanks!11:43
straycatIs the public facing mason supposed to be operational?12:24
pedroalvarezstraycat: what do you  mean with "operational"?12:31
pedroalvarezIt's working, but is not deploying images and testing them12:32
pedroalvarezso is a very-basic version12:33
straycatIt hasn't passed since the 8th12:34
fay_has anything been committed since then ?- it still seems to be checking for updates12:36
straycatAm I looking at the wrong thing?12:36
pedroalvarezyeah, nothing new in definitions.git  master since 4 days ago12:37
richard_mawthere's pending morph updates if anyone wants to do it12:37
straycatIs that the right thing?12:37
pedroalvarezstraycat: yes, it is12:37
fay_straycat, bottom right of the screen shows the last time it checked - it will only build and show a pass if something changes12:38
straycatSo "Failed to contact" is fine?12:38
pedroalvarezis "fine"12:38
pedroalvarezthat means that mason couldn;t reach g.b.o to check if there was something new to build12:39
straycatSo if it can't reach gbo to check, how can it be working?12:39
SotKit checks once a minute I think, so the connection failures are intermittent12:40
fay_it has managed to contact gbo since the fail so it was an intermittent network issue12:40
pedroalvarezthe thing is, that if there isn't anything new to build, then mason is not going to print a new row in green/red12:40
SotKI assume the long outage on Wednesday was because of the attempted upgrade of g.b.o?12:43
pedroalvarezwe shut down g.b.o. to do a backup of the disk, so yes12:45
franredrichard_maw, ssam2, the hack in morph to fix tarfile.chown works great, tvm for your help :) - I will send this patch soon12:46
petefothfranred: does that mean we will be able to deploy gerrit ?12:46
straycatfay_, *nod* but that interface doesn't tell me that12:46
franredpetefoth, ;-)12:47
franredpatches will be sent soon too12:48
Kinnisonfranred: and upgrading a system which has gerrit on it?12:48
petefothfranred: but I can't login12:48
KinnisonPresumably you don't have a user on it12:49
paulsherwoodfranred: if that's official, can we get it hooked up to please?12:49
KinnisonAlso, that IP is a local IP, so noone here apart from those inside Codethink's network will have access12:49
franredpetefoth, probably you can login with an openID, but Im gonna shutdown it12:49
franredpaulsherwood, it is still in testing time - it needs more configuration before publishing it12:50
paulsherwoodfranred: ok. what's the eta? i can barely wait!12:50
franredKinnison, what do you mean by "upgrading a system which has gerrit on it"?12:50
* petefoth googles about how to use OpenID12:51
KinnisonImagine I deployed a baserock system with gerrit on12:51
KinnisonI now want to upgrade it12:51
Kinnisondoes your work allow for that?12:51
* Kinnison knows we've seen a lot of work recently because Troves somehow became non-upgradeable, so I don't want to risk that12:51
franredpetefoth, until now you only can use yahoo accounts ;-)12:51
paulsherwooddoes franred's work involve storyboard and gerrit on one machine, or separate ones?12:52
franredpaulsherwood, no storyboard yet12:52
petefothfranred: Ah yes - now I remember. Will we be able to login using Codethink credentials soon?12:52
KinnisonI believe we're aiming for OpenID12:52
Kinnisonbecause infra needs something open12:52
petefothpaulsherwood: gerrit only I believe12:52
paulsherwoodpetefoth: this will be a public system... not codethink :)12:53
franredpetefoth, it is possible to configure LDAP but as Kinnison said we aim to use OpenID12:53
petefothpaulsherwood: good point!12:53
* petefoth goes to check what his OpenID is 12:53
Kinnisonpedroalvarez: regarding your comment on the empty trovehost12:54
Kinnisonpedroalvarez: You'd need to remove the entire trove stanza from the lorry-controller.conf12:54
franredKinnison, I can't see anything which implies that this system is not upgradable but I will have a look at it12:54
Kinnisonpedroalvarez: but regardless, (and anything you upgrade) will already have a lorry-controller.conf12:54
Kinnisonpedroalvarez: should you be touching it?12:54
Kinnisonfranred: well there's all sorts of databasey goodness12:54
Kinnisonfranred: can gerrit handle upgrading its DB if it finds it's an older version, or is there a migration process which will be needed?12:55
Kinnisonif there's a migration process, is the DB in a shared path12:55
straycatI don't know anything about UI, but I think having a "status" at the top of the page would make things clearer.12:55
Kinnisonstraycat: If you want to criticise Gerrit's UI then you need to get in line with the gajillion other people out there :-)12:55
Kinnisonstraycat: There's a bunch of cmdline tools to mitigate its awfulness12:55
franredKinnison, I will dig into it12:56
Kinnisonfranred: :-)12:56
franredKinnison, thanks ;-)12:56
fay_Kinnison, I think straycat is critising mason rather than gerrit12:56
Kinnisonfranred: I'm not saying it needs to be solved before the first review, but if that review cover-note contained indications of where we need further work, it'd be a good idea12:56
Kinnisonfay_: aah12:56
tlsathere's room for improvement (mason UI)12:57
franredKinnison, this second review does not includes the final configuration - it does only includes the PoC of Gerrit working on baserock12:57
Kinnisonfranred: aye :-)12:57
tlsabut afaik, the plan is to replace that with something else12:57
Kinnisonfranred: just note the weaknesses (or even the known unknowns :-)12:57
tlsa"off the shelf"12:57
KinnisonWell, we want to use zuul, gearman and turbo-hipster if appropriate12:57
Kinnisonbut we'll likely still need something to frontend it12:58
KinnisonNo idea what that'll be yet12:58
franredKinnison, +1 to that latest sentence12:58
pedroalvarezKinnison: no, but what about if you want to deploy a new, a different one, without an upstream trove12:58
franredbut we will work out something ;)12:58
ssam2somehow I've built a devel system based more or less off master which fails to even get to login12:59
ssam2i'm reduced to setting init=/bin/bash and trying again, because systemd emergency mode doesn't work either12:59
ssam2hopefully this is just something stupid I have done!12:59
Kinnisonpedroalvarez: aah yes, deploying a brand new standalone trove will need the entire stanza to not be present13:00
Kinnisonpedroalvarez: perhaps switch ansible so that it substitutes an entire stanza in, rather than just the hostname?13:00
pedroalvarezKinnison: I'm currently thinking about it13:02
ssam2oh, seems my mistake was to set 'console=tty0 console=ttyS0' on the kernel commandline13:05
ssam2only the serial console lets me log in, but I can't access it, either on our OpenStack host, or on the KVM machine I've tried it on13:05
ssam2I got that from here:
ssam2has anyone used the 'KERNEL_ARGS: console=tty0 console=ttyS0' line successfully in a cluster morph? or should we not be advising people to do that?13:06
pedroalvarezssam2: I wrote that, and it worked for me :S13:07
ssam2so the problem is that my machine goes into emergency mode, which opens a login prompt on ttyS0, which I can only read from13:12
ssam2putting ttyS0 first and tty0 second seems to give me emergency mode on the VT13:22
ssam2i'll change the instructions in the wiki unless anyone objects13:22
pedroalvarezssam2: please, do13:30
franreddo you know how to make git send-email works in baserock? it seems like it is trying to use busybox sendmail and it fails:
pedroalvarezseems like git-send-mail is not compatible with busybox sendmail13:46
paulsherwoodfranred: this *used* to work -
franredpaulsherwood, cheers, I will test it13:54
paulsherwoodfranred: how's gerrit looking?13:54
franredpaulsherwood, it needs to spend time in the configuration part - it needs database configuration (to think: backups and upgrade), administrator (number of admins, repositories, group and users), security...13:57
franredI want to play with it on monday adding repositories, users, rules and see how does it work13:59
*** genii [~quassel@ubuntu/member/genii] has joined #baserock13:59
straycatWhere is our mason hosted?14:43
jjardonHi, I think I found a bug: "ERROR: Ref 1.12 for cairo in stratum strata/graphics-common.morph is not a string" If I change it to 1.12.16 it doesnt complain14:43
Kinnisonnot a bug14:44
Kinnisona yaml parse quirk14:44
Kinnison1.12 is a number14:44
richard_mawjjardon: either that or quote it so it's "1.12"14:44
Kinnison1.12.16 is not14:44
* ssam2 has a Baserock container running in Docker in a Baserock VM14:45
ssam2running in OpenStack14:45
Kinnisonmuch virtualisation14:45
* straycat sighs14:45
richard_mawwe're just missing the OpenStack running on Xen14:45
Kinnisonssam2: OOI is docker using systemd-nspawn to make the container?14:46
jjardonKinnison: then I guess morph should support numbers in the ref parameter? A branch name can be a number14:46
richard_mawKinnison: no, but now we can see if nspawn works inside docker14:47
richard_mawssam2: ^14:47
ssam2Kinnison: docker has multiple backends, by default it does its own thing14:47
ssam2it used to wrap LXCV14:47
KinnisonI see14:47
Kinnisonjjardon: Potentially lossy conversion14:47
Kinnisonjjardon: consider 1.1014:47
Kinnisonjjardon: better to whinge and require quoting14:47
ssam2Kinnison: a backend for BSD Jails is also proposed or maybe even implemented14:47
pedroalvarezstraycat: we have a openstack tenancy hosted in
*** fay_ [] has quit [Remote host closed the connection]15:02
*** tiagogomes [~tiagogome@] has quit [Ping timeout: 252 seconds]15:03
*** genii [~quassel@ubuntu/member/genii] has quit [Read error: Connection reset by peer]15:32
*** genii [~quassel@ubuntu/member/genii] has joined #baserock15:37
pedroalvarezhas anybody ever hit his error?
pedroalvarezcontext: running morph in my Raspberry Pi after the native-bootstrap script has finished15:57
richard_mawpedroalvarez: yes, your kernel wasn't built with namespace support15:57
Kinnisonclone: Invalid argument15:58
Kinnisonthat's the important bit of the error15:58
pedroalvarezKinnison: yeah, and I knew someone has hit this error no long time ago15:58
Kinnisonradiofree: that's the configure output from qtwayland15:58
pedroalvarezrichard_maw: thanks15:58
Kinnisonradiofree: and is the full text of the error15:59
*** franred [] has quit [Quit: Leaving]16:01
straycatrjek, have you seen this scripts/basic/fixdep not found thing?16:01
Kinnisonstraycat: yes, that's the need for a fakebash, no?16:01
rjekYes, fakebash16:02
rjekI really want to be able to specify which compiler an architecture requires, rather than trying to fudge patches from fringe mailing lists into the GCC we're using.16:03
* Kinnison oopses and realises he used a private pastebin on a public channel, sorry about that chaps16:03
*** tiagogomes [] has joined #baserock16:03
straycatmy "bash" is a script that runs sh16:03
*** tiagogomes [] has quit [Ping timeout: 245 seconds]16:20
*** tiagogomes [] has joined #baserock16:35
*** tiagogomes [] has quit [Ping timeout: 255 seconds]16:40
*** jonathanmaw [] has quit [Quit: Leaving]16:46
straycatI don't see what this has to do with the bash hack, it's more that if I compile a file within the chroot I can't execute it.16:52
Kinnisonoh is fixdep a compiled thing, hmm16:52
rjekAlso check that you have an appropriate in an appropriate place.16:52
Kinnisonstraycat: yeah, it's most likely that the rtld is in the wrong place16:53
Kinnisonstraycat: what platform are you trying to bootstrap?16:53
KinnisonOooh mips is a total pain16:53
KinnisonBut it's likely to be the RTLD16:54
*** franred [] has joined #baserock16:54
Kinnisoncan you readelf the generated fixdep?16:54
Kinnisonand see what interpreter path it has baked in?16:54
Kinnisonat this point, it *should* be the /tools path16:54
rjekstraycat: OOI, what flags do you pass to your bootstrap compiler in terms of architecture and ABI?16:56
straycatKinnison, It's /lib/
straycatrjek, I haven't specified anything outside of whatever morph uses by default. I've just patched morph to map mips32 to mipsel17:00
straycatgcc -dumpmachine for the current os gives me 'mipsel-linux-gnu' morph's default (with a mapping to mipsel) should work17:16
KinnisonI'm afraid I don't recall all the mips stuff rjek did, also it was for mips64 so I don't know how applicable it was17:18
KinnisonI do know that mips, in general, is very poorly supported in the compilers that we have17:18
Kinnisonoften requiring patches to get it to behave right17:18
* Kinnison wishes he could help more, but is out of memory :-(17:19
*** franred [] has quit [Quit: Leaving]17:19
straycatOkay, I'll read around and see if anyone's had similar problems17:21
*** ssam2 [] has quit [Quit: Leaving]17:22
*** violeta [] has quit [Ping timeout: 276 seconds]17:42
*** ridgerunner [] has quit [Remote host closed the connection]18:02
*** benbrown1 [] has quit [Read error: Connection reset by peer]18:02
*** SotK [] has quit [Read error: Connection reset by peer]18:02
*** paulsherwood [] has quit [Read error: Connection reset by peer]18:02
*** petefoth [] has quit [Read error: Connection reset by peer]18:02
*** jmacs [] has quit [Read error: Connection reset by peer]18:02
*** tiagogomes [] has joined #baserock18:25
*** tiagogomes [] has quit [Ping timeout: 245 seconds]19:01
*** tiagogomes [] has joined #baserock21:05
*** tiagogomes [] has quit [Quit: Leaving]21:29
*** genii [~quassel@ubuntu/member/genii] has quit [Ping timeout: 240 seconds]22:13
*** genii [~quassel@ubuntu/member/genii] has joined #baserock22:15
*** flatmush [] has quit [Ping timeout: 255 seconds]22:29
*** flatmush [] has joined #baserock22:30
*** persia__ [quassel@2400:8900::f03c:91ff:feae:3452] has joined #baserock22:30
*** juergbi [] has joined #baserock22:40
*** ratmice___ [] has quit [Ping timeout: 246 seconds]22:42
*** flatmush [] has quit [Ping timeout: 246 seconds]22:43
*** flatmush [] has joined #baserock22:43
*** ratmice___ [] has joined #baserock22:57
*** juergbi [] has quit [Ping timeout: 252 seconds]23:13
*** juergbi [] has joined #baserock23:15
*** persia_ [quassel@2400:8900::f03c:91ff:feae:3452] has joined #baserock23:41
persia_ is now known as Guest2997823:42

Generated by 2.15.3 by Marius Gedminas - find it at!