IRC logs for #baserock for Monday, 2015-03-16

*** zoli__ has joined #baserock04:45
*** zoli__ has quit IRC05:15
*** zoli__ has joined #baserock05:40
*** mike has joined #baserock07:08
*** mike is now known as Guest4731507:08
*** petefoth has joined #baserock07:27
*** rdale has joined #baserock07:50
*** a1exhughe5 has joined #baserock08:01
*** fay_ has quit IRC08:30
*** fay_ has joined #baserock08:31
bjdooksHi, who do I need to prod to get a latest armv7be image?08:35
bjdooksbaserock-8-armv7a-be.tar.bz2 last uploaded 15jul201308:35
persiabjdooks: I think you need someone from ops, but I think they stopped uploading things for which there were not public distbuild clusters & cache.08:36
*** CTtpollard has joined #baserock08:36
persiaI know that at least a couple of them have some Jetsons around, so it might be possible for one of them to bring it up to date.08:37
persiaActually, it's probably important to do that: I don't think that a tarball that old can be used to build a current one.08:42
bjdookstftpboot 0x3000000; tftpboot 0x3800000 zynq-zybo.dtb; bootm 0x3000000 - 0x380000008:54
*** jonathanmaw has joined #baserock08:55
persiaIs that the BE environment booting on the Zynq hybrid ARM/FPGA board?08:55
persiaNifty!  I've seen a couple designs using that chip, but nothing with a traditional userspace.08:56
bjdooks 4 files changed, 25 insertions(+), 17 deletions(-)08:56
bjdooksof which part of that is already in net-next08:56
bjdooksLinux vexpress-tc2 4.0.0-rc3-00003-gfac9018 #15 SMP Mon Mar 16 08:43:14 GMT 2015 armv7b GNU/Linux08:58
bjdooksjust rebased to -rc408:59
*** bashrc has joined #baserock09:02
*** mdunford has joined #baserock09:07
*** gary_perkins has joined #baserock09:23
*** gfinney has joined #baserock09:33
wdutchI've just made a clean definitions branch but it's failing at stage1-gcc build-commands09:56
*** ssam2 has joined #baserock09:57
*** ChanServ sets mode: +v ssam209:57
persiawdutch: Could you share a bit more context?10:00
wdutchpersia: I'm trying to build base-system-x86_64-generic, it goes splat at stage1-gcc build-commands, should I paste the whole output?10:04
*** lachlanmackenzie has joined #baserock10:08
*** Krin has joined #baserock10:10
ssam2wdutch: just the last 100 or so lines would be fine10:12
ssam2what you have there is a series of messages from nested 'make' instances going 'something failed in a subprocess', but it doesn't contain the original error message10:13
radiofreewhat happens to the build log when a chunk fails?10:13
*** edcragg has joined #baserock10:13
ssam2it goes in morph.log with something like 'ERROR LOG FROM FAILED BUILD' prefixed to each line10:13
*** pacon has joined #baserock10:15
radiofreethat's a bit unwieldy, could it also be copied to the failed staging area?10:16
ssam2no reason why not10:16
ssam2wdutch: looks like `makeinfo` is failing with an error10:17
ssam2i've no idea why. maybe we updated it recently?10:17
ssam2seems we updated texinfo on 3rd feb, but surely we'd have hit this issue already if that was the cause10:18
radiofreei recompiled stage1-gcc several times last week, on arm though10:20
rdalemaybe the version of texinfo on the host hasn't been updated since then?10:21
radiofreei have 5.2 on my 15.10 host10:22
pedroalvarezI wonder why anyone making a clean definitions branch is building gcc-stage110:22
persiawdutch: what are you using as a cache server?10:23
wdutchpedroalvarez: maybe I'm using that term wrongly, I've just checked out definitions and hav made no modifications10:23
wdutchpersia: how can I find that out?10:24
persiaLook at your morph.conf10:24
SotKwdutch: check if is artifact-cache-server set in your morph.conf?10:24
wdutchartifact-cache-server =
persiaThat makes it especially strange that you would need to rebuild that, unless something else is odd.10:25
petefothwdutch: or /src/morph.conf if you followed the steps in
persiawdutch: Try building an unmodified reference system.10:25
SotKwdutch: are you using latest morph?10:26
straycati think connectivity issues might be involved10:26
straycatmorph only attempts one cache lookup, if that fails then it will fallback to a build10:26
* straycat got a ERROR: <urlopen error [Errno 111] Connection refused> this morning whilst morph was doing a cache lookup10:27
pedroalvarezthere are log's entries of someone trying to fetch stage1-gcc, but a version that is not present in the cache server10:28
straycatthen again, if it's connectivity it would likely just work on a second attempt10:30
* SotK wonders if wdutch is using an old version of morph10:30
wdutchI'll try reinstalling morph10:31
pedroalvarezwdutch: these instructions would be enough:
wdutchpedroalvarez: thanks10:33
pedroalvarezanyway, if morph is not complaining about missing "build-depend" fields, then is a recent enough version of morph10:33
pedroalvarezor an old version of definitions10:33
persiaWhat could have happened to morph that upgrading would fix it?10:33
persiaBecause for nearly any answer, we ought consider that a bug, and revert it.10:33
SotKcache key algorithm may have changed, but I doubt it since its a while since it did10:34
persiaYes, but we should revert that.10:34
persiaWe should never change it without changing the key in the code to indicate it changed.10:34
SotKwhich key is that?10:36
* SotK is probably confused10:37
persiaNo, rather, I haven't been following the code closely.10:39
persiaAt one point in the past, we had a version string for the way we calculated cache keys, which I thought was in CacheKeyComputer, but looking at the code in blob c3a01b9ed87b6db90c5b93909421d335c2e127bc , I don't see it anymore.10:40
ssam2there is a versioning field *in* the cache key10:41
ssam2but changing that causes cache keys to change. there's nothing at present that could tell a user 'warning: cache keys are different now'10:42
ssam2i'm not sure how that would even work10:42
paulsherwoodme neither10:46
*** pacon has quit IRC10:47
ssam2persia: I agree that it'll be worrying if upgrading morph causes wdutch's problem to magically go away, though10:47
wdutchit didn't10:47
*** pacon has joined #baserock10:48
ssam2wdutch: if you've got time to debug, try cloning gcc-tarball.git somewhere in your baserock devel system and building it manually10:49
ssam2using the same configure and build-commands that the stage1-gcc.morph uses10:49
ssam2if it that succeeds, then morph is doing something weird. if it fails, something is broken in your baserock system (and possibly all of them)10:50
ssam2or something is up with stage1-binutils, but given that it's 'makeinfo' failing, that seems unlikely10:51
wdutchI'll give it a go10:51
persiaIt would be really cool if morph had a subcommand to process a chunk morphology against the current working directory to ease this sort of replication.10:51
petefothpersia: why not create story for that?10:52
persiaBecause I have a very hairy yak standing between me and today :(  But I'll put it on my list.10:53
*** pacon2 has joined #baserock10:59
mwilliams_ctssam2: I'm having a look at the openid error when used with storyboard today ("I could not reach..."). Trying to deploy a development instance, but packer seems to take a long time "pulling docker image fedora 20". is this expected?11:01
mwilliams_ctlong time ~30 minutes and still going11:02
*** pacon has quit IRC11:02
ssam2cool, thanks for looking it11:02
ssam2I've kind of stopped using 'docker' for developing these systems11:02
pedroalvarezwdutch: I'd be intereseted on the sha1 of definitions.git you are using, and `morph --version` sha111:02
ssam2mwilliams_ct: I just deploy them straight to openstack now11:03
ssam2so might be easier for you to do the same11:03
mwilliams_ctssam2: OK, I'll give that a try thanks. I'll try and add a separate patch to README.mdwn if that goes well while I'm at it11:03
ssam2also, Packer is quite fiddly, feel free to work around it. I thought it might be better than `morph deploy`, but it's actually just equally annoying in different ways11:03
ssam2thanks, updating the readme would be cool (just removing the 'development' instructions might make sense)11:04
ssam2also, note that infrastructure.git embeds a copy of django_openid_provider and has a couple of fixes that aren't upstream. this isn't ideal, but upstream doesn't seem very active and there are several equally inactive forks, so I've not been too motivated to persue upstreaming things so far11:07
*** jonathanmaw has quit IRC11:07
mwilliams_ctmuch obliged, I'll poke you if I get stuck (but hopefully not too much ;) )11:07
*** jonathanmaw has joined #baserock11:09
jjardonHey, in case anyone is interested, I've just found this app to track our gerrit instance from your phone: Mgerrit11:23
wdutchfrom my baserock VM in kvm I can git clone and ssh but I can't ping the outside world, is this normal? I set it up following this guide
richard_mawwdutch: running it directly with kvm like that blocks the ping protocol.11:31
* richard_maw generally recommends the virt-manager approach instead of direct kvm11:31
*** XuLiu has joined #baserock11:32
rjek 111:34
*** CTtpollard has quit IRC11:44
*** CTtpollard has joined #baserock11:46
*** wschaller has joined #baserock12:04
jjardonssam2: some patches in gerrit already have a +2; is there anything I can do to merge them?12:33
ssam2when you say +2, what do you mean?12:35
ssam2a +2, or multiple +1's ?12:35
ssam2in Gerrit's world 1+1 doesn't make +212:35
*** pacon2 has quit IRC12:37
jjardonssam2: +212:42
jjardon(given by you IIRC)12:42
ssam2oh, you should just be able to click the 'Submit' button12:43
ssam2if you see it12:43
ssam2it's hidden if you can't click it12:43
jjardonoh, ok, I just found it, thanks!12:45
*** a1exhughe5 has quit IRC13:03
*** mdunford has quit IRC13:17
*** a1exhughe5 has joined #baserock13:18
*** mdunford has joined #baserock13:21
straycatssam2, thanks for merging the fix, though i hadn't done the fixup richard_maw suggested, in practice it can't be triggered but probably still worth making the fix14:00
ssam2oh, thanks for pointing that out14:02
straycatas for the git@ thing, it works without it, but should also work with, i have a habit of specifying it because you can't tell from the url whether it's a mercurial repo or a git repo14:02
ssam2ok. seemed not to work for me, but i didn't leave it very long14:02
* straycat probably should have acted sooner but was a little preoccupied14:02
ssam2i'll do the msg= fix directly in master, since i was going to do it at merge time anyway14:03
straycat:) thanks14:03
straycat*because otherwise14:05
richard_mawstraycat: in practice it's very unlikely to be triggered, but you could trigger it by having a filename with a % character in it, which is perfectly legal14:15
straycatahh yes that's true :/ forgot we had filenames in the key14:22
straycatsomeone could put a % in the repo too if they wanted14:25
richard_mawprobably ref too14:30
straycatahh but it's a sha by that point i think14:30
ssam2krin: does zookeeper require Gerrit ?15:02
ssam2oh, never mind15:02
ssam2was wondering why github:franred/gerrit-installation-binaries appeared in the zookeeper stratum, but it's because it needs JAva15:03
franredssam2, yep, it is just for that15:04
ssam2franred: i'm planning on sending a patch to remove gerrit from definitions.git, I hope you don't mind :)15:04
ssam2i don't think we need it now it's in infrastructure.git15:04
franredssam2, I don't mind, but java is required for zookeeper so I imagine you are not going to be able to rid off the github reference15:05
ssam2yeah :(15:05
franredunless until we are able to bootstrap Java15:05
* ssam2 discovers that Gerrit lets you easily diff the second version of a patch against the first version of a patch15:07
petefothMy '[PATCH] Add link to Import Tool wiki pages' patch for Import Tool now has 2 * +1s. Do I need to do anything further to get it merged (other than wait for someone with commit access to get a round tuit)?15:11
ssam2just remind people to do it15:14
ssam2i'll do it15:15
ssam2perhaps you can thank me by submitting your next patch through instead15:15
petefothssam2: thanks. My post was intended as a 'reminder' :) Registering with Gerrit is on my to-do list (submitting more patches *isn't yet though).15:17
petefothHow hard would it be to allow gerrit to be used (optionally) for reviewing possible changes to I think that could be useful if we wanted to have a tracked discussion about proposed 'specification' pages such as the one I sent recently. I think that gerrit is a better way of tracking such discussions than the ML (though I recognise that others may not agree).15:20
CTtpollardIt sounds possible15:22
*** zoli__ has quit IRC15:27
*** zoli__ has joined #baserock15:27
bashrcsearching for a missing cluster: baserock-firehose/baserock-firehose.morph15:28
pedroalvareznothing like that in master. Maybe in another branch/repo?15:33
*** wschaller_ has joined #baserock15:36
bashrcappears here: perhaps it was never added
ssam2petefoth: not impossible, we could use gerrit-replication to push changes to branchable after they're reviewed in Gerrit15:38
*** wschaller has quit IRC15:38
persiaThe difficulty with using gerrit for wiki edits is that there's a good chance some other wiki edit happened (e.g. from a browser) in the meantime, potentially causing merge issues.15:40
petefothssam2: I think it would be worth doing but, even if others agree, I don't think there's any great rush. It can probably wait until the next time the OPs team don't have enough to do :)15:40
persiaIf we did do that, I'd prefer no merge automation: just using gerrit for discussion and manual merge (unless we're willing to remove all web-edit features).15:40
ssam2when you mean 'merge automation', do you mean disable the 'Submit' button in Gerrit?15:41
ssam2or something else?15:41
petefothpersia: I don't think it's often that people will be changing the same parts of the same pages, but I may be wropng about that. I don't feel strongly about the merge, but I think you are correcxt that it shouldn't happen automatically on receprt fo a +215:41
persiassam2: I mean anything that causes gerrit to try to push a merged branch to branchable.  The Submit button is part of that.15:42
petefothI think we *should* have a 'Submit' button, so that gerrit can do the grunt work of the merge15:42
persiapetefoth: I agree it's not often, but I think the chance is increased if there is an ongoing review for a couple days.  And although rare, I have had to resolve merge conflicts with the baserock wiki in the pat.15:42
petefothnut I don't hink we need to agree on that now :)15:43
bashrcI think the final decision would always be manual15:43
persiaAs long as we allow non-reviewed changes, I agree.15:43
petefothbashrc: agreed15:43
petefothpersia: oh definitely! I would only envisage using it for stuff where the author thinks "we really need some agreement on this before it goes live'15:44
persiapetefoth: Makes sense.  I just don't trust automation to get merges right in that sort of timeframe.15:45
petefothpersia: I agree - technology? Just say 'No!' ;)15:47
*** zoli__ has joined #baserock16:03
jjardonssam2: thanks for the reviews16:41
jmacsI can't seem to sign into gerrit using my Google account; should that work?16:42
ssam2no, it's limited to
ssam2richard_maw: how do you feel about jjardon's v3 simple-network patch? it'd be really handy to have it merged for some work I'm doing16:53
* jjardon finds gerrit much better to send/review different versions of patch series16:53
persiajjardon: The main complaint about this when gerrit was first proposed was the need for Change-ID:, and that it meant that if you pushed, you needed to update your local copy to gain the Change-ID:.  How do you find this workflow?16:54
richard_mawssam2: ok, I'll take a look16:56
wdutchI've heard that somebody made a stratum of GNU utilities but I can't find it in definitions or an obvious branch, does anybody remember anything about it?16:56
jjardonpersia: there is a hook that do that for me, so I basically dont care about the Change-ID line :)16:56
jjardonwdutch: coreutils?16:56
persiajjardon: Heh, cool.16:57
*** XuLiu has quit IRC16:59
wdutchjjardon: if that's it it's not what I need, I was wondering if there was a GNU grep, I git grepped GNU grep but didn't see anything, I was told it might be helpful16:59
mwilliams_ctwdutch: are you just going for maximal alliteration there?17:00
wdutchmwilliams_ct: I think I broke out into jazz improve there17:01
wdutchlol at herbal aromas17:02
pedroalvarezwdutch: do you really need GNU grep?17:02
wdutchpedroalvarez: I do it's a dependecy for what I'm building17:03
pedroalvarezif so, you can try to send a patch to add it to coreutils?17:03
wdutchwill do17:03
*** a1exhughe5 has quit IRC17:03
pedroalvarezlooks like we have it already lorried :
wdutchpedroalvarez: that one has a lot of prereqs, I was hoping to mirror the tarball which has fewer17:04
jjardonwdutch: in general, Its prefereable to lorry (and build) from git repos instead tarballs17:05
* jjardon remembered we should move tar to the coreutils stratum as well17:06
richard_mawjjardon: biff17:07
richard_mawssam2: ^ I guess17:07
jmacsOK, got logged into gerrit now. Is the idea that replies on gerrit will get emailed to baserock-dev eventually?17:08
jjardondont think so. To the author and the people that is watching the specific project only17:09
ssam2hmm, not to baserock-dev17:09
ssam2it doesn't send any emails right now though, which is a known issue.17:09
ssam2would be too noisy to send them to baserock-dev I think17:10
ssam2thanks richard_maw! jjardon: do you have time to do the fixes richard asked for?17:11
ssam2i can maybe tomorrow if not17:11
jmacsDoesn't that mean baserock-dev contains a partial view of the patch's progress, then?17:11
* richard_maw has applied his usual disclaimer that he doesn't require nit-picks to be taken care of before merge17:12
jjardonssam2: I have17:13
pedroalvarezjmacs: well, gerrit won't contain patches anymore when we move to it (if we move)17:15
ssam2you mean baserock-dev won't contain patches any more17:15
ssam2jmacs: yes, patch review is split between baserock-dev and now17:15
jmacsOh, that's fine then. I thought gerrit was going to continue alongside the mailing list.17:16
ssam2I hope not. I thought it'd be too disruptive to force everyone onto gerrit right away, but unless people strongly object i'll disable push access to* in a few weeks17:17
*** jonathanmaw has quit IRC17:23
*** gfinney has quit IRC17:28
*** gary_perkins has quit IRC17:31
*** ssam2 has quit IRC17:35
*** Krin has quit IRC17:43
*** bashrc has quit IRC17:52
*** Guest47315 has quit IRC18:06
*** rdale has quit IRC18:21
*** mdunford has quit IRC18:32
*** zoli__ has quit IRC18:36
*** edcragg has quit IRC18:41
*** mdunford has joined #baserock18:45
*** wschaller_ has quit IRC19:01
*** lachlanmackenzie has quit IRC19:02
*** mdunford has quit IRC19:25
*** zoli__ has joined #baserock20:01
*** zoli__ has quit IRC20:08
*** zoli__ has joined #baserock20:27
*** zoli__ has quit IRC21:30
*** gfinney has joined #baserock22:56
*** radiofree is now known as justache23:56
*** justache is now known as radiofree23:57

Generated by 2.14.0 by Marius Gedminas - find it at!