IRC logs for #baserock for Monday, 2015-03-23

*** mike has joined #baserock07:39
*** mike is now known as Guest525107:40
*** a1exhughe5 has joined #baserock08:10
*** gfinney has joined #baserock08:15
*** jonathanmaw has joined #baserock08:22
*** mdizzle has joined #baserock08:36
*** mdunford has joined #baserock09:01
*** bashrc_ has joined #baserock09:01
*** rdale has joined #baserock09:07
*** rdale has quit IRC09:17
*** rdale has joined #baserock09:19
*** mdizzle has quit IRC09:29
*** mdizzle has joined #baserock09:31
*** mdizzle has quit IRC09:36
*** mdizzle has joined #baserock09:37
*** franred has joined #baserock09:47
*** tiagogomes_ has joined #baserock09:49
*** CTtpollard has joined #baserock09:51
*** ssam2 has joined #baserock09:57
*** ChanServ sets mode: +v ssam209:57
*** gary_perkins has joined #baserock09:58
*** paulwaters_ has quit IRC09:59
*** paulw has joined #baserock10:03
ssam2wow, lots of patches in Gerrit now10:23
*** wschaller has joined #baserock10:23
*** flatmush1 is now known as flatmush10:28
CTtpollardis the stance still no to a bot that announces gerrit merges?10:29
richard_mawkejiahu: I've worked out what's gone on with your gawk problem. Upstream's latest release build is broken10:29
KinnisonSo we need to roll backwards on gawk a bit, or is there a fix in their repo?10:30
richard_mawgawk-4.1.0-367-gcd99e34 is the earliest point where it was fixed10:30
richard_mawKinnison: development head has a fix, but no release includes it yet10:30
kejiahurichard_maw: thanks, I will try that one10:30
straycatCTtpollard, i'd like one10:33
richard_mawkejiahu: in the absence of a proper release, it may even be better to pick the head of the gawk-4.1-stable branch10:33
straycatCTtpollard, i'd like one for submissions too10:34
radiofreebuilding an armv7b build-system, get this error with libusb
CTtpollardI think I brought it up before, but it was a mixed response10:37
radiofree"Too many mounts (maximum of 50)" ?10:37
richard_mawold version of linux-user-chroot10:37
radiofreeyeah i am building this on a baserock 8 system :\10:38
straycatCTtpollard, i would argue anyone who doesn't want to receive messages from a gerrit bot can put the bot on their ignore list10:38
radiofreei have bootstrapped a new armv7b system, perhaps i should start again from there10:38
CTtpollardstraycat: my view exactly10:38
radiofreerichard_maw: is still the version we're using?10:39
richard_mawradiofree: yes10:40
ssam2tpollard,straycat: good point actually, I hadn't thought of that and no longer object10:48
*** jonathanmaw has quit IRC10:56
*** Krin has joined #baserock10:57
*** jonathanmaw has joined #baserock10:58
* pedroalvarez fails to fetch a commit from a review in gerrit to test it11:00
SotKI think you need to use git-review to get the commits somehow11:02
SotKor fetch it from a weird ref11:02
ssam2but because 'git remote update' / 'git fetch' only pull refs/heads/* and refs/tags/* by default, you have to manually do `git fetch gerrit refs/changes/nn/nn; git checkout FETCH_HEAD -b change-nnn`11:14
ssam2or use git-review11:15
jjardonmmm, seems some commit has not be applied? (I do not see the xserver upgrade in definitions master) but gerrit said it did:
pedroalvarezodd and worrying11:28
* Kinnison doesn't like the sequence of reviews on that11:28
Kinnisonis that how Gerrit works?11:28
Kinnisoni.e. that the author gets to +2 their patch?11:29
pedroalvarezin this case, the author, which is also in the mergers group, acted as a merger and decided that the two +1's were valid, and then +2ed the patch and merged11:30
richard_mawKinnison: I don't think there's anything fundamentally wrong with that, since +2 is just the final "yes, please merge this bot!", and sometimes as a reviewee you want to review the reviews to see if you think they're appropriately rigorous11:30
richard_mawKinnison: but AIUI the intention is to add Zuul to do the +2 later anyway11:30
KinnisonAah I see11:31
Kinnisonso the two +1s, plus the absence of any -1s permitted the merger to push the "merge it" button which is recorded as a +2 ?11:31
*** edcragg has joined #baserock11:31
* straycat notices that large changes seem to be being submitted as single commits11:31
KinnisonI fear that's how gerrit works11:32
SotKI think the problem was that the "merge it" button doesn't appear until there is an explicit +211:32
KinnisonSotK: aah11:32
straycatWe do need to figure out how to submit series to it, otherwise I'd sooner use the ml personally11:32
*** mdunford has quit IRC11:33
SotKI don't think you can11:33
SotKGerrit's "patch series" support is the topic stuff11:33
pedroalvarezI've already seen "patch series" in gerrit11:34
kejiahurichard_maw: HEAD of 4.1 stable fixed the gawk issue, the build is carrying on to see if there is any more issue. thanks for your help11:35
richard_mawno worries kejiahu11:35
* richard_maw just attempted to tab-complete the phrase "no worries", since he uses it so often11:35
straycattopic stuff?11:38
* richard_maw just got linked to
SotKstraycat: the name in brackets in the Branch column of the changes page11:39
richard_mawthe rough idea appears to be "submit your patch series with this command-line tool and it'll submit all your changes and put dependencies in to string them together"11:40
* Kinnison worries that the default behaviour of gerrit is not to support patch series11:42
* Kinnison thought work had been done to allow them11:42
bashrc_gerrit isn't ideal. I don't know what the rationale was for using it11:44
KinnisonIt's better than everything we had before :)11:44
Kinnisonand it's already there11:44
richard_mawand OpenStack is using it11:44
Kinnisonand used by many projects, which means there's lots of engineers familiar with it11:44
* Kinnison is just a whingy git :)11:45
richard_mawyou know what they say, there's two kinds of tools, those which everyone complains about, and those which nobody uses11:46
*** mdunford has joined #baserock11:48
* straycat expected to be able to review a series from the cli with the git-review tool, and to view patches with the same topic in some sane way11:53
persiastraycat: Have you looked at gertty?11:57
straycatno i'll check it out thanks11:58
ssam2I think it's actually not too bad having to submit things as individual commits11:59
ssam2we have cases where someone submits a patch series to the mailing list and someone else goes "this is too big to review, please split it up into smaller parts"12:00
ssam2Gerrit kind of enforces splitting up big changes12:00
straycati think it's pretty terrible because it completely screws my ability to linearly explain what i'm trying to do12:01
ssam2it can makes things more awkward, yeah. Perhaps in that case, drop a cover letter to the mailing list listing all of the changes and explaining the order that they make sense12:03
bashrc_unfortunately the consequence of that could be people submitting large patches as a single commit12:03
straycatthat's basically every case, then12:03
persiaIndeed: for large things, sending mail to the list with explanations can be useful.12:03
straycatand putting it in the cover letter means it's not in the git history12:03
persiabashrc_: That ought be cause for -112:03
straycat(the explanation of my changes)12:03
ssam2straycat: cover letters already aren't in the git history12:03
ssam2and we use them12:03
persiastraycat: Yes, but if you use a topic, they should generally land together, in order.12:03
straycatssam2, i put the explanation in the patch12:04
ssam2then I don't see a problem...12:04
straycatthat's what a series allows me to do, to build up a linear explanation that goes into the git history for anyone to review and digest later12:04
ssam2you can submit a change to Gerrit that depends on another change12:05
ssam2first you push change A, then you push a branch with both change A and change B12:05
straycatsure, if a topic works and the gerrtty tool allows sane review of that then that's cool, but i don't agree with your initial statement that it's useful that we're being coerced into submitting things as individual commits12:05
straycatif that were useful we'd be doing it already12:06
ssam2fair enough12:07
* SotK submitted his partial builds patch series by doing `git push gerrit HEAD:refs/for/master/baserock/adamcoldrick/partial-builds` whilst I had my baserock/adamcoldrick/partial-builds branch checked out and I'm pretty sure that came out in the right order with the dependencies right12:08
ssam230 seconds of gerrit downtime coming up while I test backup automation again. sorry12:27
* pedroalvarez is sitll worried about why things are not in g.b.o after they have been merged in gerrit12:28
jjardonyeah, me too. ssam2 (or other) maybe you can take a look to the logs and see if there is something strange?12:29
ssam2good point, will do12:29
ssam2probably a merge conflict between gerrit's master and's master12:29
*** wschaller has quit IRC12:53
*** wschaller has joined #baserock13:10
SotKso, its impossible to deploy a minimal system at the moment13:27
richard_mawSotK: what's broken?13:27
SotKit doesn't use systemd, but it has simple-network.configure in its configure extensions13:28
SotKwhich expects systemd to be used13:28
SotKthings like "ERROR: [Errno 2] No such file or directory: '/srv/distbuild/tmp/deployments/tmpIy3eIZ/tmpmfch7p/overlay-deploy-minimal-system-x86_64-generic-rootfs/etc/systemd/network/50-e*.network'" happen13:29
richard_mawSotK: is it failing because there's no /etc/systemd/network directory, or because /etc/systemd/network/ doesn't exist?13:29
richard_mawjjardon: ^^13:29
richard_mawI don't care much about whether it drops networkd config in which won't be used, but we need to make it work even when the /etc/systemd/network directory doesn't exist13:30
*** petefoth has quit IRC13:31
persiaI thought we were careful to complain about all the systemd-networkd patches and make sure that got tested with minimal.13:41
persiaWould adding minimal to the mason runs help avoid this?  We really want to support the non-systemd case.13:41
ssam2adding the minimal systems to clusters/ci.morph might indeed make sense13:48
straycatis anyone here using gertty? i think i need to generate a http password for it, but when i try to do this on it gives a 'working...13:49
straycat' notifaction, but no password gets generated13:49
ssam2the 'generate password' thing in gerrit seems really buggy13:49
ssam2try going to a different page, and going back to the settings page again, keep clicking the button13:49
straycatit just worked13:49
ssam2it worked for me in the end13:49
pedroalvarezand it just worked for me too13:49
ssam2i've not investigated reporting a bug upstream yet, but there's definitely something wrong with it13:50
* bashrc_ notices that in 14.x versions the morph definitions were all over the place13:51
straycatyou mean chunk morphs in chunk repos?13:51
bashrc_no separate subdirectories13:52
*** petefoth has joined #baserock13:53
pedroalvarezit's much better now13:58
jjardonSotK: I can not look at that at the moment; can you file a bug in the storyboard an assign it to me, so I do not forget?14:10
SotKjjardon: sure14:10
jjardonSotK: thanks14:10
straycatgertty didn't seem too great so i'm trying gerrymander except i have an issue caused by the following:14:35
straycat richardipsum  salo  ~  $  ssh review gerrit stream-events14:35
straycatCapability streamEvents is required to access this resource14:35
* pedroalvarez is still trying to get gertty working :/14:39
straycatgiven it ran for around 30seconds with no output while presumably populating its database i don't hold much hope for it14:39
Kinnisonstraycat: admins are given streamevents automatically, everyone else has to be explicitly granted it14:40
straycat  huzzah!14:42
straycati guess someone will have to grant me streamevents if i want to use the 'watch' command14:43
ssam2I can look into streamevents if you want, presumably there's some reason gerrit doesn't allow it by default14:47
ssam2if I can't find a reason, i'll enable it for everyone14:47
* SotK notes it will probably need to be enabled for Zuul when we get to that14:48
*** wschaller_ has joined #baserock14:57
*** wschaller has quit IRC15:00
*** gfinney has quit IRC15:03
*** gfinney has joined #baserock15:03
*** gfinney has quit IRC15:05
*** gfinney has joined #baserock15:05
*** gfinney_ has joined #baserock15:06
*** jonathanmaw_ has joined #baserock15:36
*** jonathanmaw has quit IRC15:36 seems unresponsible over SSH. not sure why yet15:44
* straycat manages to comment from cli with ` richardipsum  salo  ~  $  ssh review gerrit review --code-review 0 -m "'This is a test comment from a cli'" 6fe11cd320332f9c24ca17ba2f9c1bdb323cc213'15:45
* Kinnison is confused as to what those characters are in straycat's prompt15:46
straycatmad unicode courtesy of powerline shell15:46
KinnisonWhatever you're pasting my shell insists isn't valid utf815:47
straycatoh? i thought it was15:47
Kinnisonperhaps screen is squiffing it15:48
kejiahuis down?15:54
ssam2there seems to be an issue at the hosting provider for some instances. and are both unresponsive for me.16:03
ssam2I've raised a support ticket16:03
*** jonathanmaw_ has quit IRC16:03
ssam2I have an existing SSH session into the 'frontend' machine, and while it's responsive, I can't run any commands16:03
ssam2so seems like IO is broken16:03
kejiahuit was quite unstable since this morning. and now it's completely unaccessible16:03
*** jonathanmaw has joined #baserock16:03
*** brlogger` has quit IRC16:03
*** brlogger has joined #baserock16:03
Kinnison16:03 <+statusbot> Status update: We're investigating an issue that's affecting access to instances hosted on our compute and storage platforms.  Further updates to16:05
Kinnison                   follow. -- http://status.datacentred.io16:05
Kinnisonjust appeared on the hosting company's channel16:05
DavePageKinnison: I was just about to paste that :)16:05
pedroalvarezin the meantime I managed to ssh in git.baserock.org16:08
pedroalvarezand looks like cgit is working again16:08
KinnisonSo minor hiccough16:09
pedroalvarezI hope :016:09
* straycat doesn't see a way to do inline comments from a cli yet16:09
straycatSotK, send it to the ml?16:10
straycator point out that the previous is on the ml if you submit to gerrit16:10
SotKI think I'll send it to the ml16:11
pedroalvarezi know is not ideal for some users, but it's possible to do comments in the web UI of gerrit16:11
SotKI worry that submitting it to Gerrit will cause at least one of (a) about 20 commits being sent to Gerrit, (b) merge conflicts16:11
straycatpedroalvarez, yup but no way to do that from the cli16:13
straycatSotK, which would be 20 new changes, but at least on the same topic? >.>16:13
ssam2SotK: is there really no way to break it up into separate topics16:13
ssam2Gerrit is used by several big projects, there must be ways of doing this sensibly16:14
SotKssam2: my branch depends on my OSTree branch which is on the mailing list16:14
ssam2ah, so it depends on something outside of Gerrit16:14
SotKif I push just the two commits for partial deployment they won't make any sense16:14
ssam2probably fair to send it to the list as well then16:15
SotKhopefully its the last one I send to the list! :)16:15
jjardonssam2: any idea what happened with the "missing" commits? should I send the patches again?16:16
ssam2just going to investigate now that the infra is back up16:16
KinnisonApparently Gerrit 2.11 and later support submitting comments over ssh16:16
jjardonssam2: thanks, sorry for rushing :)16:17
ssam2Gerrit's REST API is also really good, wouldn't be too hard to write a wrapper tool if none of the existing ones suit you16:19
jjardonKinnison: nice, unfortunately there is not a stable 2.11 release yet (but a promising v2.11-rc1 tag is already in the "stable-2.11" branch)16:19
* Kinnison gets the impression 2.11 has been a long-time coming16:20
KinnisonGiven it was last july that issue 602 was marked "fixed in 2.11"16:21
straycatKinnison, you can already submit comments over ssh, but not inline comments16:24
Kinnisonstraycat: yes, 2.11 is supposed to extend that suppotr16:26
ssam2jjardon: i don't see any missing changes in master of morph or definitions16:26
ssam2 is merged to master of morph.git in and gerrit.baserock.org16:26
straycatKinnison, okay i see16:27
jjardonssam2: what about ?16:29
jjardon(sorry for the previous one)16:30
pedroalvarezI failed to see that it was for morph.git :/16:32
ssam2jjardon: hmm, that does seem to be missing16:33
tlsa  <-- can anyone help me with pushing to gerrit?16:40
tlsaMy changes have change-ids16:40
straycatwhy are we not using https on
KinnisonLack of SSL certificate right now I imagine16:43
straycatright, so that's why i can't use gertty i think16:43
* straycat sighs16:44
jjardontlsa: mmm, it should work. try yo remove tlsa from for/master/tlsa/certify-reproducibility ?16:45
ssam2I've requested Codethink get a certificate for us, it's being looked in to16:45
ssam2tlsa: could you paste the output of 'git log master..HEAD' somewhere?16:46
tlsajjardon: my branch is tlsa/certify-reproducibility16:49
ssam2tlsa: are you trying to push all of those 4 at once?16:49
tlsassam2: yes16:49
ssam2I think you need to push each new change to refs/for/master/tlsa/xx individually16:49
ssam2although if that's the problem, the error message given is a complete joke16:49
tlsaI thought more than one worked from what SotK said earlier16:50
jjardonDoesnt matter whats the name of tour branch, you are setting the "topic" of the group of commits there16:52
ssam2tlsa: I think that you can push a branch which contains existing unmerged changes, but not a branch that contains more than one new change16:52
tlsajjardon: oh, ok16:52
ssam2jjardon: do you know when 'submit' was pressed for ?16:53
tlsajjardon: same rejection without the "tlsa/"16:53
ssam2jjardon: today about 10:28:39 ?16:53
ssam2if so, no errors in the log (except for 'failed to send email')16:54
pedroalvarezhe pointed it out at 11:2316:55
ssam2could be but that suggests there'd be an error in the log file16:55
ssam2and it purports to be fixed in 2.816:56
jjardonssam2: yeah, 10:28 sound rigth16:57
jjardontlsa: Its definitely possible to send a branch with several commits: gerrit will order them correctly with the same "topic"16:58
* jjardon taking a look to the logs again16:58
ssam2there seem to be intermittent errors in the Gerrit logs where it can't access the database17:02
rjekrelated to the other data centre problems?17:06
*** jonathanmaw has quit IRC17:07
*** a1exhughe5 has quit IRC17:10
ssam2jjardon: I've manually fixed and made it actually get merged17:10
jjardonssam2: cheers17:10
ssam2had to click 'cherry-pick', then +2 it myself17:10
ssam2I've no idea why this happened, so please look out for it happening again17:11
richard_mawssam2: could it be a race between the fetch+merge and the push involved? If there was a change from another source to morph.git and the merger isn't capable of handling conflicts then it might just get dropped17:12
ssam2richard_maw: the merger should be capable of handling and reporting conflicts, but it's possible that there was a race17:13
jjardonssam2: I wonder if we can change the policy to rebase before merge? I feel that we will have a lot of usefulness merge commits if not (before makes sense because we put the review info there)17:13
ssam2changing policy might help indeed17:14
straycatssam2, we could have a self-signed for now?17:14
ssam2straycat: that would be extra work and would only result in people complaining about the self-signed cert. may as well do it properly17:14
radiofreeooh, lovely coloured dmesg in baserock master17:15
* straycat shrugs17:15
straycatalso a little surprised there's no comment on the RFC for branching morph, given that's a pretty big change17:15
ssam2at present rate I should have some time free to do code review in about 201617:15
straycatfair enough17:16
* straycat goes home17:16
*** gfinney_ has quit IRC17:28
*** mdizzle has quit IRC17:37
* pedroalvarez is sad because his comments in the ML regarding mox3 were ignored17:41
pedroalvarezthe patch was also sent to gerrit17:41
tlsaooh, I've done something that has changed the reason why gerrit rejects my patchs17:43
tlsa[remote rejected] HEAD -> refs/for/master/certify-reproducibility (n/a (unpacker error))17:43
tlsawas (no new changes)17:43
Kinnisonthat's the git packer change17:44
Kinnisonthere's a flag to pass17:44
Kinnisonrdale: had the same issue17:44
rdalei needed a '--no-thin' option when doing a push17:49
tlsayep, just found that on stackoverdlow17:50
jjardonmaybe of the interest of the infrastructure team:
tlsawell, that finally tel me push the series, but I'm not sure how the original (no new changes) rejection went away :s17:51
ssam2jjardon: feel free to start writing a custom database backend for Gerrit :)17:52
* jjardon runs17:52
ssam2tlsa: it does seem that Gerrit's error messages are pretty unhelpful. Perhaps you can report a bug17:52
*** gary_perkins has quit IRC17:58
*** bashrc_ has quit IRC18:01
*** edcragg has quit IRC18:09
*** Guest5251 has quit IRC18:10
*** mdunford has quit IRC18:14
*** ssam2 has quit IRC18:20
*** tiagogomes_ has quit IRC18:21
*** Krin has quit IRC18:25
jjardonmmm, getting the cache from feels very slow, anyone more with this problem?18:39
pedroalvarezjjardon: save some logs to look at how long takes to fetch some artifacts tomorrow18:43
jjardonalso mason is red, same error as other times:18:43
pedroalvarezAlso, this can be a problem of your network18:43
jjardon2015-03-23 16:03:49 Progress: Transferring pulseaudio-doc to shared artifact cache18:43
jjardon2015-03-23 16:04:02 Build of pulseaudio-doc failed.18:43
jjardonmaybe, will go home and try there18:44
pedroalvarezMason error might be related with fetching artifacts18:44
pedroalvarezCache server misbehaving..18:44
pedroalvarez17G free, no space problems this time18:45
*** tlsa has quit IRC19:04
*** tlsa has joined #baserock19:04
*** wschaller_ has quit IRC19:06
radiofreepedroalvarez: could you not give it more HD space?22:18
radiofreeit seems like you're constantly fighting with lack of space22:18
pedroalvarezit wasn't space this time, but yes, I've been thinking about that23:51
pedroalvarezmason failed this time for the same reason that had problems today: The hosting company had a minor hiccough.23:52
pedroalvarezalso, the new version of tar is causing problems when building lsof (which uncompresses a file in the configure commands):

Generated by 2.15.3 by Marius Gedminas - find it at!