IRC logs for #baserock for Wednesday, 2016-09-21

*** gtristan has quit IRC04:20
*** gtristan has joined #baserock04:37
*** fay has joined #baserock06:48
*** fay is now known as Guest2149906:49
*** rdale has joined #baserock07:04
*** ctbruce has joined #baserock07:21
*** locallycompact has joined #baserock07:37
locallycompactIt's just this isn't it
*** toscalix has joined #baserock07:54
*** CTtpollard has quit IRC07:55
locallycompactI'd like to get rid of kind in some capactiy because system and stratum have exactly the same fields now07:57
*** Guest21499 is now known as faybrocklebank07:57
locallycompactNow where we had chunks: List<Chunk> and strata: List<Stratum> we have instead contents: List<Either Stratum Chunk>07:59
*** CTtpollard has joined #baserock08:02
locallycompactUnless people would deeply deeply prefer contents: Either List<String> List<Chunk>08:12
locallycompact*Either List<Stratum> List<Chunk>08:14
*** gtristan has quit IRC08:22
*** locallycompact has quit IRC08:48
*** CTtpollard has quit IRC09:10
*** gtristan has joined #baserock09:11
*** CTtpollard has joined #baserock09:12
leemingStratum {union} Chunk, seems like a sensible thing to me. Although a little uneducated/experienced in the matter09:23
*** gtristan has quit IRC09:29
*** locallycompact has joined #baserock09:32
*** gtristan has joined #baserock09:42
*** jjardon_matrix has quit IRC09:52
* paulsherwood prefers List<Either Stratum Chunk>10:10
* locallycompact too10:11
paulsherwoodif by that you mean a list of things, here each thing can either be a single component, or a collection of single components10:11
locallycompactrather than homogenous lists10:11
paulsherwoodrather than one list of collections and one list of single components10:12
* paulsherwood is spelling it out to make sure we fully understand each other10:12
*** dabukalam has quit IRC10:19
*** cphang has joined #baserock10:22
*** jjardon_matrix has joined #baserock10:23
cphangNewbie baserock problem here: Built a dev system, but it's failing when I try to upgrade it. It's failing when I try to remove the remote mount point. Root has passwordless ssh access to
cphangERROR: ssh command `ssh root@ -- env BASEROCK_SYSTEM_CONFIG_SYNC=/tmp/tmp.Gy4zd5/systems/devel21Sept/orig/usr/bin/baserock-system-config-sync /tmp/tmp.Gy4zd5/systems/devel21Sept/orig/usr/bin/system-version-manager deploy /systems/devel21Sept/orig` failed ERROR: extensions/ssh-rsync.write failed with code 1:10:23
pedroalvarezthat error is not useful at all10:26
pedroalvarezis there anything else in the log?10:26
*** franred has joined #baserock10:27
cphangNope, there's an additional function traceback, but nothing else.10:30
pedroalvarezcphang: you could try to disable the cleanup in extensions/ssh-rsync.write and try to run the failed command manually10:33
pedroalvarezhoping that doing that will give you more logs10:33
cphangpedroalvarez: Okay I'll give that a go thanks!10:35
pedroalvarezlet us know if you need more help/pointers10:44
pedroalvarezwith more logs I'll be more useful10:44
pedroalvarezcphang: oh btw, you may have some info in /var/log/baserock-system-config-sync.log10:46
cphangpedroalvarez: Okay I had a look I could maybe do with some pointers. Couldn't find anything in config-sync.log apart from whether files were sync'd or not. Is there a flag for me to write in ssh-rsync.write, or to put in when I use morph upgrade, or in the cluster morphology file? Thanks for your help! :)11:07
pedroalvarezhum.. not sure if --debug will help here, I dobut it11:11
pedroalvarezwhat  instructions are you following? I'll try to reproduce11:13
cphangJust the baserock quickstart guide. Initially used the script to upgrade, when I got the error message I tried to use the morph upgrade on it's own as I could see the build was successful.11:16
locallycompactpaulsherwood, necessary that contents be specified in order? could we just allow this?
pedroalvarezcphang: hmm. it worked for me
pedroalvarezwe need to figure out why it's failing in your case, how do you feel about going with my suggestion and stop the cleanup after failure and try to run the failing command manually?11:24
paulsherwoodlocallycompact: sadly, order is significant for current *system* definitions, because we have overlaps in systems (ie multiple chunks contain their own versions of a given file... so system ends up with the last one installed)11:26
paulsherwoodorder is not significant for chunks, so long as the transitive dependencies are carried through11:27
paulsherwoodlocallycompact: i would strongly prefer that order is not significant...11:27
cphangSure. I've noticed a discrepency between your and my upgrade log. I think it's cleaning up, and then trying to remove the mount point?11:27
paulsherwoodbut that leads back to yesterday's discussion, where we warn on all of the overlaps but choosing not to build the system is useless for the user11:28
pedroalvarezcphang: exactly11:28
pedroalvarezcphang: is there anything useful after that?11:28
pedroalvarez(after the error message, i mean)11:28
cphangpedroalvarez: Nope same as before, so Upon failure I'll try and run the command? I'm just wondering if cleanup of the tmp/ folders before removal of the mount point could be the problem? It means I can't run the command afterwards manually too. Apologies if I'm missing something!11:33
*** jonathanmaw has joined #baserock11:34
pedroalvarezcphang: the cleanup is a consequence of a failure, otherwise there is nothing to clean up11:36
pedroalvarezwhat I mean is that, removing tmp/folders is not in this case the cause of the failure11:37
*** dabukalam has joined #baserock11:37
pedroalvarezcphang: so, if you go to extensions/ssh-rsync.write, there you will find all the upgrades code11:38
pedroalvarezif you comment out the line 83 and 84 you will disable the cleanup11:39
pedroalvarezand if you comment out the line 72, you will disable the removal of mount point11:40
pedroalvarezand 62 to disable the removal of the folder of the mount point11:41
cphangpedroalvarez: Ahh okay sorry, I didn't realise. I thought from the log the failure point was at removing the mount point, rather than it was coming to the end of a clean-up procedure. I'll do as you say and get back to you! Thanks again for your help11:41
*** dabukalam has quit IRC11:46
*** dabukalam_ has joined #baserock12:07
*** dabukalam_ has quit IRC12:14
*** dabukalam_ has joined #baserock12:18
*** dabukalam_ has quit IRC12:26
locallycompactpaulsherwood, I changed it so it writes out all build-depends explicitly and then sorts, now it succeeds at parsing and resolving b-ds but doesn't seem to be staging build-essential into core12:50
*** dabukalam has joined #baserock13:06
*** dabukalam has quit IRC13:12
*** gtristan has quit IRC13:12
*** dabukalam has joined #baserock13:22
*** lachlanmackenzie has quit IRC13:22
*** dabukalam has quit IRC13:23
*** lachlanmackenzie has joined #baserock13:40
*** CTtpollard has quit IRC14:04
*** CTtpollard has joined #baserock14:06
cphangpedroalvarez: Apologies for the delay, had some other problems to sort out. So it looks like it's failing at the system-version-manager because it's not there in the first place. Coincidentally, to get to this stage by commenting out the cleanup, if I try it once more it fails at an earlier stage, even if I restore the original ssh-rsync file14:32
pedroalvarezcphang: does the folder '/tmp/tmp.GPd9bl/systems/2016-09-21/orig/usr/bin' exist?14:34
pedroalvarezin that log it looks like it's failing at the same point14:35
cphangNo, the tmp.GPd9bl folder is completely empty14:35
richard_mawit's used as a mount point, rather than putting things in it14:36
pedroalvarezso maybe you haven't disabled the code that does the umounting14:37
cphangRight. I thought I did, but I'll try and reproduce it again.14:39
*** locallycompact has quit IRC14:45
cphangpedroalvarez: okay so keeping the folder mounted, systems contains a default, devel and factory folder, but devel is empty.14:54
cphangSo calling the system-version-manager also fails14:55
pedroalvarezaha, "devel" is the VERSION_LABEL, yes14:56
pedroalvareztrue, there are more cleanups  to disable14:58
pedroalvarezlines 103 and 104 in this case14:59
cphangAhh okay.15:00
pedroalvarezsorry, missed that15:01
cphangNot a problem, again, thanks for all the help!15:01
*** locallycompact has joined #baserock15:04
cphangpedroalvarez: Seems to be another case of broken paths, I disabled the clean up in L129 as well just in case but it gave the same message.15:13
pedroalvarez uh? /etc/grub.d/15_ostree ?15:22
pedroalvarezsorry, didn't know that file existed15:22
richard_mawwell it's crashing because it doesn't15:23
pedroalvarezi don't really understand what's going on there15:24
*** ctbruce has quit IRC15:26
* richard_maw no longer knows, but it's failing in merge-config, which tries to do a three-way diff for the contents of /etc for the two snapshots15:26
richard_mawperhaps the logic there doesn't handle the file not existing in one version of it properly15:27
cphangWell I'm following the quick-start instructions to the letter, so is it something funky with the kvm setup perhaps?15:28
pedroalvarezmaybe is something funky with our code15:28
pedroalvarezcphang: I would really like to see your  /var/log/baserock-system-config-sync.log15:29
pedroalvarezis that possible?15:29
*** gtristan has joined #baserock15:30
pedroalvarezhm.. something is removing the files.. ?15:39
* pedroalvarez checks one thing15:42
locallycompacthere's an example of something that should be possible with the new spec
pedroalvarezcphang: do you have " /etc/grub.d/15_ostree" in your current system?15:48
paulsherwoodlocallycompact: have you taken into account the build-mode stuff?15:51
paulsherwood(maybe that's what's causing the missing build-essential)15:52
locallycompactI'm just using ybd as is with a small hack to let include-mode through15:52
cphangyes I do15:52
cphangpedroalvarez: yes I do15:52
locallycompactit has all these contents fields internally so I just tried it, gets surprisingly far15:53
locallycompactbut something wierd with staging15:53
paulsherwoodlocallycompact: if you set log-verbose: True it may give you more clues15:54
pedroalvarezeugh, it's a symlink..15:54
locallycompacthmm there's no ybd conf in upstream15:55
paulsherwoodput it in your definitions dir?15:56
paulsherwoodwhat is 'upstream' in your context15:56
paulsherwoodoh, yes15:57
*** dabukalam has joined #baserock15:59
*** jonathanmaw has quit IRC16:02
*** faybrocklebank has quit IRC16:07
pedroalvarezok I've got a theory16:19
pedroalvarezfacts: factory has the symlink, and new version has the symlink too, but the contents of the file have changed16:20
pedroalvarezno no, sorry, that's not it16:21
cphangpedroalvarez: Just to let you know as well, there are various builds now on system-version-manager, but I think again due to broken links they're not removable. I had to give them different names, otherwise the upgrade would fail when mounting16:30
pedroalvarezyeah, makes sense16:31
pedroalvarezI haven't figured out what is going on16:31
cphangShould I try setting up with virt-manager or virtualbox instead? I don't understand the difference in behaviour between your and mine's use case16:35
pedroalvarezcphang: what are you using at the moment?16:36
pedroalvarezplease send me a link to the instructions you are following and I'll try to reproduce from scratch16:36
cphangKVM, QEMU16:37
cphang setup with KVM, using fallocate for the second drive16:37
locallycompact16-09-21 00:02:21 [12/309/309] [gdbm] Installing dependencies16:38
locallycompact []16:38
paulsherwoodlocallycompact: presuming you can repeat locally, you could try import pdb ; pdb.set_trace() to look around at the point you're expecting build-depends to be populated16:50
locallycompactneed to clone everything but yeah16:51
paulsherwoodlocallycompact: can avoid that by setting mode: no-build16:51
paulsherwoodthat will fudge all the cache-keys, still run the assembly process16:51
paulsherwood(but as empty artifacts)16:52
locallycompactthat seemed to run through16:53
paulsherwoodyes, but with empty build-depends, so put in the debug to catch at the right point16:53
*** tiagogomes has quit IRC17:46
*** toscalix has quit IRC17:59
*** lachlanmackenzie has quit IRC18:24
*** gtristan has quit IRC18:27
*** lachlanmackenzie has joined #baserock18:31
*** rdale has quit IRC18:41
*** locallycompact has quit IRC19:03
*** mwilliams_ct has quit IRC21:07
*** vgrade has quit IRC21:34
*** vgrade has joined #baserock21:34

Generated by 2.15.3 by Marius Gedminas - find it at!