IRC logs for #baserock for Tuesday, 2014-11-25

*** juergbi [] has joined #baserock00:07
*** rdale_ [] has joined #baserock03:30
*** rdale [] has quit [Read error: Connection reset by peer]03:30
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock06:11
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]06:11
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock07:12
* petefoth loikes that Taiga has built in access to, and it apprears you *don’t* need to be logged in or registered to use it07:28
*** mike [] has joined #baserock07:55
mike is now known as Guest8200007:56
*** wdutch [] has joined #baserock08:14 doesn't need login, that's correct08:55
*** bashrc [] has joined #baserock08:59
petefothinteresting discussions in #storyboard about configurable task (and maybe story) statuses and priorities09:00
petefoth for a lightly snipped transcript fro anyone not in that channel09:01
*** bashrc [] has quit [Client Quit]09:03
*** Guest82000 [] has quit [Read error: Connection reset by peer]09:03
*** bashrc [] has joined #baserock09:03
*** mike [] has joined #baserock09:04
mike is now known as Guest5669109:04
*** mariaderidder [] has joined #baserock09:06
*** Guest56691 [] has quit [Quit: Leaving]09:14
*** sherm_ [] has joined #baserock09:15
radiofreepaulsher1ood: memory09:22
radiofreei have actually come across that *once* with mesa, however it's pretty rare09:22
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]09:22
*** zoli_ [] has joined #baserock09:22
*** zoli_ [] has quit [Changing host]09:22
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock09:22
radiofreesimply rebuilding should fix it, or setting max-jobs to a number lower than the number of cores you have in your vm09:23
radiofreewe should add swap to br09:24
paulsher1oodradiofree: thanks09:27
*** tiagogomes [~tiagogome@] has joined #baserock09:30
radiofreepetefoth: a lot of that discussion seems to be "well that's how it works in launchpad"09:36
petefothradiofree: yes, but it is also covereing where *they* want to go. Thje references to LP are there because they all worked with it and wanted to make something that didn’t have LP’s limitations. paulsher1ood is working on^h^h with them too 09:38
paulsher1oodpetefoth: i'm not workign on them, just trying to contribute and understand their thinking :)09:39
petefothpah - words!09:40
radiofreepetefoth: ah right, i get it, it was just on the first read through it seemed like "it's like that in lp" was a defence for a feature being implemented in a certain way, rather than "it's like that in lp, we should change that here!"09:46
petefothradiofree: more like “LP did it that way. Was that OK or do we need to do it differently”09:47
petefothWhich seems to me like a sound approach.09:48
*** Krin [] has joined #baserock09:57
rdale_is systemd no longer always included?10:03
radiofreerdale_: i don't think a minimal system has it10:06
rdale_my enlightenment strata failed to build because systemd wasn't there - i've added a build-depends of systemd - does that include a runtime dependency too?10:07
rdale_i mean i've added 'foundation.morph' as a dependency to get systemd10:07
SotKrdale_: foundation.morph needs to be in your system definition too for it to be included at runtime10:08
*** locallycompact [] has joined #baserock10:24
richard_maw^ is why I did the run-depends patch10:26
* richard_maw attempts to stop harping on10:26
petefothrichard_maw: don’t do that! :) I like harps10:26
*** sambishop [] has quit [Ping timeout: 240 seconds]10:27
pdarHiya, I've been having trouble building the systems/ceph-system... after branching defenitions10:42
pdarmaster on friday. Ive determined that this is because some libraries ceph requires are no10:42
pdarlonger being built, despite their morphology being unchanged. I guess that10:42
pdarto baserock has10:42
pdareffective way that I can find what change has caused this problem?10:42
paulsher1oodpdar: git bisect?10:43
paulsher1ood(if you've never done a git bisect, probably worth reading about it anyway, it's awesome :-) )10:44
pdarpaulsher1ood: ooh thanks, I havent heard of it before. Ill take a look :)10:45
richard_mawI don't recall any such change, are you sure you aren't building a new version of ceph that has more dependencies now?10:47
paulsher1oodthat sounds more likely, actually10:51
paulsher1oodpdar: can you paste the error scroll somewhere?10:51
pdarrichard_maw: I have been doing that in a different branch and that works ok. This problem arrose when working with a branch of master I made to look at jims changes to a ceph.configure.10:55
pdarpaulsher1ood: I can, would that be the error as displayed in /src/morph.log?10:59
paulsher1oodpdar: yes. or jsut what spewed on your screen after '[a/b] [ceph] Building chunk ceph'11:01
*** sambishop [] has joined #baserock11:01
pdarpaulsher1ood: Okey dokey, I'll just recreate the error and pop it in a pastebin.11:06
paulsher1oodtvm. is a thing :)11:06
pdarpaulsher1ood: Ok, heres the error log:
pdarthanks for taking a look :)11:26
richard_mawboost thread library not found11:26
pdaryes, its not being built anymore for some reason, 11:27
pdarit is built in my older version of defenitions ok, but not with the one I branced on friday11:28
richard_mawhm, the version I'm looking at definitely includes boost in the stratum11:28
richard_mawit's possible that boost have split up the library into another repo, but I'd like you to look at
richard_mawI suspect the "Cache-related issues" section may be relevant11:29
* SotK thought we use a tarball import for boost11:29
* SotK was wrong11:29
richard_mawpdar: have you ever un-cleanly shut-down the system you're building in?11:29
radiofreeSotK: we do!11:30
pdarSotK: Yes, the version ceph uses is the boost-tarball 11:30
radiofreethere is a boost.git, but i think that fails because there's about a million submodules it attempts to pull in11:31
SotKheh, I found boost.git and assumed we were using that now11:31
pdarrichard_maw: It is possible, is there a way I could check?11:32
radiofreei suppose all of those would need to be lorried for boost from git to work :\11:33
richard_mawpdar: not really, though if you check and it starts working, then the answer was yes11:33
paulsher1oodpdar: can you paste the morph file that contains your definition of ceph, if it's not confidential?11:34
paulsher1oodgit grep ceph should show you where it is11:34
pdarpaulsher1ood: Its just the strata/ceph-service/ceph.morph in definitions:
*** fay_ [] has quit [Read error: Connection reset by peer]11:42
richard_mawwow, seems to guess which language is in use at random11:42
pdarrichard_maw: I tried building with --no-ccache and got the same error11:43
*** fay [] has joined #baserock11:43
richard_mawit's probably either the hard-link cache or a corrupted chunk artifact then11:43
fay is now known as Guest491911:43
petefothwelcome Guest4919 :)11:44
Guest4919 is now known as fay_11:44
pdarThe morphology defining the boost build instructions, strata/ceph-service/boost.morph, is set to supress errors I believe, and so the boost libraries dont build fine, but also dont complain about it. I think..11:46
pdarrichard_maw: so I should try clearing my hard-link cache?11:47
pdarah, I cant find the chunks subdirectory, has it changed name?11:49
richard_mawrun `morph --dump-config | grep tempdir`, as is in the instructions to get morph to tell you where it is11:51
pdarrichard_maw: Sorry I read the wrong command from the page11:52
pdargah, same error with no hard-linked cache.11:59
richard_mawrun `find . -name '*.chunk.*' -size 012:00
richard_mawin the artifacts subdir12:00
richard_mawof the directory listed when you run `morph --dump-config | grep cachedir`12:00
richard_mawhello straycat12:00
straycathello richard_maw :)12:00
petefothsome write extensions (e.g. kvm.write) have good documentation in Doc Strings. Would it make sense to move that information into the, converting it to YAML as I go?12:01
straycatoh, i was highlighted but for no reason :s12:02
richard_mawpetefoth: it would be better, as that way the documentation is discoverable without having to look at the code12:03
petefothrichard_maw: thanks12:04
pdarrichard_maw: Thanks, but it didnt work either. I also checked for short chunks too but didnt find anything.12:07
straycatrichard_maw, Are my yarns invalid?12:07
richard_mawstraycat: no, but I think they could be improved12:08
richard_mawthe goal is to have something that reads like english12:08
paulsher1oodpdar: did you see boost build before you did this? eg
paulsher1oodpdar: specifically, did you see [Build 111/135] [boost] Extracting upstream:boost-tarball12:13
pdarpaulsher1ood: not recently. It just starts buildinf from [Build 112/135] [ceph] ...12:15
paulsher1oodpdar: have you changed strata/ceph-service.morph at all?12:19
paulsher1oodoh, actually - i get the same error12:19
paulsher1oodhere's another system needs adding to mason12:21
*** elevenarms [~elevenarm@] has joined #baserock12:22
pdarno, I'm fairly sure I know why the error is happening. Boost appears to build fine but here we see that the error messages from building the luibraries are suppressed:
pdarI guess in the past when this hack was introduced, some libraries failed to build but they would not effect ceph??12:24
paulsher1oodpossibly, but... naughty mdoffman :)12:24
pdarRemoving the hack causes the build to fail at boost12:26
*** elevenarms [~elevenarm@] has quit [Quit: Be back later ...]12:30
paulsher1oodpdar: hmmm. almost certainly worth asking mdoffman for help here12:39
pdarpaulsher1ood: Okey dokey, thanks for all your help! :-)12:43
* paulsher1ood feels like he's been no help, really12:50
* paulsher1ood wonders if anyone has successfully built a ceph system on baserock recently.... jmacs for example?12:50
rdale_  maybe it would be better to build boost with cmake rather than jam12:52
paulsher1oodrdale_: good idea12:56
rdale_shall i have a look at the error we're getting with jam though?12:57
paulsher1oodrdale_: if you don't mind12:59
radiofreehave you tried upgrading boost?13:07
jjardonHi, why morph needs an internet connection when deploying?13:07
richard_mawit needs to work out which artifact you meant with the specified definition file, which invovles looking at git. You should be able to turn this off by passing --no-git-update13:08
richard_mawwe tried forcing --no-git-update, but that turned out to be a bad idea because if you built your system with distbuild, then you don't have the caches locally to look up what the artifact you want to build is13:09
richard_mawI am aware that this sucks13:09
radiofreemaster of boost-tarball contains a newer version13:12
radiofreethough browsing that in cgit just crashed my laptop13:12
*** elevenarms [~elevenarm@] has joined #baserock13:12
petefothhow / where / by which other thing does the .check extension assoicated with a .write extension get called?13:13
richard_mawfor the type field in the cluster morphology, it looks for an extension called $kind.check13:13
richard_mawif it exists, it is executed with the location field as argv[1]13:14
richard_mawthey can either be located in definitions.git or morphlib/exts/13:14
richard_mawit is only called at deploy time, it is called before the rootfs tarball is extracted or configuration extensions are run13:15
petefothrichard_maw: thanks. When you say ‘*it* looks’ do you mean ‘the write extension looks’ or ‘some other thng looks’? I couldn’t see a reference to the .check extensio in either kvm.write or in Extensions.py13:16
richard_mawpetefoth: I meant `morph deploy` by *it*13:16
* petefoth goes to look at the code for `morph deploy` (and thanks richard_maw again)13:17
*** cosm [] has quit [Ping timeout: 245 seconds]13:23
*** elevenarms [~elevenarm@] has quit [Quit: Be back later ...]13:42
pdarpaulsher1ood: jmacs didnt seem to have any problems when I spoke to him yesterday, hes away till thursday so cannee check though.13:55
pdarrdale_: thanks for taking a look, Have you found anything?13:55
pdarradiofree: I've not tried upgrading boost no, I'll give it a go13:56
rdale_it fails building the 'wave' library for me13:56
rdale_so i'm trying out giving the script a list of libs to build without wave13:57
*** cosm [] has joined #baserock14:02
jjardonthanks for the info richard_maw . Is this tracked somewhere so It doesnt get lost?14:05
petefothrichard_maw: the latest version of talls about ‘The INITRAMFS_PATH option’ which doesn’t seem to occur in either kvm.write of kvm.check. IS that mainly because the file has got out of step with the implementation< or am I missing something?14:07
* petefoth remembers that irc is asyncronous, especially when richard_maw has gone to lunch14:10
richard_mawpetefoth: no, it's because the bits which handle INITRAMFS_PATH are part of a python module that kvm.write depends on14:22
petefothrichard_maw: OK. Which module - tempfile? urlparse?14:23
richard_mawsorry, I should've been more specific, it uses morphlib.writeexts14:24
petefothrichard_maw: np -just spotted that one myself14:24
richard_mawwhich is defined in morphlib/writeexts.py14:24
petefothis it the case that *all* the parameters must be supplied either by the user in the call to `morph deploy` or as defaults in the cluster morphology/definition? Or do I need to check the code for each to see if they have hard-coded defaults?14:43
petefoths/parameters/parameters for write extensions/14:44
richard_mawthere's no hard-coded defaults, but location becomes $1, type is removed and MORPH_LOG_FD is added14:45
* petefoth will take some time to digest that14:46
doffmpdar: Ok, Boost hack. Did you have a specific question? When I added boost they were right in the middle of the process of splitting it up in to many separate repositories.14:47
doffmThats probably finished now. We should think about separating out the morphologies when we get the chance.14:47
paulsher1ooddoffm: the key issue is that ceph doesn't build with the hack14:48
pdardoffm: ok, yes.14:48
radiofreedoffm: we have a trove mirror of the git repo14:48
doffmpaulsher1ood: Thats a pretty big issue. :)14:48
radiofreehowever there's quite a lot ....
radiofreecurrent HEAD of boost-tarball master worked for me on ARM at least14:49
radiofreewhich is 1.5614:49
doffmpdar: OK, yeah thats pretty hacky. I'm guessing that ceph have just added dependencies to more boost libraries, ones that were previously failing to build.14:53
doffmAre still failing to build.14:53
pdardoffm: The hack wasnt working with current master and boost 1.54. Several more boost libraries fail to build than I've seen with previous versions of definitions14:55
rdale_atomic, log, thread and wave are the ones that don't build14:57
doffmpdar: The boost project organization has changed alot in the last year. The github repo that we are mirroring (boostorg/boost.git) is now just a shell that pulls in the project submodules.14:58
doffmI'd ignore the silly hack I put in. Too much has changed, it was very hacky to begin with. Just need to work out why those submodules arn't building.14:59
doffmWhich I'm afraid I'm not sure of. I can try building 'thread' here if you would like.15:00
radiofreedoffm: we're *not* using the git mirror15:00
radiofreewe're using a tarball release, which i upgraded to 1.56 a few weeks ago, try that pdar (without the || true hack)15:00
doffmradiofree: pdar: Get rid of the || true hack. Its just stupid error supression to get things working.15:01
doffmIt still won't work. But it will tell you whats broken.15:02
pdardoffm: okey dokey, thanks!15:02
pdarradiofree: I shall try boost 1.56 tarbal and see if that solves my problems, thanks!15:03
radiofreewhen was the last time this was built pdar?15:04
pdarradiofree: boost 1.54 and ceph working in harmony??15:05
radiofreeerm.. sure15:05
radiofreemore specifically boost 1.54, have you built that recently?15:05
pdarI only started having problems building when I started using master of defenitions on friday, before I was working on master from November 4th15:07
radiofreei'm going to go with glib being updated then15:08
radiofreerichard_maw: if i make the changes to the svn lorry you suggeted is it a +1 from you? or should i resubmit?15:10
richard_mawif that's the one with "branches/*/c5" or whatever, then yes15:11
richard_mawbut a quick paste of the lorry in question would eliminate uncertainty on my part15:11
pdarradiofree: ahh ok, thanks15:12
petefothrichard_maw: so `location` is mandatory for all write extensions?15:14
richard_mawthough you can put an empty string there if the write extension takes all its arguments by the environment variable parameters15:14
petefothrichard_maw: nod. 15:16
petefothis ‘arguments’ the preferred word, ratehr than’parmaterers’ or ‘settings’? (I don’t mind which we use - I just want us to be consistent throughout our documents)15:17
petefothor even ‘parameters’ :)15:18
richard_mawI guess strictly, arguments refers to the $1 and $2 passed to the write extensions, while parameters is a superset containing the environment variables too15:18
petefothOK - I’ll go with ‘pararararameters’15:19
radiofreerichard_maw: i'm looking at
radiofreetags/*/c5 looks like it's fine for them all15:21
radiofreeapart from CRYPTOPP_5_6_115:21
radiofreeactually for 5_6_0 as well15:21
radiofreethere's no c5 folder in there15:21
straycatrichard_maw, How come you didn't vote on that patch?15:21
richard_mawstraycat: probably forgetfullness, which patch is this?15:22
straycatthe one that makes the arg to build/deploy cmds relative to the working dir15:22
richard_mawI'll send an e-mail15:24
*** mariaderidder [] has quit [Ping timeout: 240 seconds]15:24
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]15:28
richard_mawstraycat: conditional +1 now, I'll do another pass through the code to see if I want to upgrade that to an unconditional +115:29
straycatoh, cool :]15:30
*** mariaderidder [] has joined #baserock15:39
petefothOK - I’m being dim again - why does the KVM write extension have/need a HOST part in its location custom URL *AND* a HOSTNAME parameter? 15:44
* petefoth has been looking at code for too long15:44
richard_mawthe HOST part of the location is the VM host, HOSTNAME is for the system you're deploying15:45
petefothrichard_maw: that sounds palusible. Ta!15:47
petefothso it’s the Hostname of the Guest system :)15:48
radiofreehowever, for i don't think the tags will work?15:50
richard_mawunfortunately, we can either pick some tags being empty of content, or some being in c5 subdirectories15:52
richard_mawwe'd need changes to lorry to be able to specify otherwise15:52
richard_mawon balance, I'd probably prefer the tags worked and had odd layouts to non-functional tags, so change "tags" to "tags/*" and you have a +2 to merge15:53
petefothmore dimness: so the NAME part of the custom URL, is the name of the VM from the perspective of KVM and HOSTNAME is how the running VM appears to the rest of the network? 15:55
petefothstraycat: thanks15:57
* petefoth is getting there, slowly15:57
*** mariaderidder [] has quit [Ping timeout: 265 seconds]15:59
straycatoh right the yarns from v1 weren't quite right16:01
straycati didn't notice16:01
* straycat sighs16:01
petefothso does look sane? (not asking for a full review - just a check I’m on the right lines)16:02
richard_maws/syslinux/the bootloader/ and you have a +1 from me16:04
petefothrichard_maw: I said I didn’t want a full rerview :)16:04
richard_mawsame amout of work on my part16:05
*** petefoth [] has quit [Quit: petefoth]16:10
*** mariaderidder [] has joined #baserock16:13
radiofreeah i can't merge to lorry16:22
richard_mawradiofree: do you want to be able to?16:29
richard_mawwell, you can now16:30
radiofreerichard_maw: thanks16:38
radiofreewas to merge the patches16:38
* perryl feels it may be worth saying that the robocop reference in the firehose rfc was more tongue in cheek than a proper suggestion :)16:43
*** tiagogomes [~tiagogome@] has quit [Quit: Leaving]16:56
robtaylorlets call it hugbot 16:57
robtaylorargh. i mean
perrylrobtaylor, +100 :)16:58
perryli'm mainly hoping to avoid discussion of the name of the tool taking over completely17:00
robtaylorbikeshedding is always easier17:01
paulsher1oodactually, i've been meaning to say, sometimes the 'bikeshedding' card is unfairly played. sometimes little details like naming have a big impact17:05
robtaylornot very often ime17:06
straycat is now known as cat034-117:07
* robtaylor imagines a bikeshedding session over the name 'Linux'17:08
cat034-1 is now known as straycat17:08
jjardonHi, now that we have the latest xserver in baserock, I was planning to strip down it and its deps so it has the minimum functionality to run inside a wayland session (through xwayland). Is this ok or do you think its worth to have a complete x11 stratum?17:08
jjardonrobtaylor: GNU/Linux!! ;)17:08
robtaylorjjardon: indeed, there's a bikeshed that'll never end ;)17:09
paulsher1ood+1 (for both jjardon's suggestions) :)17:09
*** tiagogomes [~tiagogome@] has joined #baserock17:13
*** petefoth [] has joined #baserock17:17
*** Krin [] has quit [Remote host closed the connection]17:22
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock17:49
*** tiagogomes [~tiagogome@] has quit [Quit: Leaving]17:56
*** bashrc [] has quit [Quit: Lost terminal]17:58
*** mariaderidder [] has quit [Quit: Ex-Chat]17:58
*** locallycompact [] has quit [Ping timeout: 264 seconds]17:59
*** sherm_ [] has quit [Quit: Leaving]18:01
* richard_maw wrote to experiment on migrating processes to another tree.18:15
richard_maw`--replace --filter TARGET=/ --mount-source=/dev/sda --mount-type=btrfs -o subvol=/systems/TEST/run` will make it create a mount tree somewhere in /tmp that is the same as your current one, but with /systems/TEST/run mounted at / instead of what you currently have18:17
*** cosm [] has quit [Ping timeout: 255 seconds]18:23
*** rdale_ [] has quit [Read error: Connection reset by peer]18:50
*** cosm [] has joined #baserock19:06
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]20:34
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock20:51
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]21:04
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock21:12
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]21:13
*** zoli_ [] has joined #baserock21:13
*** zoli_ [] has quit [Changing host]21:13
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock21:13
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]21:26
*** genii [~quassel@ubuntu/member/genii] has joined #baserock21:27
*** zoli_ [] has joined #baserock21:49
*** zoli_ [] has quit [Changing host]21:49
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock21:49
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]21:59
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock22:07
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]22:41

Generated by 2.15.3 by Marius Gedminas - find it at!