IRC logs for #baserock for Thursday, 2015-02-12

*** zoli__ has quit IRC00:35
*** zoli__ has joined #baserock00:47
*** zoli__ has quit IRC00:58
*** zoli__ has joined #baserock01:09
*** zoli__ has quit IRC01:28
*** zoli__ has joined #baserock01:51
*** zoli___ has joined #baserock02:00
*** zoli__ has quit IRC02:00
*** JPohlman1 has joined #baserock02:47
*** richard_1aw has joined #baserock02:47
*** JPohlmann has quit IRC02:52
*** richard_maw has quit IRC02:52
*** zoli___ has quit IRC02:56
*** zoli__ has joined #baserock02:56
*** zoli__ has quit IRC03:07
*** zoli___ has joined #baserock03:07
*** zoli___ has quit IRC04:37
*** zoli__ has joined #baserock05:33
*** zoli__ has quit IRC05:34
*** zoli__ has joined #baserock05:39
*** mdunford has joined #baserock07:08
*** sherm_ has joined #baserock07:21
*** gfinney_ has joined #baserock07:49
jjardonKernel live patching merged!
pedroalvarezNote: I rebooted yesterday. It had like 5 cvsps processes consuming 90% of cpu each.08:17
*** jonathanmaw has joined #baserock08:51
*** franred has joined #baserock08:55
*** bashrc has joined #baserock09:01
persiaI was just reading the mailing list for another project, and discovered that they had configured their gerrit to work in "cherry-pick mode": I'm not quite sure what Gerrit config does this, but apparently it reduces the number of merge commits.09:29
persiaThe cost being that each change has to be safely atomic to land.09:29
persiaI've mixed feelings about this: I like the lack of merge commits, but I also see the point of having multiple commits that are related.09:30
persia(and I don't know how such a model works with the Mason and Zuul stuff)09:30
*** tiagogomes_ has joined #baserock09:32
jjardonI actually like the fact there is always a merge commit when landing series of patches: easier to keep track of reviewers and easier to revert09:33
persiaI don't trust humans to make merge commits, and I don't know that robots would add as much metainformation.09:35
* persia has seen all sorts of annoyances in human-managed merge commits, including spurious changes to "fix" issues in the previous series, making tools like `git blame` less helpful, etc.09:35
* rjek doesn't trust humans or computeres to do anything, and thus generally accepts that some patching up will have to be made from time to time.09:35
persiaYes, but not *every* merge (which is roughly what we do today)09:36
persiaThat's just putting up with a level of messiness that nobody needs.09:36
*** edcragg has joined #baserock09:47
*** edcragg has quit IRC09:52
*** straycat is now known as ofcourseistilllo09:52
*** edcragg has joined #baserock09:53
ofcourseistilllothis nick gets truncated by freenode, propose we move to oftc09:54
* rjek sets his nick to the text of War & Peace.09:55
*** ssam2 has joined #baserock09:58
*** ChanServ sets mode: +v ssam209:58
paulsherwoodpedroalvarez: s/veryfiying/verifying/ ?09:59
pedroalvarezpaulsherwood: argh! where?10:01
*** mariaderidder has joined #baserock10:02
paulsherwoodin the video10:02
pedroalvarezpaulsherwood: double argh!10:02
paulsherwoodthere's also veyfing10:02
* robtaylor is veryfing10:02
paulsherwoodbut nonetheless, very impressive :)10:02
robtaylorwhich video is this ooi?10:04
pedroalvarezLooks like I'm not doing well my workarounds to not talk in the videos :)10:04
*** gary_perkins has joined #baserock10:07
paulsherwoodpedroalvarez: do you want to fix the speelings (:-), before folks start publishing the link?10:08
pedroalvarezpaulsherwood: I'll do that as soon as I can10:10
radiofreeyou need sound pedroalvarez!10:11
*** Krin has joined #baserock10:19
pedroalvarezfranred: I've +1ed the six patch and the coreutils dd patch (just in case you don't se my emails, something is going wrong with my mail client/server)10:26
franredpedroalvarez, thanks :)10:31
CTtpollardfor the cluster morph on should it have disk_size and cloud_init?10:32
CTtpollardI thought cloud_init made it the maximum size possible?10:32
ssam2morph still needs to know a DISK_SIZE10:32
CTtpollardahhh, does this need to be changed then
ssam2it works by creating a disk image locally, then uploading it to openstack. And it's not smart enough to work out the needed size of the disk image for you10:33
CTtpollardfor the openststack bit?10:33
ssam2yes, I think so10:33
ssam2although it does have a note about DISK_SIZE already10:33
CTtpollard'DISK_SIZE - the size of the disk image to create. If you deploy an image with cloud-init (as in the following example), the disk will be resized to fit the maximum space given in OpenStack.'10:33
CTtpollardI've added it to the wiki, cheers ssam210:36
pedroalvarezI remember adding that note, but I also remember adding DISK_SIZE to the cluster example10:36
pedroalvarezCTtpollard: ta!10:36
robtaylorrdale_: seen ? may be useful for cribbing libstdc++ configuration11:05
rdale_yes, i've been looking at that11:06
*** zoli__ has quit IRC11:06
rdale_there is a script called 'stage2-glibc-fix-specs' for glibc - i think i need an equivalent to that to take a gcc spec file, and then fix up the location of ld.so11:06
rdale_but i can't find that script in defintions yet11:07
tiagogomes_rdale_ that script it is on glibc repo:
rdale_ah right, thanks11:11
tiagogomes_rdale_ did you check where musl installs
rdale_lib/ -> /tools/lib/libc.so11:14
rdale_i need to generate a .spec file in the same dir as libgcc.a is in, and then i fix up the in there to point to the musl one with a script like the glibc one11:14
*** a1exhughe5 has joined #baserock11:19
jmalkreading trove docs; - makes use of many placeholder variables (YOUR-TROVE-HOST, trove_id, new-code, ab-cd (= trove_id), TROVE-HOST, TROVE_ID) to refer to a smaller number of things. is this the kind of thing I should tidy up, or is there a documentation authority I should consult before wading in?11:22
*** inara has quit IRC11:27
*** inara has joined #baserock11:33
rdale_woo hoo! the stage2-musl-fix-specs that i copied from the glibc one has fix my configure error11:39
robtaylorrdale_: niiiice11:40
*** zoli__ has joined #baserock11:43
pedroalvarezthat file has a "NASTY HACK" mark, Is possible to that in other way?11:44
*** zoli__ has quit IRC11:50
ssam2jmalk: in general, feel free to tidy it up11:55
tiagogomes_I tried to do in another when I was updating gcc. The spec is needed only for stage1-gcc. So I tried to create the spec file in the stage1-gcc chunk instead. The idea was that by no including stage1-gcc in the build-depends of some chunk would cause the spec file to be not installed11:55
ssam2jmalk: if petefoth is around its good to check with him first. but he's not around til Monday11:55
tiagogomes_However, I found out that stage1-gcc chunk is installed even if it is *not* on the build-depends of some chunk11:55
jmalkssam2: thanks11:58
jmalkssam2: and I'll get in touch with him when he's back online11:59
*** zoli__ has joined #baserock12:22
*** zoli__ has quit IRC12:24
*** wschaller has joined #baserock12:37
*** zoli__ has joined #baserock12:41
*** richard_1aw is now known as richard_maw12:49
*** zoli__ has quit IRC13:06
*** zoli__ has joined #baserock13:09
*** ssam2 has quit IRC13:36
*** zoli___ has joined #baserock13:37
*** zoli__ has quit IRC13:37
*** ofcourseistilllo is now known as straycat13:42
*** a1exhughe5 has quit IRC13:52
*** ssam2 has joined #baserock13:54
*** ChanServ sets mode: +v ssam213:54
*** ssam2 has quit IRC14:00
*** sherm_ has quit IRC14:03
jmalkI see as a very minimal "Index of /" html page, is that meant to be the case? I thought it used to be meatier14:12
SotKjmalk: its been like that as long as I remember I think14:13
jmalkSotK: ok just checking nothing was broken14:13
*** sherm_ has joined #baserock14:16
*** ssam2 has joined #baserock14:17
*** ChanServ sets mode: +v ssam214:17
tiagogomes_anyone knows why we built libapr and libapr-util from tarballs and not git?14:23
tiagogomes_s/git/svn/ in this case14:23
tiagogomes_franred mmm, yep, that's my name :)14:26
franredI've sent a patch to build apr (apr-util is inside of apr in master) for apache httpd server14:26
franredtiagogomes_, ^^14:26
tiagogomes_franred ah, what's the name of the patch series14:28
franredAdd apache-httpd server stratum14:28
franredtiagogomes_, ^^14:28
tiagogomes_franred I see, why didn't you use delta:apr ?14:30
*** ssam2 has quit IRC14:30
franredtiagogomes_, because I didn't see that we have apr in the repo....which is called libapr ¬.¬14:32
franredboth repos should be the same14:32
franredI think, I will check it14:32
tiagogomes_the looked the same at first glance14:32
*** ssam2 has joined #baserock14:43
*** ChanServ sets mode: +v ssam214:43
jjardonred in !14:54
jmalkI'm still confused by upgrade to devel instructions - ; there's no hint as to where to "get the branch containing the devel system" from, or what to "use the cluster morphology" for. Does anyone have any input on how to achieve these things, and therefore how to improve the doc please?14:57
jjardonMaybe this is related with the problem? "2015-02-12 13:37:14 Progress: Ready to build perl-misc: waiting for a worker to become available"14:57
*** a1exhughe5 has joined #baserock15:15
Zarait's easier to use the simple build-deploy workflow, if possible15:22
Zarabaserock source is here:
Zaraso you should be able to find the url of the relevant devel system there, in 'systems'15:24
jmalkZara: thanks very much. that's what I'm currently doing to make progress, but want to not have dead-end documentation if possible15:25
ZaraI suspect it doesn't say what to do next because I couldn't get morph deploy working without at the time, so I left that bit for someone who knew what they were doing. :P15:31
pedroalvarezjjardon: I've looked into mason and looks like it's building again15:31
jmalkZara: alas, that person is not me. thanks for your advice15:32
Zarajmalk: heh, yeah, it's unofficially of the form 'do the easy thing, here's a pointer for how to do the harder thing if you know this stuff better than I do'.15:33
jmalkZara: makes sense15:33
*** ssam2 has quit IRC15:43
*** ssam2 has joined #baserock15:43
*** ChanServ sets mode: +v ssam215:43
*** ssam2 has quit IRC15:56
*** fay__ has quit IRC15:56
*** fay_ has joined #baserock15:58
*** jonathanmaw has quit IRC16:00
pdarHiya, what does morph do if you dont give it configure/build/install commands?16:12
SotKruns whatever the default configure/build/install commands are for the build-system you put in the chunk morphology16:13
SotKif you didn't make a chunk morphology then it autodetects the build system based on files in the chunk repo16:14
SotKdefault configure/build/install commands are here:
pdarthanks SotK. I had a feeling it wasnt just magic...16:16
jmalkhaving used to upgrade to a devel system, it is running as TEST - what do I need to do to make it my 'actual' system? just a case of `system-version-manager set-default TEST`?16:22
pedroalvarezjmalk: after running cycle the default should be already TEST16:23
pedroalvarez`system-version-manager get-default` or `system-version-manager list` in newer versions16:24
jmalkpedroalvarez: you're right, sorry. so I can just start using it as I was using the non-devel system (factory)?/16:24
pedroalvarezjmalk: a reboot might be needed :)16:25
jmalkyep. thanks!16:25
SotKsystem-version-manager get-running will tell you which version you're booted into16:25
jmalkSotK: yes thanks.16:27
tlsaI've installed a fresh VM, following
tlsaIt said to make a 2nd disc of 30Gb.  I made one of 40Gb.16:32
tlsaOn my first build attempt (using the cycle script), I get:16:32
tlsaERROR: Insufficient space on disk: /tmp/morph_tmp requires 4000000000 bytes free, has 112625664016:32
jmalktlsa: I think that refers to available ram16:32
tlsaam I doing something silly?  Or do I need to make the disc a a lot bigger than the instructions say?16:32
tlsaoh :s16:33
bashrctlsa: you could try: btrfs filesystem resize max /16:34
tlsano change16:34
tlsabut I think it should be the /src disc that its building on?16:35
pedroalvareztlsa: you have to create a morph configuration file (/etc/morph.conf)16:37
tlsapedroalvarez: ah, yeah, thanks16:38
tlsapedroalvarez: yep, all fixed16:40
tlsascripts/ systems/xfce-system.morph clusters/upgr16:42
tlsaERROR: Unknown architecture armv8l64 in morphology systems/build-system-armv8l64.morph16:42
jmalktlsa: how did you go from the wiki link you posted to using the cycle script?16:43
jmalktlsa: (seems like missing out the morph.conf will mess people up, so it needs to be a step between the two on the wiki if poss)16:44
Zarait is on the wiki, between the two steps16:44
Zarabut the first step goes back to the quickstart page at present, so it's possible to skip past the morph conf step-- or to think it's not relevant16:45
tlsaI was jumping around, because I didn't know what I was doing -- I didn't know if I was using an ancient release so wanted to update before doing anything else16:45
jmalkZara: yes, you're right, tlsa thanks16:46
Zara(so might be worthwhile updating the link back to quickstart so it goes to the 'login to the vm' section of the quickstart page)16:46
Zara(I'm still not happy about the amount of back and forth between pages, but at present there are a lot of instructions to follow :S)16:47
jmalkZara: yes, it's a complex process16:47
jmalkZara: by which I mean both, "working with the system" and "making the documentation navigable" are complex processes16:48
tlsaI get the same "Unknown architecture armv8l64" error tryting to build the systems/weston-system-x86_64-generic.morph system16:51
pedroalvareztlsa: to what exactly have you updated to?16:52
pdarfranred: Is it necessary to add the whole nfs strata to deal with the runtime dependency of memcahced on libevent?16:52
tlsapedroalvarez: I've not yet updated anything16:52
tiagogomes_is there an easy way to build a chunk that needs access to the source of another chunk16:52
*** a1exhughe5 has quit IRC16:52
franredpdar, unless we move libevent to databases or memcached to nfs... yes16:53
pedroalvareztlsa: ah, ok, the support for that architecture has been added recently to morph16:53
tlsaright, so latest defintions needs new morph16:53
pedroalvareztlsa: so either upgrading your system, or using latest morph will do16:54
*** gfinney_ has quit IRC16:54
*** gfinney_ has joined #baserock16:54
straycattiagogomes_, you could modify the chunk to add the chunk it needs as a submodule i guess?16:54
tlsapedroalvarez: how can I upgrade my system, if the morph I have doesn't work?16:54
pedroalvareztlsa: you can still use latest morph without doing the whole upgrade:
tlsayeah, I'll have to do that16:55
pedroalvareztlsa: if interested on upgrade it, visit
pedroalvarezZara: is this explained better anywhere else? ^16:56
tiagogomes_straycat mmm...16:56
tlsaI was following the build-deploy-cycle instructions when I got that error16:56
tiagogomes_I'll think about it16:56
pdarfranred: how come?16:57
tiagogomes_tlsa, I'm curious, what are you doing with armv8l6416:57
tlsanothing at all16:57
tlsaI was building an x86_64 system on an x86_46 system16:57
franredpdar, to add libevent to your system you need to add the nfs stratum unless you create a libevent stratum - which I don't think it has any sense16:58
Zarajmalk: editing individual instructions for clarity is quite straightforward (so go ahead and sort things out where they're unclear!); putting things in the right places is a bit trickier.16:58
pedroalvareztlsa: hm.. morph validates everything before building it seems to me16:59
tlsapedroalvarez: yep, looks that way16:59
franredpdar, we try to not duplicate packages in different strata to not have to actualize all of them when you need to do it...16:59
franredpdar, so yeah, you need to add nfs stratum to the system if you want memcached use the libevent package16:59
pedroalvarezerm.. if it is a runtime dependency, it has to be added [too] to the system17:00
pedroalvarezotherwise it won't be installed on the produced system17:01
tlsapedroalvarez: might be worth updating the br image downloads on the website -- to make sure its morph is compatible with definitions17:02
*** sherm_ has quit IRC17:02
Zarapedroalvarez: not that I can find atm (I would say that it's normally fine to just use the commands listed in workflow and skim the overview).17:02
Zarathere is a note in the import tool quickstart about it17:03
*** zoli___ has quit IRC17:03
Zaraeg: Now build and deploy the system, and try it out!17:03
Zaraeg: change directory back to definitions/, and run:17:03
Zarascripts/ systems/web-system-x86_64-generic.morph clusters/upgrade-devel.morph17:03
pedroalvarezZara: thanks :)17:03
Zarathis assumes that the you've done the previous steps and have a system ready to deploy, so it really just gives you the syntax for using cycle.sh17:03
Zarathere's also a bit about building a system on the import tool quickstart page, haha17:04
*** CTtpollard has quit IRC17:04
pedroalvareztlsa: we kind of assume that Morph inside of a release of baserock can build the definitons of that release, but can't garantee that version of Morph will build always master of defintiions.git. That's why we suggest to use latest Morph in the quick-start page.17:06
pdarfranred: Hmm, I understand its good to avoid duplication of package decleration across strata, but it also seems unnecessary to include a bunch of packages you dont need in your system too.17:07
franredpdar, well that is the cost of the stratums ;-)17:08
straycatyes they are kind of bad in this regard17:08
straycatbut you can get around it by splitting the strata up17:08
pdarfranred: craaazy baserock17:08
pdarI guess for some strata the overhead of carrying extra packages is negligable, but when its not, is that where you have to think about reorganising strata?17:12
*** gfinney_ has quit IRC17:14
jjardonwelcome systemd  v218 :) thanks pedroalvarez !17:15
jjardonpedroalvarez: btw, sorry for being nippicking but my surname is Jardón, not Jardon17:16
pedroalvarezmine  is Álvarez17:16
franredpdar, yes, reorganizing strata is what happen some times...more than you think :)17:17
pedroalvarezjjardon: sorry about that, from some keyboards and inside of baserock is difficult to put 'ó'17:18
franredpedroalvarez, I though capital letters does not have written accent :P - but could be an deprecated rule17:18
mwilliams_ctperhaps a Spanish investigation is needed to find out franred?17:18
franredmwilliams_ct, pedroalvarez is right ;-)
jjardonpedroalvarez: np! its a known bug: ;)!/story/417:19
mwilliams_ctfranred: huh who'd have though pedroalvarez would know his own name? :)17:20
*** sambishop has quit IRC17:28
pedroalvarezmason is working hard these days, eh?17:30
tlsawhat's the URL for mason these days?17:31
pdarfranred: I guess this disadvantage of strata is outweighed by a benefit of strata, right?17:34
tlsaoh, wow, I thought that mason had been replaced17:34
franredpdar, you are right ;-)17:35
jjardonpedroalvarez: lets keep it busy ;)17:35
pdarwhy arent runtime dependencies stated in the chunks defined in stratum morphologies?17:46
pdarnamely why dont we keep track of the runtime dependencies like we keep track of build dependencies?17:52
*** Krin has quit IRC17:53
pedroalvarezpdar: richard_maw has done some work to suppor that, but there was some discussion and it hasn't never been merged17:55
*** franred has quit IRC17:55
jjardonpdar: AFAIK, currently, runtime dependencies are defined when you build the system morph file17:56
*** bashrc has quit IRC17:56
tiagogomes_we don't support runtime dependencies yet, so every build dependency will be a runtime dependency as well18:01
*** sambishop has joined #baserock18:03
*** mariaderidder has quit IRC18:17
pdarpedroalvarez: thanks, I have now located the email thread and get reading18:21
*** wschaller has quit IRC18:24
*** edcragg has quit IRC18:31
jjardontiagogomes_: you can define the runtime dependencies when you contruct your system morph file18:32
jjardonwell, kind of18:32
jjardonin that file you dont define dependencies but what should be in the system18:32
*** tiagogomes_ has quit IRC18:33
*** tiagogomes_ has joined #baserock18:34
pdarjjardon: looking at some system morph files i get the impression you dont directly state that there are runtime dependencies that might not allow your system to work, but practically you account for them by including the necessary strata18:36
*** edcragg has joined #baserock18:37
*** tiagogomes_ has quit IRC18:40
jjardonYeah, that why I said "kind of": you define what strata should be in the system, they do not have to have build dependencies between them18:41
straycatruntime dependencies aren't defined anywhere18:42
jjardonIts at least how i understand it works18:42
straycatmissed the "kind of", sorry18:47
jjardonstraycat: :)18:50
*** mdunford has quit IRC18:50
*** gary_perkins has quit IRC19:05
*** rdale has joined #baserock20:22
*** rdale_ has quit IRC20:25
*** edcragg has quit IRC20:29
*** rdale has quit IRC20:31
*** rdale has joined #baserock20:32
*** rdale has quit IRC20:34
*** rdale has joined #baserock20:35
*** rdale_ has joined #baserock20:37
*** rdale has quit IRC20:37
*** rdale has joined #baserock20:47
*** rdale_ has quit IRC20:48
*** rdale has quit IRC21:35
*** zoli__ has joined #baserock22:14
persiajmalk: If something in the docs is obviously wrong, just fix it.  petefoth is the most active in the documentation area, so probably has good guidance on plans and strategies, so check for big stuff, but don't wait until next week to fix a spelling mistake :)23:33
*** zoli__ has quit IRC23:53
*** zoli__ has joined #baserock23:59

Generated by 2.14.0 by Marius Gedminas - find it at!