IRC logs for #baserock for Monday, 2014-12-08

*** zoli_ [] has joined #baserock06:37
*** zoli_ [] has quit [Changing host]06:37
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock06:37
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]06:40
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock07:04
*** petefoth [] has joined #baserock07:27
*** fay_ [] has quit [Remote host closed the connection]07:54
*** fay [] has joined #baserock07:56
fay is now known as Guest7248707:56
Guest72487 is now known as fay_08:23
*** rdale [] has joined #baserock08:58
*** bashrc [] has joined #baserock08:59
*** bashrc [] has quit [Quit: leaving]09:06
*** mike [] has joined #baserock09:06
*** bashrc [] has joined #baserock09:06
*** mariaderidder [] has joined #baserock09:06
mike is now known as Guest8445609:06
*** tiagogomes [] has joined #baserock09:16
*** tiagogomes [] has quit [Ping timeout: 265 seconds]09:22
*** tiagogomes [] has joined #baserock09:22
*** franred [] has joined #baserock09:23
*** franred [] has quit [Remote host closed the connection]09:28
*** franred [] has joined #baserock09:44
*** ssam2 [] has joined #baserock09:54
Mode #baserock +v ssam2 by ChanServ09:54
*** mdunford [] has quit [Read error: Connection reset by peer]09:54
*** mdunford [] has joined #baserock09:54
* petefoth added a placeholder for the 'Rationalise upgrade functionality' story in
*** locallycompact [] has joined #baserock10:04
*** locallycompact [] has quit [Ping timeout: 252 seconds]10:52
*** locallycompact [] has joined #baserock11:04
bashrclooks like doesn't work. Certificate verification failed11:22
rjekRemove the s.11:22
bashrcI did11:22
rjekWithout the s, there is no certificate verification to occur :)11:23
bashrcif it's inperable then it should probably be delisted11:23
bashrcas listed here
pedroalvarezit's operable, but IIRC the certificate is self signed11:24
pedroalvarezthe certificate of the trove11:24
persiabashrc: You may need to disable certificate checking in git, or add the certificate to the certificate list11:25
bashrcok, however a git clone on it fails11:25
* persia idly wonders if there should be a chunk with the cert that is included in the development system11:25
bashrcpersia: I could, but really should users have to do that?11:25
persiabashrc: Which users?11:26
persiaI think users of devel systems would have easier lives if we included a cert for baserock.org11:27
persiaI think users of the baserock tooling are better served by using upstream ca-certificates from mozilla.11:27
persiaI'm not excited enough to try to cause a CA to be accepted by Mozilla.11:27
bashrcyes, it's not really tennable in 2014 to be asking people to be using insecure repos11:27
persiaIt isn't insecure.11:27
persiaThere just isn't a chain of trust.11:28
persiaPersonally speaking, I'd prefer to see more self-signed and peer-signed certificates, because I don't trust the practices of some of the organisations approved by Mozilla (and in at least two cases, know how to arrange them to issue certificates with information that does not match a legal entity)11:28
persiaSo from that perspective, maybe it is a bug in git that it just fails, rather than giving a useful opportunity to accept a certificate?11:29
bashrcI don't trust those practices either, but there isn't a better system in place at present11:29
persiaYep, but as a result, I don't know the best path for everyone :(11:30
bashrcit's not a bug in git, failing on inability to verify a certificate is pretty standard in a lot of software11:31
bashrcif baserock were to be used in any security critical environments only having http repos or downloads without checksums would be a problem11:46
Zara_hi, I'm trying to update baserock to the newest version; I tried following the instructions in the 'upgrading a baserock installation' section, but the command I'm hoping to use (baserock-import) still isn't found, suggesting it hasn't updated. Do I need to run the command from somewhere specific?11:53
ssam2do you mean the 'baserock-import' command? if so, no, it should be in /usr/bin11:55
ssam2so you should be able to run it anywhere11:55
Zara_sorry, that wasn't clear, I meant the morph upgrade command11:55
persiabashrc: Yes.  The checksums are a known issue that we've talked about solving.  For certificates, I think we need a plan.11:56
bashrcthe easiest solution is just to adopt the conventional system until such time you can figure out some better arrangement11:57
ssam2Zara_: ah, right. Yes, you have to run it inside the checkout of the branch you want to upgrade to11:58
Zara_ssam2: ah, ok, I originally tried it in the import branch but I'm guessing I'd actually want the newest definitions branch?11:59
ssam2for testing the import tool, the best branch to use right now is 'sam/update-import-tool'12:00
ssam2(which it looks like I named incorrectly.)12:00
ssam2if that branch had been reviewed and merged to master, then master would be the best branch to use12:00
* petefoth has a task to review the tutorial in his todo list
Zara_could someone tell me the exact morph checkout command to use in this case (or point me to something which explains how they work)? 12:09
Zara_petefoth: I'll note what needs adding as I go along. :)12:11
paulsher1oodZara_: better to use morhp branch? so 'morph branch baserock:baserock/definitions your-branch' should give you latest definitions in your-branch12:11
Zara_paulsher1ood: I was trying to use checkout from within a testbranch, is the the same thing?12:12
Zara_oh wait, his branch is *in* the definitions repo; I think that's what I was missing12:14
franredZara_ no is not the same, morph checkout will checkout the branch on the server,  morph branch will create a local branch with that name, AFAIK12:15
paulsher1oodmorph branch baserock:baserock/definitions your-branch his-branch12:15
paulsher1oodwill give you his-branch in your-branch... i hope that makes some sense12:15
Zara_I'll try it out and see :)12:18
bashrcis there a way of updating baserock systems without having to create and flash an entire new image?12:22
Zara_ah, seems to have worked; had to make a new testbranch since it acts weird if there's stuff already in the branch (or possibly if there's already a branch with that name)12:22
DavePagebashrc: I was under the impression that creating and flashing an entire new image was a feature, not a bug12:24
*** mdunford [] has quit [Ping timeout: 260 seconds]12:25
bashrcDavePage: seems like a lot of work if you just want to make some minor bug fix12:27
petefothbashrc: making a new system shouldn't take long (because cached build atrifacts), and you need to do it to check that your change hasn't broken anything12:28
pedroalvarezand you don't need to flash anything to upgrade your baserock system if it's using a btrfs filesystem12:30
bashrcI'm not familiar with btrfs. Does it have some image update feature?12:34
pedroalvarezthe baserock upgrade process needs btrfs12:38
paulsher1oods/upgrade/default upgrade/12:38
paulsher1oodother processes are possible :)12:38
pedroalvarezyeah :)12:39
Zara_I'm getting a 'deployment failed as system is not yet built' message. Suspect it's something to do with this instruction 'Make sure the 'morph' field below matches the system you are upgrading'; I'm not sure what that refers to.12:40
Zara_(what the 'morph field' refers to12:40
Zara_(my guess is the '(hostname)')12:41
pedroalvarezZara_: so you want to upgrade your system, don't you?12:41
pedroalvarezto do that, you have to build the system using `morph build` and then deploy it as an upgrade to your running system12:42
Zara_ah, ok, that should definitely go in the tutorial! :P12:42
ssam2Zara: it'd be really cool if you could fix the 'how to upgrade your VM' guide to be more helpful to people who are relatively new to Baserock12:44
ssam2after you've done the upgrade of course12:44
Zara_yeah, I'll get on it :)12:45
pedroalvarezZara_: regarding to the 'morph' field - It's the following line:12:47
pedroalvarez- morph: systems/devel-system-x86_64-generic.morph12:47
pedroalvarezso, if you want to upgrade your system to a system that is not a devel-system, you only have to change that field12:47
Zara_oh, so that refers to the thing *above* it on the page-- I thought it was referring to something in the upgrade command!12:50
Zara_so it's a field in the cluster morphology12:51
Zara_I'll make a note to move that up12:51
Zara_the build fails when it reaches setup.py12:52
SotKcan you paste the full error in a pastebin somewhere?12:53
pedroalvarezI assume that is the of the baserock-import chunk12:54
Zara_SotK: sure, do you have the url for the br pastebin handy?12:54
SotKZara_: ^12:55
Zara_SotK: thanks :)12:56
Zara_I do not know the significance12:57
rjekVHSIC Hardware Description Language.  It's a language for designing chips.12:58
pedroalvarezrjek: please, I know is not the best pastebin option, but it works fine for now12:58
rjekpedroalvarez: Can it be configured to default to plain text rather than randomly selecting a colour scheme that maximises unreadability?12:58
pedroalvarezrjek: probably12:59
SotKZara_: that looks like a problem with the import tool stratum12:59
Zara_SotK: yeah, seems like it, though I'm not sure if that means I tried to build the wrong thing12:59
SotKZara_: nope, the stratum is wrong I think :)13:00
SotKthe baserock-import chunk has no build-depends listed, so is probably being built before ansicolor has been built going off that error13:01
Zara_SotK: (heh, I meant I might have tried to build an unfinished version or something and there's a working one lurking around elsewhere)13:01
Zara_though I hope it's not me for a change :P13:01
SotKwhere are you building it from?13:03
persiabashrc: I'm not sure what you mean by "the conventional system".  We do provide ca-certificates in the devel image, which works for any https URL signed by the CAs listed there.13:03
Zara_SotK: I followed the instructions above, so it should be from my own branch containing the definitions from Sam's branch, sam/update-import-tool13:05
Zara_the system itself was the x86_64 generic devel thing13:05
SotKthe stratum does indeed look wrong in that to me13:09
SotKZara_: try adding ansicolor (and the other chunks in the baserock-import stratum) to the build-depends field of the baserock-import chunk in that stratum13:10
Zara_Ok, I'll give it a go :)13:11
* SotK wonders why the 0.10.0 tag for pbr produces version 0.11.0.dev48.g49ff27c, and the 0.9.0 tag produces version 0.10.013:15
Zara_SotK: cool, that worked :)13:16
SotKZara_: good :)13:16
SotKZara_: maybe submit a patch to fix it since its broken in master too by the look of things?13:17
*** mdunford [] has joined #baserock13:18
Zara_SotK: I think ssam2 said it hadn't been merged to master, yet, so hopefully he can catch it before that13:24
Zara_(yay, build ends successfully!)13:24
pedroalvarezyeah, is not in master, otherwise mason would be failing13:26
pedroalvarezand is not:
* SotK is confused13:32
SotKperhaps mason built them in the right order first try?13:35
Zara_hm, I did something wrong; I destroyed my baserock.13:39
Zara_well, my devel system; luckily I have a copy13:39
* SotK has never seen that before13:41
* persia has had that happen, when running out of space on a btrfs volume13:41
persiaIf one runs several sequential upgades without removing some results with system-version-manager, it is easy to fill the folume.13:42
Zara_persia: how would I find out if that's what happened?13:44
bashrcpersia: there was some mention of btrfs space issues in a meeting I was in this morning13:45
persiaFrom the state you're in now, I think you'd have to mount the backing store from another system, and then use btrfs-tools to dig through.13:45
persiaDo you  have any unpushed git repos in that environment?13:45
bashrccouldn't there be some delete buffer, such that the disk never gets totally filled?13:45
persiabashrc: Depends on what one tells the system to do.  Doesn't work if I try to deploy 6 2.5GB filesystems to a 10G backing store.13:46
persiaThat said, we should have better errors when we break things (and less catastrophic breakage), but when running everything as root, we don't have any reserved space protections.13:47
bashrceverything as root?13:48
persiaYes.  Currently, morph expects to be run as root.13:49
Zara_when I logged into my copy system, I got a msg saying 'FACTORY login:[    4.8566991] systemd-journald[90]: File /var/log/journal/[long key]/system.journal corrupted or uncleanly shut down, renaming and replacing. I'm not sure if that's to do with the update break, or if it's because I then had to force off the vm (it no longer recognised shut down or reboot)13:49
persiaWhat is your "copy system"?13:50
Zara_persia: when I was trying to update my devel system on friday, I managed to instead make a copy of it called 'newbr'13:50
Zara_possibly it's not actually a copy and just is a generic devel system build13:51
persiaHard to say.13:51
Zara_I haven't investigated that thoroughly yet; I'm just relieved I don't have to do everything from scratch13:51
persiaIf you have precious git repos there, I'd recommend trying to mount the disk image from another system.13:52
persiaIf you don't have precious data, I'd recommend starting over with a new devel VM.13:52
Zara_persia: luckily, I haven't used baserock much yet, so it should be ok to start over-- is there a way to delete the broken factory system?13:53
Zara_(without deleting the whole develvm and the other systems on it, since that'd save me a few mintues)13:53
persiaDeleting "factory" isn't a good idea.13:54
persiaAnd I only know how to delete other systems using `system-version-manager`, but that requires the ability to run commands.13:55
Zara_I can run commands from my copy (just checked, it seems to be a copy after all)13:56
persiaTry `system-version-manager --help`13:57
persiaIf you set the default to the running system, you can delete the othes (except "factory")13:57
jjardonHi, ok to lorry libusb?
Zara_persia: Ok, I've done that for now. :)14:01
paulsher1oodpersia: deleting factory is fine now14:01
paulsher1oodso long as you have a working alternative running/default system14:01
persiaChecking my mail log, it seems that morph a3c7fea2e1a7e10460d58d9fcbf8b395c0b17893 or newer is safe t use to upgrade systems without "factory"14:01
franredjjardon, looks ok to me +114:01
persiapaulsher1ood: Teaches me to check my logs: I just should ask, as someone else is always faster :)14:02
Zara_so, if I delete factory, will that let me clear the space without me needing to spend a long time learning to use lots of btrfs tools? (I'd like to reattempt the upgrade if possible)14:03
jjardonpaulsher1ood: franred thanks, pushed14:03
paulsher1oodZara_: yes, run system-version-manager remove factory14:04
persiaZara_: Yes, as long a ou have new enough morph14:04
paulsher1ood(which will only work after set-default to something else)14:04
Zara_ok, let's find out (since if not, I'll need a new vm anyway! =D )14:04
Zara_ah, I get an error: 'ERROR: error accessing '/tmp/tmpEk4n_e/systems/factory/orig'14:05
Zara_though it's gone from the list14:05
Zara_ah, never mind, the copy seems to be incomplete; it's still saying 'morph not found' (I must have somehow affected it today as well). looks like it's time to start from scratch.14:18
ssam2zara: good catch on the build failure of baserock-import14:20
Zara_at least something useful came out of this! :)14:21
ssam2I added manpage generation at the last minute which means that all the dependencies now need to be available at build-time14:21
pedroalvarezhm.. I wonder if systemd-timesyncd.service is the replacement of the ntp[d] service we had before15:10
pedroalvarezIt's not working in my systems15:10
pedroalvarezbecause: ConditionVirtualization=no was not met (currently investigating)15:10
pedroalvarezlooks like this unit is defined to not work in virtualized environments, so maybe I'm not looking to the right place15:12
KinnisonVirtualised machines usually have their clocks sync'd to their hosts15:13
pedroalvarezKinnison: that's the reason of why my vm is 5 minutes ahead, thanks!15:14
persiaYour host is 5 minutes ahead?15:14
Zara_hm, would it also be useful for me to add a wiki tutorial for installing baserock using the virtual machine manager gui? it's a bit fiddly and both times I've done it, it's taken me longer than it had to.15:25
paulsher1oodZara_: using which virtual environment?15:26
Zara_kvm, I think15:27
Zara_the virtual machine enviroment help says libvirt15:27
* Kinnison thinks Zara_ is using virt-manager15:28
Zara_yeah, it doesn't have to be kvm, but mine is (I think)15:28
Zara_Kinnison: yeah15:28
paulsher1oodso is not sufficient? (i only use virtualbox)15:28
Zara_it didn't work for me when I tried it some months ago, which is why I ended up using the gui, though that might have been due to a mistake on my part15:29
* SotK found that kvm guide unhelpful when he was first using Baserock, and someone (Kinnison perhaps) told me it was easier to just use the gui15:29
KinnisonIt's way easier to use virt-manager than to use kvm directly15:30
SotKAlso, is useless15:31
radiofreedoes virt manager make it easy to setup your network?15:31
rjekkvm is based on qemu by Fabrice Bellard, and thus has... a less than ideal command line interface.15:31
radiofreei'm still using /etc/qemu-ifup scripts...15:31
rjek(also see: ffmpeg)15:31
Kinnisonvirt-manager lets you control network setup yes15:31
Kinnisonit's not massively capable, but you can do the simple stuff like bridge vs. nat vs. isolated15:31
radiofreehmm i'll have to look into that, the way i'm used to is -net nic,model=virtio and -net tap15:32
radiofreethen a script (/etc/qemu-ifup) which gets run automatically to do the brctl stuff15:33
radiofreeit works well enough to allow you to quickly boot things from the command line, is the script15:34
franredSokt, looks like lockfile is now here -->
franredthanks by the way, I will fix it15:42
SotKfranred: it is15:43
SotKfranred: thanks 15:43
franredSotK, ^^15:45
franredand another nice soul for a quick review15:45
SotKhmm, actually, is it worth getting it from the repo on
franredSotK, why not?15:48
franredit is the place where was moved15:48
franredSotK, ^^15:49
perrylignore that...15:50
SotKfranred: the readme in links to
SotKthey are the same repo though, so +1 whichever you pick15:51
pedroalvarezthe HEAD commit of master of the loickfile repo is not in the opesntack repo15:55
paulsher1oodhow does that happen?15:55
franredbecause they change the repository and we haven't detected? SotK I've changed for this:
SotKpaulsher1ood: they moved the repo to (mirrored in a new repo on github) and added a commit to the old github removing everything and adding a note in the README explaining where it went15:56
SotKfranred: +115:57
SotKpaulsher1ood: (we were lorrying from the old github)15:57
pedroalvarezoh, I didn't realise that the commit was to remove everything15:57
pedroalvarezmakes sense now to me: +1 franred 15:58
franredpedroalvarez, SotK, thanks, it is pushed15:59
SotKthanks franred 16:00
franredSotK, the odd thing is that I am building that repository in openstack sucessfully16:02
franredIm building it from 4eb27075e0523b6ed2027444fd42d83a6e9b14c5 master, which is the last commit before they move clean the repository :)16:02
franreds|move clean|move/clenan|16:03
radiofreejjardon: not sure if you're aware but we can use mainline libdrm now16:21
radiofreetegra stuff has been merged16:21
SotKdoes anyone happen to know exactly what the issue was that means that the python import-tool doesn't work for things which use pbr?16:31
ssam2only Richard Ipsum16:31
ssam2my guess is that it works in a fundamentally different way to 'pip' so the tool would need extra code to understand whatever format it uses16:32
ssam2that code should be fairly easy to write, because pbr seems like it has quite a sane input format16:32
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]16:33
franrednot sure how difficult would be, but I know that pbr uses normally "requirements.txt" file to populate the install_requirements for setup_tools16:34
SotKI only ask because when I was trying to get Zuul installed in a hacked up way a few weeks ago I couldn't do anything due to an error similar to those linked in TODO.python16:34
franredSotK, I think Richard also added a link with the problem that he found on the interaction between pbr and setup_tools16:35
SotKI got around that by updating the version of pbr I was using16:35
SotKbut maybe it wasn't the same problem16:35
ssam2I don't know if he tried that16:35
jjardonradiofree: nice! All of it? (I see more patches in our branch than the ones applied in master)16:35
SotKthe version we are using in baserock at the moment is pretty old I think16:35
pedroalvarezI remember that there was a way to check the kernel configuration of baserock system, but I forgot how to do t hat16:36
radiofreei'll give it a go16:37
pedroalvarezdoes anyone remember how to do that?16:37
ssam2is it in configure-commands in /baserock/ ?16:38
pedroalvarezssam2: yeah, but that doesn't include the ones that are included by default16:39
paulsher1oodpedroalvarez: there was a docker script i used, but iirc it got dropped 16:39
pedroalvarezoh, looks like `cat /proc/config.gz |gunzip` is what I was looking for16:45
Kinnisonzgrep exists too16:47
pedroalvareznice! :)16:47
pedroalvarezthis file is generated becaue IKCONFIG is enabled in the kernel16:47
*** wdutch [] has quit [Quit: Quit]17:13
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock17:25
*** tiagogomes [] has quit [Quit: Leaving]17:32
*** locallycompact [] has quit [Ping timeout: 245 seconds]17:36
*** tiagogomes [] has joined #baserock17:40
*** tiagogomes [] has quit [Client Quit]17:40
radiofreejjardon: words \o/17:41
jjardonradiofree: +117:41
radiofreepatch "Use libdrm master"
radiofreetested on ARM17:46
jjardonradiofree: remove the "unpetrify-ref" and you have +217:47
radiofreeis unpetrify-ref not a thing anymore?17:48
KinnisonI thought it still was17:48
radiofreeit's a least useful for translating the sha to a branch17:48
jjardonradiofree: no if you are building from master, at least is what remember from a richard_maw review of one of my patches17:50
franredplease leave the unpetrify-ref, it is useful to know where the sha1 comes from17:50
radiofreepersonally i'd prefer to leave it in, that sha could be from anywhere, it's a good indication of the branch17:50
Kinnisonwith an unpetrify-ref of 'master' it's a good hint that it may be unreproducible17:51
*** bashrc [] has quit [Quit: Lost terminal]17:51
KinnisonSo it's useful information17:51
ssam2radiofree: +1 if you keep the unpetrify-ref :)17:51
radiofreei'll test it in an x86 vm tomorrow first... you never know....17:52
pedroalvarezradiofree: thanks for bearing that in mind :)17:53
*** mdunford [] has quit [Ping timeout: 250 seconds]17:56
*** mariaderidder [] has quit [Quit: Ex-Chat]17:57
jjardonis there a doc somewhere that explain what is each parameter? So we can refer to it to explain what "unpetrify-ref" is for17:59
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]18:04
ssam2jjardon: lots of stuff is likely to change soon18:05
ssam2but documents how things have been in the past18:06
*** ssam2 [] has quit [Quit: Leaving]18:06
jjardonthanks! but unfortunately there is not mention to "unpetrify-ref" there ;)18:08
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock18:10
jjardonwhat Sam mean about "lots of stuff is likely to change soon" ?18:10
jjardonany big changes comming?18:13
*** Guest84456 [] has quit [Quit: Leaving]18:24
*** genii [~quassel@ubuntu/member/genii] has joined #baserock18:26
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]19:00
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock19:41

Generated by 2.14.0 by Marius Gedminas - find it at!