IRC logs for #baserock for Monday, 2015-09-28

*** inara has quit IRC00:03
*** inara has joined #baserock00:07
*** bfletcher_ has joined #baserock01:34
*** doffm_ has joined #baserock01:36
*** Zara_ has joined #baserock01:36
*** bfletcher has quit IRC01:38
*** Zara has quit IRC01:38
*** doffm has quit IRC01:38
*** persia_ has quit IRC04:39
*** persia has quit IRC04:39
*** persia has joined #baserock04:40
*** persia has joined #baserock04:40
*** persia_ has joined #baserock04:40
*** gtristan has joined #baserock05:55
gtristanhi jjardon :)06:00
gtristanjjardon, anyway I'm not entirely sure what is your timezone... this weekend was a major korean holiday, only looked briefly into getting started06:04
gtristanI'm just about to try to build your branch06:05
*** paulw has joined #baserock06:37
jjardonhey gtristan! :)07:07
jjardonmy branch has been merged, but still have issues when you try to run it; AFAIK one of the problems where the driver were not able to access the hardware for a problem in the logind configuration in the xserver07:09
jjardonalso, more components should be added to the stratum, like gdm, gnome-control-center... and glade of course ;)07:10
gtristanjjardon, ah, well, let me try get through building it :)07:13
gtristanbetter get the hang of baserock and ybd07:14
gtristanalso, any idea why ybd asks for root (from the 'Quick Start' on the github page) ?07:14
gtristanseems ominous, I'm trying to build without root and so far going smooth07:15
jjardongtristan: AFAIK you need root to create/enter the chroot07:15
jjardongtristan: I normally use the chroot offered here to build stuff:
gtristanjjardon, ybd mentions that there is no need for the chroot (lists this as one of the main differences from 'morph')07:19
gtristanjjardon, well, stop me if I'm off-base but, I am trying to build the gnome 'system' with ybd, my expectation is the output will be an image, which I can then try to deploy & test either on USB or a VM07:20
jjardongtristan: I guess you should ignore me and ask paulsherwood about ybd then :)07:22
gtristanyeah he specifically asked I use ybd :)07:22
jjardonwhen you build a system, then it can be deployed in several things, check the cluster directory to diferent types of deployments07:23
gtristanso we'll see where this gets, there is a lot of downloading, and I expect the build to fail *somewhere* with the root requirement07:23
jjardonso, in our case ybd build systems/gnome....  and then ybd deploy clusters/gnome-system-x86_64-deploy.morph07:24
jjardonin this case this will generate and image that you can dd in a USB and try in real hardware07:24
jjardonor boot up it directly in a vm07:24
gtristanI see, was just looking that file07:25
gtristanlocation: /gnome-system-x86_64.img07:25
gtristanwould that mean... it wants to deploy an image at / ?07:25
gtristanI guess I will find that out07:26
gtristanmaybe thats the root requirement, would be silly though07:26
jjardonmake a note about all the weird things you found; Id suggest to put all of them in a page in the wiki somewhere, or send them to the mailing list07:27
gtristanI am building a text file atm :)07:27
*** fay_ has joined #baserock07:33
gtristanjjardon, what timezone are you in btw ?07:33
SotKIIRC root is needed for some mounts that get done when sandboxing the build07:33
SotKmy knowledge may be out of date though07:33
pedroalvarezIt would have been morph faster and :)07:34
SotKalso, ybd doesn't have 'build' or 'deploy' subcommands, just doing `ybd clusters/gnome-system-x86_64-deploy.morph` should be enough to build and deploy the gnome system07:35
gtristanI see07:35
jjardongtristan: Im in UK, so UTC+107:36
gtristanjjardon, ah ok... I thought maybe you were elsewhere from seeing your gmail presence... you get up early !07:37
jjardongtristan: too early for my taste, yes ;)07:38
wdutchI'm taking CIAT offline briefly for an update07:47
wdutchit's back on07:51
*** tiagogomes_ has joined #baserock07:53
wdutchI can't see any reason why this exited with 1
wdutchpossibly due to the change in delta/xorg-lib-libXt at 06:35UTC?07:56
wdutchalthough on second thought ybd looks to be exiting properly so maybe it's my own fault ...07:57
paulsherwoodgtristan: i think that some of the commands in definitions require root (especially the fhs dirs)08:04
paulsherwoodwould be better if this could be done without root of course08:04
paulsherwoodand as SotK says some of the sandboxlib commands may need root08:05
paulsherwoodmorph has the same problem iiuc08:06
gtristanpaulsherwood, I see, honestly, this has been spending over an hour in a git clone of the linux kernel sources (I'm about 65% through that now) :-/08:06
paulsherwoodwow. how fast is your connection?08:06
paulsherwoods/fast/slow/ :)08:07
*** mariaderidder has joined #baserock08:07
gtristanI suppose that at least, assuming the build fails for lack of root somewhere... these sources will be cached somewhere in ~/.cache/ybd08:07
gtristanAha !08:07
paulsherwoodyes correct08:07
gtristanpaulsherwood, I see I'm connected to the wrong lan with my laptop... I'll try a switch08:07
gtristanshould normally be 25megabit08:08
paulsherwoodgtristan: for safety please use last tag of ybd to get started. there may be some wierdness in latest master08:08
paulsherwoodgtristan: clone normally takes a few mins for me08:10
*** gtristan has quit IRC08:12
*** jonathanmaw has joined #baserock08:17
pedroalvarezI'm currently wondering how my part is going to be connected to the rest of ciat08:30
jjardonpedroalvarez: your part?08:31
pedroalvarezthe testing part of ciat08:31
jjardontalking about ciat; is there any ETA about when it is going to build master of definitions?08:32
*** gtristan has joined #baserock08:32
jjardonand where are the morphologies sources  that its building at the moment?08:33
pedroalvarezjjardon: I don't think we are aiming for that yet08:33
jjardonpedroalvarez: ah, ok. what is the aim then?08:35
pedroalvarezjjardon: update components in definitions (master) and, build systems with those changes, run some tests, and then give some feedback of the integration08:36
pedroalvarez(that sounds like building master of definitions, i guess, but it's not)08:38
*** rdale has joined #baserock08:38
Kinnisonjjardon: It'll be a week or two before we consider switching to definitions master for CIAT08:39
Kinnisonjjardon: we have a number of other tasks to complete first, including supporting multiple pipelines for different integration sets08:39
*** toscalix__ has joined #baserock08:44
jonathanmawI got an E-mail this morning from a guy asking me a question about Baserock, which is somehow a matter of life and death08:51
rjekjonathanmaw: I would reply asking if you have his permission to forward it to the mailing list.08:52
jonathanmawthe life or death question is whether it's possible to compile BR on arm9 architectures, specifically the freescale imx708:52
rjekI think we've done it before, but not recently08:53
*** toscalix__ is now known as toscalix08:56
wdutchBaserock: a matter of life and death. New tagline right there08:58
wdutchspeaking of branding, are there any more tshirts? I never got one :(08:59
*** franred has joined #baserock09:00
*** Lachlan1975 has joined #baserock09:07
*** Zara_ is now known as Zara09:13
jjardonpedroalvarez: Kinnison cool, thanks for the info09:20
Kinnisonjjardon: As soon as we can, I'd like for us to have multiple integration pipelines, but that will result in many candidates for merges being offered up so we need to be prepared to deal with the influx of patches09:21
*** ssam2 has joined #baserock09:27
*** ChanServ sets mode: +v ssam209:27
*** bjdooks has quit IRC09:53
*** bjdooks has joined #baserock09:54
radiofreedon't know if this is relevant to CIAT
radiofreeAnitya seems interesting though09:58
KinnisonInteresting idea09:59
KinnisonI don't think Debian has a service for uupdate checks09:59
* Kinnison looks09:59
*** mariaderidder has quit IRC10:03
richard_mawpedroalvarez: is the latest version of your work to get Baserock on AWS pushed to baserock/pedroalvarez/xen-guest ?10:14
pedroalvarezrichard_maw: yes. I didn't do too much10:17
pedroalvarezrichard_maw: I tried to deploy an image, and fix any problem by modifying the deployed image10:17
*** fay_ has quit IRC10:17
*** mariaderidder has joined #baserock10:18
pedroalvarezand the only hacks I did were:10:20
pedroalvarez- create symlinks for all the systems/default/run/* folder in /10:20
pedroalvarez- Copy /boot/grub folder i from an Ubuntu image into /boot10:20
pedroalvarez(I was wondering if it was failing because stupid tests, or because AWS actually tries to do somethiing with that)10:21
ssam2radiofree: looks really interesting!10:21
pedroalvarezwdutch: I think I need help from you10:21
wdutchpedroalvarez: wassup?10:22 has to change a lot to support the multi-testing10:22
ssam2radiofree: the idea of having all the infrastructure communicate thru a federated message bus is really interesting too10:22
wdutchpedroalvarez: yep, looks like it10:23
pedroalvarezwdutch: and I'm a bit confused about how to proceed10:24
ssam2and seems it has multiple backends to detect a 'release' on different hosting sites:
wdutchpedroalvarez: I'm not familiar with the state of testing but as a black box, on a deployment it needs to know which tests to run and where to run them, am I correct?10:26
pedroalvarezwdutch: it is currently having as input the image and the sha1, and with multi-testing this would need this kind of info:
pedroalvarezyes, where is the image, name for it, and where to run them, yes10:27
richard_mawtiredcat: I can't represent the look of glee I have on my face from your patch to make morph produce a progress bar for artifact downloads in the textual medium10:28
pedroalvarezwdutch: the "test to run" bit, not sure, it's defined in another file. I'm not sure about that. It looks like a lot of data already10:29
wdutchokay, as it stands the buildslave creates the openstack instances?10:30
wdutchthen ssh's into that?10:31
pedroalvarez(as before I think)10:31
wdutchI have to confess to having been a bit too preoccupied to pay close attention10:31
pedroalvarezthat's not a problem :)10:32
wdutchall the trigger does is clone the testing tool and start the test, could it not pass the yaml file or a url to it rather than all the arguments it currently does?10:33
pedroalvarezThat could work, but then we have to figure out when those files are generated10:34
wdutchyep, the problem of configuring the pipeline is tricky, I'm drafting an email at the moment on the topic hoping the have a discussion on baserock-dev10:35
wdutchare they going to be generated? or would it be better to have a collection of predefined cases?10:37
wdutchoh you need to IDs put in at least10:37
pedroalvareztrue, makes sense to have a set of predefinied cases10:40
*** fay_ has joined #baserock10:42
toscalixradiofree: it sems anitya notifies you when there is a change on a repo/project you pre-define. Through the docu, I do not see it goes further10:58
radiofreeyeah, but i guess we pre-define the project by its presence on g.b.o?11:00
radiofreealso it builds the component11:00
ssam2does it? I thought it just sent a message on the fedora message bus thingy11:06
ssam2which caused another component to file a bug in bugzilla saying ''someone needs to update this package''11:06
paulsherwoodKinnison: do we really *need* overrides for some reason? now i understand now bitbake supports that, it seems to create more problems than it solves11:23
paulsherwood(eg folks fork implicitly every time they override)11:23
radiofreessam2: at least according to the blog it does "Anitya doesn’t stop there; it also downloads the new version and tries to run a test build. It then leaves a message in the Bugzilla bug to tell the maintainer if the new release is a simple update."11:24
ssam2ah, cool. didn't see that11:24
*** Lachlan1975 has quit IRC11:39
*** paulw has quit IRC11:41
* SotK notices his weekend javascript experiment fails to take into account "interrupted" builds11:45
toscalixssam2: I didn't see it either but I think that the tool that builds is not anitya but the common build tool that fedora uses11:47
toscalixwe would need to check the code to find out the level of integration it has. The blog provides a different view compared to the git description11:52
ssam2toscalix, radiofree: Seems that Anitya just sends messages (and is thus a pretty generic tool), then 'the new hostness' kicks off a build on Fedora infrastructure following the message12:01
ssam2'the new hotness' todo list has'Kick off koji scratch builds of new upstream releases and comment in existing bugs with the status.' as a TODO item, but there does seem to be some code already12:01
Kinnisonpaulsherwood: I think it would reduce the pain of "branching a system"12:22
*** gtristan has quit IRC12:33
richard_maw seems to be a good reference for making AWS images. It looks like we don't need to have a partitioned disk image, but ec2-import-instance may be the wrong tool.12:50
paulsherwoodKinnison: while increasing the pain of maintaining the system?12:50
Kinnisonpaulsherwood: unclear, engineers are used to leaf-type resolution, but if you dislike it then perhaps it's not good12:51
Kinnisonpaulsherwood: as I said, I'd not thought through all the ramifications, just thought it might be nice and so suggested it for consideration12:51
paulsherwoodI just think it's like what bitbake has done12:52
paulsherwoodI'm not sure myself12:52
pedroalvarezYay! multi-testing working
pedroalvarezrichard_maw: I tried another one, that failed telling me to use the import-instance command12:53
paulsherwoodpedroalvarez: w00t?!12:54
pedroalvarezpaulsherwood: semi-woot :)12:55
paulsherwoodis that booting a baserock system on AWS?12:55
pedroalvareznope, but booting various systems for one test12:56
paulsherwoodok, still w00t12:56
pedroalvarezrunning some tests in one of them, and others in the other one12:56
richard_mawpaulsherwood: that's the goal of my task, the guide I found was for CentOS, but swapping out the steps for installing the rootfs with one provided earlier ought to be sufficient12:56
paulsherwoodyes but that's cheating a bit??12:57
paulsherwoodstill if it works :)12:58
richard_mawnot if the rootfs provided earlier was a Baserock one, then it's just a stepping stone towards integrating it in properly12:58
pedroalvarezrichard_maw: do you think that integrating grub will be needed?12:59
richard_mawpedroalvarez: I'm unsure. Potentially the PV-GRUB thing is just a replacement for needing to specify the AKI alongside, by having it read the grub config to determine the kernel, which then would attempt to boot it with the arguments separate.13:01
pedroalvarezI see13:02
KinnisonPV-GRUB is not PVM not HVM13:07
Kinnisonwe're using HVM AMIs13:07
KinnisonHopefully that should be less doom-onna-stick13:08
* richard_maw has more keywords for refining his search for documentation now13:09
*** gtristan has joined #baserock13:12
wdutchif I've built a set of artefacts from a system morph, is the sha of definitions and the name of the system enough to know exactly what's been built?13:37
KinnisonIt should be, yes13:37
KinnisonProviding the definitions are properly locked down13:38
Kinnisonif they have non-sha ref: entries then no, it's not enough13:38
richard_mawassuming the same version of the build tool, definitions locked down, and that the build tool is deterministic about what it determines it should build13:38
wdutchcool, ciat now logs this information along with the slave it was done on13:38
wdutchso we know up to artefacts being deleted what is where ready for deployment13:39
wdutch*up to and not including13:39
paulsherwoodybd creates a trees file which would close the non-sha ref holw13:39
wdutcha change in systemd has successfully and quietly gone through the whole ciat pipeline :)13:51
*** franred has quit IRC13:53
ssam2wdutch: nice! how long did it take, out of interest ?13:54
richard_maw1:40 approx by the looks of it13:55
ssam2not too bad. better than the 27 hours that our previous design of Mason would have taken :-)13:55
richard_mawthat'll be because it's building a GDP system rather than something smaller13:55
richard_mawssam2: the previous implementation at least, there's been designs for mason that were roughly the shape of what it is now about 3 iterations ago13:57
*** franred has joined #baserock13:58
tiagogomes_mmm, bison is failing to build for me14:02
tiagogomes_./bootstrap: eval: line 1: rsync: not found14:02
Kinnisontiagogomes_: that implies a lack of networking and a bootstrap which expects networking14:03
Kinnisontiagogomes_: :-(14:03
KinnisonWhich reminds me...14:03
richard_mawtiagogomes_: I thought bison was one of those programs which claimed to need rsync, but actually just need some method to get gnulib, and we already provided it14:04
Kinnisonwdutch: How hard might it be to build and install linux-user-chroot onto the ciat boxes?  Currently we're using ybd in "less reliable than the titanic" mode14:04
wdutchI don't want the ciat box to sink :(14:05
* wdutch will look into it14:05
richard_mawtbh, if you can do that, you can probably run morph on there14:06
pedroalvarezmorph build might be a bit slower14:11
KinnisonShort-term it's likely to be, yes14:11
pedroalvarezwe have a script to create various distbuild-workers, that would be faster14:12
richard_mawpedroalvarez: AIUI that handles parallel build by having multiple artifact caches and transferring artifacts between them, so it less efficient than proper cache locking14:13
richard_mawand morph doesn't cope with something else succeeding in putting something in the cache first14:13
paulsherwoodybd does14:13
richard_mawthat was my point14:16
pedroalvarezalso, morph distbuild still builds the same thing various times :/14:17
tiagogomes_bison build fix:
wdutchis it okay to manually build linux-user-chroot from source on aws or do we need a less snowflakey solution?14:31
ssam2it's pretty trivial to build from source, i would go for it14:34
ssam2jjardon: you should read
ssam2I guess this wouldn't have happened if "bootstrap" / "prepare source tree" commands weren't mixed up with configure-commands as they are currenlty14:36
jjardonssam2: yeah. Also thats a problem now since we added to the other "prepare source tree" scripts (autogen[.sh])15:05
ssam2yeah.. more reasons it should have been done as a change to VERSION15:05
ssam2or waited until we could use definitions version 7 and the DEFAULTS file, so it would be obvious from definitions.git what had changed15:06
jjardonssam2: we can revert those changes if you want then15:06
ssam2maybe. It only matters if we're going to do a stable release soon I think15:07
* paulsherwood wonders when next genivi release is15:08
jjardonssam2: Id say revert them, so we exercise our revert process; I do not actually know how this work with gerrit; do we have to do something special?15:15
ssam2jjardon: it's pretty easy, you can click 'revert' on the change screen and it'll create a new change that does the revert15:20
ssam2do you want to try it then?15:20
jjardonssam2: oh, so webby :) Im pretty busy at the moment, feel free to do it or I will do it tonigh if you agree with the revertion15:22
ssam2i'll do it15:22
ssam2although.. i think the real issue though is use of pre-configure-commands where you should have used configure-commands15:23
ssam2if bison.morph had always used configure-commands (to override the default) instead of pre-configure-commands (to extend the default) then it wouldn't have happened15:24
ssam2maybe it's more useful to do a patch that uses configure-commands instead of pre-configure-commands when we want to override the defaults15:24
*** toscalix has quit IRC15:34
pedroalvarezpaulsherwood: 16 october15:48
tiagogomes_ssam2, is a big commit :)15:51
*** mariaderidder has quit IRC15:55
richard_mawtlsa: I was impressed by when pedroalvarez showed it in a meeting16:01
pedroalvarezthat was tlsa and SotK this saturday :)16:02
pedroalvarezbtw, buildbot still thinks that is on
pedroalvarezwdutch: I guess you didn't manage to find how to configure that16:03
wdutchit does?16:03
* richard_maw clicked buttons and was impressed by it showing links to logs16:03
SotKwdutch: indeed, the log links are all wrong :(16:03
wdutchso it does16:04
pedroalvarez has on it16:05
pedroalvarezindeed! ta!16:05
pedroalvarezha! links wokring :)16:05
* wdutch will put in master for now16:05
SotKwdutch: thanks!16:05
wdutchweird it is in master16:06
wdutchnot sure what happened there16:06
* SotK wonders if there is a way to obtain the different paths through the pipeline from the JSON API for the purpose of arranging the coloured boxes in something other than "ascending order based on the first character"16:10
* richard_maw doesn't think that information exists16:10
pedroalvarezI wonder if there is a way to get what builds were triggered for another build16:11
richard_mawlater builds are triggered by the last thing that a build does is process the trigger16:12
richard_mawthere's no configuration in place for "after this job succeeds, do these"16:12
richard_mawsince at least for the firehose case, this isn't statically defineable16:12
jjardonssam2: agree, what about this  route: use configure-commands in chunks for now -> add new "prepare source tree" to buils steps -> fix chunks16:13
pedroalvarezso there isn't a easy way of getting all the information of the builds triggered for a firehose integration16:13
jjardonssam2: AFAIK wait for definitions version7 will  make this much more easy, right?16:14
* richard_maw reconfigured CIAT so that kernel and systemd updates trigger armv8 builds16:15
richard_mawbase systems currently, since I forgot we also wanted to change that to the build systems16:15
jjardonwhat are the plans of the CIAT? trigger a rebuild for every commit? or every new tag? for every component or only for a few ones?16:17
richard_mawmore usefully: It depends onthe component, but we want to be able to do all those things16:18
jjardon:) nice16:18
ssam2jjardon: maybe:)16:19
richard_mawwe're also hoping to do distributed builds for ARM, since we can't get meaty enough builders from any of our current vendors16:20
jjardonwhy not make the  x86 builds distributed as well, so everything is building in the same way?16:22
jjardonssam2: is any patch pending in morph before a release can be made?16:23
pedroalvarezdo we have any http server installed in ciat.b.o?16:23
* radiofree has a meaty arm on his desk that no one seems to want to use16:23
richard_mawpedroalvarez: you've got python haven't you?16:23
ssam2jjardon: the version 7 patches are still not merged16:23
ssam2in Morph16:23
pedroalvarezrichard_maw: hm... and lack of experience16:24
* pedroalvarez will research16:24
pedroalvarezSimpleHTTPServer module will do it16:25
richard_mawI think you can just do `python -m SimpleHTTPServer` and it serves the current directory16:25
pedroalvarezrichard_maw: thanks :)16:25
pedroalvarezSotK: quick question, do you have the sources of the UI you developed in a repo I can clone?16:29
pedroalvarezSotK: If not, nevermind16:29
SotKpedroalvarez: not pushed anywhere sadly16:29
pedroalvarezI have already copied the sources, no worries :)16:30
*** tiagogomes_ has quit IRC16:30
* pedroalvarez pushes it to git://
*** rdale has quit IRC16:37
* SotK imagines that has a much nicer commit history than my repo, despite being (I assume) only one commit16:38
paulsherwoodpedroalvarez: looks nice and small, at least :)16:39
*** nowster has quit IRC16:44
pedroalvarezOk, that's now in :)16:44
pedroalvarezI really hope I can use it as base16:44
paulsherwoodradiofree: based on what i now know about how to tune for build times, do you fancy re-running a build on your box?16:46
*** fay_ has quit IRC16:46
paulsherwoodi think instances should be cores/10 rounded up, and max-jobs should be cores/instances16:47
radiofreeYes I can try tomorrow16:51
* radiofree points out it isn't his machine16:51
paulsherwoods/your box/the box on your desk/16:53
*** franred has quit IRC16:55
paulsherwood1, 5, 3, 2, 416:58
paulsherwood??? :)16:58
paulsherwoodpedroalvarez: ^^16:59
myselfWhat is, the firing order of a Lancia straight-5?16:59
*** ssam2 has quit IRC16:59
*** nowster has joined #baserock16:59
paulsherwoodthat's the (strange) order of
richard_mawdunno, probably back by midnight17:00
richard_mawwrong channel17:00
* paulsherwood was trying to parse richard_maw's comment17:00
* richard_maw apologises for causing paulsherwood to have to type17:01
paulsherwoodi cant help myself anyway :)17:02
richard_mawit looks like it's in the right order to me17:02
* paulsherwood is safari-ing17:02
paulsherwoodcorrect on chromw17:03
richard_mawcould be that the data structure the items are in is has an unordered implementation in safari17:03
*** jonathanmaw has quit IRC17:04
pedroalvarezpaulsherwood: yup, will do17:04
paulsherwoodrichard_maw: is that arm build using last tag of ybd, or master?17:12
* paulsherwood should have realized, he can check for himself... 15-09-28 00:00:00 [SETUP] version is HEAD 15.39 (15.39 + 0 commits)17:16
paulsherwoodinstances: 5 won't work very well, though if that's running on what i think it is17:17
paulsherwoodfaster would be no instances setting, and run the job on multiple build machines one running kbas, with the others uploading artifacts to it17:20
*** zoli__ has joined #baserock17:59
*** SotK_ has joined #baserock18:33
SotK_paulsherwood: my bad, the sort function returns a Boolean rather than an integer18:33
SotK_Chrome doesn't care but Safari does18:34
*** zoli__ has quit IRC18:35
*** zoli__ has joined #baserock18:37
*** SotK_ has quit IRC18:38
*** zoli__ has quit IRC18:44
pedroalvarezoh, I though paulsherwood was requesting a different ordering for the UI18:52
SotKif someone can provide a way to know a sensible arrangement that's better than numerical order automatically it should be easy to implement19:10
paulsherwoodSotK: i think folks are looking into something to show parallelisation, so i suggest wait to see what they come up with19:12
*** Stanto has quit IRC19:29
* paulsherwood creates a new baserock vm... and finds networking doesn't seem to work out of the box... it used to?19:31
*** Stanto has joined #baserock19:31
paulsherwoodalso nf_conntrack: automatic helper assignment is deprecated and it will be removed soon...19:35
* SotK fixes the ordering and also pushes his repo to his github19:42
SotK( is the fix, if there is anyone around with access to the CIAT box/the repo Pedro made)19:45
pedroalvarezSotK: thanks! I will apply the change19:45
pedroalvarezdone, not sure if I have to restart the http server19:52
SotKshouldn't have to19:58
paulsherwoodyes, that's fixed20:00
* SotK wants to add a page for all the builds in a given "builder"20:04
paulsherwoodanyone know what the trick is for networking these days? it hasnt found an ip address, ping says 'bad address'20:08
pedroalvarezpaulsherwood: does the vm have an IP?20:10
pedroalvarez /o\20:11
pedroalvarez`ifconfig -a`?20:12
paulsherwood (with -a)20:13
pedroalvarez`cat /etc/systemd/network/*`20:16 files there  define how systemd configures your network20:16
pedroalvarezSotK: I'll be having a look this week at how to improve the UI20:18
paulsherwoodsorry Name= e*20:18
pedroalvarezyup, that's fine20:19
pedroalvarezI assume that EOF is not actually in the file20:19
paulsherwoodcorrect :)20:19
paulsherwoodthere's nothing more in the file except a blank line between the two sections20:20
pedroalvarezbah, then the problem is another thing20:20
paulsherwoodi'm using latest virtualbox, i wonder if that could affect things?20:20
pedroalvarezwell, I hope virtualbox is connecting the VM to a network20:21
pedroalvarezI have no idea how to check that though20:21
myselfboot another OS you know and trust, in an identically-configured VM.20:22
pedroalvarezthat would be a good way to test VBox, indeed20:24
myselfIf it's configured for bridged networking, make sure you have the appropriate host adapter selected, and be aware that many of the "nicer" wireless controllers (the sort found in business) won't hand out more than one IP address to a single wifi MAC; I don't know if this goes for wired gear too.20:24
myselfYou can always configure it for NAT, which hides the guest behind the host's IP; traffic appears to come from the vbox process on the host, rather than another box behind the same NIC/MAC.20:24
myself(which is to say, in virtualbox, set the network connection type to NAT)20:25
myselfIn that mode, other hosts on the LAN won't be able to reach inward to the VM, but the VM should "see out" and be able to work mostly-normally. It'll get a DHCP address from virtualbox's internal range.20:25
paulsherwoodmyself: thanks for the ideas20:27
paulsherwoodenabling 'promiscuous mode' and restarting seems to have fixed it20:28
myselfnice :) I'm just lurking for baserock stuff, but have more than a little virtualbox experience20:28
paulsherwoodi think i may be the only mac vbox 5 user20:36
myselfSo, one other strategy, is to use a USB NIC with USB passthrough to give the guest a "real nic" all to itself, give it a dedicated port on your switch, and eliminate all the network virtualization entirely20:39
pedroalvarezwow, more than a fix sounds like a pain21:09
*** gtristan has quit IRC21:23
*** gtristan has joined #baserock21:27
myselfYeah, but on the other hand, it can enable some amusing tricks. The packets literally never touch the host in a way that the host would recognize them as network traffic, so you can use a VM to do network-stack testing on the host without worrying about the fact that the test box is a VM, for instance. It's nice for cutting layers out of the question.22:18

Generated by 2.15.3 by Marius Gedminas - find it at!