IRC logs for #baserock for Wednesday, 2015-02-04

*** zoli__ has joined #baserock06:30
*** petefoth has joined #baserock07:23
*** jjardon has quit IRC08:24
*** jjardon has joined #baserock08:25
*** tiagogomes_ has joined #baserock08:34
*** grahamfinney__ has joined #baserock08:35
radiofreequestion about the UUID used in the fstab08:42
radiofreeis there someone on first boot of a system that changes/modifies that?08:43
radiofreei'm changing morph upgrade to allow you to use rawdisk image files, it works, the only problem at the moment is the fstab08:43
radiofreethere's already one in a rawdisk image, but because the UUID is different nothing mounts08:44
radiofreebut if i use that rawdisk image, dd it to the board, it will work right?08:44
*** mariaderidder has joined #baserock08:46
SotKradiofree: the UUIDs are calculated here:
*** mdunford has quit IRC08:49
SotKthat function is called in the rawdisk write extension, so I guess the UUIDs are set at deploy time08:50
* SotK could be wrong though08:50
radiofreeSotK: yeah, so the problem i have is the rawdisk image already has an fstab, so when you try to do an upgrade it just copies over that existing fstab with the wrong UUIDs08:51
*** zoli__ has quit IRC08:51
*** mdunford has joined #baserock08:51
radiofreei was thinking about making "complete_fstab_for_btrfs_layout" in rawdisk.write use a temporary directory to generate the fstab, then manually copy that over as the last stage of the upgrade?08:52
radiofreeideally i want to mount the rawdisk image as read-only, i don't want to modify anything on that (i.e the fstab)08:52
radiofreei was thinking about making "complete_fstab_for_btrfs_layout" in rawdisk.write... i mean ssh-rsync08:53
* radiofree will send patches for it anyway08:54
*** sherm_ has joined #baserock08:55
*** tiagogomes_ has quit IRC08:57
* SotK wonders if this could be caused by it getting the UUID before actually doing anything to the system08:59
*** mdunford has quit IRC08:59
*** bashrc_ has joined #baserock09:02
*** fay__ has quit IRC09:04
*** fay_ has joined #baserock09:05
radiofreeSotK: for upgrades, it gets the UUID from the system it's uprading09:07
radiofreewhich is fine, but ssh-rsync assumes there's no /etc/fstab in the system you're trying to upgradew to09:08
radiofreein my case, my upgrade source is a rawdisk image (/src/weston-foo-arm.img)09:09
radiofreewhich already has an fstab, so that just gets copied over09:09
radiofreei'll make the change i suggested and send a patch, it won't change the existing behaviour at least09:09
*** jonathanmaw has joined #baserock09:10
*** mdunford has joined #baserock09:13
*** zoli__ has joined #baserock09:19
*** gary_perkins has joined #baserock09:47
*** ssam2 has joined #baserock09:48
*** ChanServ sets mode: +v ssam209:48
*** tiagogomes_ has joined #baserock09:51
tiagogomes_ssam2, thanks for merging the newer Make tarball, can you promote it please?09:56
*** bfletcher has joined #baserock09:56
*** franred has joined #baserock09:58
petefothAre people here still using Trello or Taiga? What is the current status wrt to a TaskTracker and / or a defect tracker for Baserock?09:58
ssam2tiagogomes_: yes, in a moment09:58
ssam2petefoth: I've not used either Trello or Taiga09:58
tiagogomes_ssam2 ta09:59
ssam2petefoth: I've made some progress on setting up Storyboard, but I want to get Gerrit working before I go back to looking at Storyboard09:59
ssam2it should be pretty simple once Gerrit is done though10:00
ssam2since some of the work, such as setting up automated database backups, actually overlaps anyway10:00
petefothssam2: thanks. I'll keep working with the StoryBoard incubator page for now then10:01
ssam2sounds good10:02
ssam2tiagogomes_: moved delta/make-tarball to the top of the queue10:02
ssam2Can anyone give my patch series 'definitions: Remove all mention of Ye Olde Mason' a quick review?10:03
ssam2I included a ref change for trove-setup, which actually I don't think I should have done10:03
ssam2because the ref change reflects merging a branch in trove-setup.git and thus it should point to the merge commit and not the head of the branch that was merged10:03
ssam2other than that the patch series is removing two lines from one cluster morph so it's trivial to review10:04
pedroalvarezpetefoth: I'm still using Trello10:05
petefothpedroalvarez: thanks10:05
*** franred has quit IRC10:11
*** Krin has joined #baserock10:15
*** franred has joined #baserock10:15
*** pacon has joined #baserock10:16
ssam2thanks for the review SotK!10:19
*** franred has quit IRC10:21
SotKssam2: you're welcome10:24
nowsterCan anyone help with this: (gcc build problem)
nowstereither ld should be looking in /tools/lib* or the gcc libraries should have gone in /usr/lib64/10:27
*** franred has joined #baserock10:27
pedroalvareznowster: is there a link in /lib/ and it's broken?10:30
*** lachlanmackenzie has joined #baserock10:31
pedroalvareznowster: this symlink is created in stage2-gcc (at the bottom of this file
nowsterpedroalvarez: thanks!10:35
pedroalvarezso, was it broken then?10:35
nowster  if [ "$(echo $TARGET | cut -c -6)" = "x86_64" ]; then  probably needs mips64 in there too!10:36
pedroalvareznowster: oh yes, indeed10:36
nowsterthanks for the pointer10:37
Kinnisonnowster: I'd check with tiago what he did for armv8l64 too10:37
nowsterKinnison: tiago's working with gcc4.9.x. I'm still using gcc4.6.210:37
nowsterchanging it to:10:42
nowster  case "$TARGET" in10:42
nowster      x86_64* | mips64*) libdir=lib64 ;;10:42
nowster      *)                 libdir=lib ;;10:42
nowster  esac10:42
pedroalvarezI wonder why we are currently doing the ' vut -c -6' to target10:48
pedroalvarezI see, because $TARGET is the arch triplet10:49
Kinnisonlinux- ?10:49
ssam2jjardon: were you planning on making a branch that removes 'compression' field from everything in lorries.git ?10:51
ssam2i can do it, but I don't want to have us both send patches at the same time again10:51
Kinnisonif you do, remember that Strip is also deobsolete :)10:51
ssam2is it?10:51
ssam2oh, right.10:51
Kinnisontype and url is pretty much all that tarball uses10:52
ssam2what does Lorry do now?10:52
ssam2if I have a tarball with contents in ./ ?10:52
DavePage_Is there a high-res version of the Baserock logo at ?10:52
pedroalvareznowster: I think that is a sensible solution. Although I wonder if there is a better solution nowadays with newer gcc10:52
pedroalvarezDavePage_: I don't think so. Let me check if I have something around.10:53
pedroalvarezDavePage_: I've some svg files around10:54
pedroalvarezDavePage_: sent10:57
DavePage_Thanks pedroalvarez10:58
*** rdale_ has joined #baserock10:59
*** rdale__ has joined #baserock11:01
*** mdunford has quit IRC11:02
*** rdale has quit IRC11:02
*** rdale_ has quit IRC11:04
jjardonssam2: I was planning on do it yes, but go ahead if you already started11:07
ssam2i hadn't started, go ahead11:10
ssam2i'm doing cleanups in lorry.git first11:10
ssam2note we can remove both 'strip' and 'compression'11:11
KinnisonDoes morph place the artifact ID into the environment during builds?11:12
ssam2not as far as I know11:12
ssam2might be cool, but I can't think why you'd need it11:12
*** rdale has joined #baserock11:14
jjardonssam2: ok, on it11:14
jjardonok to update the location of usbhid-dump?
richard_mawKinnison: unless it's been added recently, it doesn't11:15
ssam2jjardon: +111:16
pedroalvarezjjardon: do they share the same history? (same sha1 in tags for example?)11:16
richard_mawssam2: You could put the cache key in the build-id of the binaries, so you could work backwards from an executable to what built it11:16
*** mdunford has joined #baserock11:16
*** rdale__ has quit IRC11:17
ssam2richard_maw: what's build-id ? is that some special field that the ELF executable file format provides?11:17
richard_mawssam2: also, you could get ccache to look at the build-id of the toolchain to calculate its cache keys11:17
richard_mawssam2: precisely11:17
ssam2we've still not proved whether or not ccache is actually useful. I think it's been broken a *long* time and we haven't done anything about it11:18
Kinnisonrichard_maw: noted.11:18
richard_mawssam2: we did prove it was worth it for building the kernel a while back, but we haven't continued to test whether it's useful, and I don't recall where the results are now11:20
ssam2ah, right11:20
jjardonpedroalvarez: yes11:20
pedroalvarezjjardon: +1 then11:20
*** wdutch has joined #baserock11:21
wdutchdoes there exist a list of the chunks in definitions?11:24
ssam2we don't manually maintain a list somewhere in text form, if that's what you mean11:24
tiagogomes_anyone knows if we have a package in baserock which installs sys/acl.h11:25
ssam2grepping for 'sys/acl.h' in /baserock/*.meta in a Baserock system will tell you11:26
ssam2I don't know off hand, though11:26
tiagogomes_ssam2 thanks11:26
pedroalvareztiagogomes_: ?11:30
tiagogomes_pedroalvarez righto, but that isn't in baserock, yet11:30
pedroalvareztiagogomes_: no, it's lorried though11:31
tiagogomes_acl is needed to compile last btrfs-progs11:31
SotKwdutch: `git grep "\- name:" -- "strata/*.morph"` will get you a list of them I think11:31
tiagogomes_hmm, if is lorries I may try to add it quickly11:31
wdutchSotK: ta11:32
tiagogomes_the lorry clean up patch is nice. I wounder whether we will switch to yaml for lorries at some point11:48
ssam2would make them a bit easier to write, I guess11:50
*** pacon has quit IRC11:59
*** edcragg has joined #baserock12:03
*** petefoth_ has joined #baserock12:11
*** rjek_ has joined #baserock12:13
*** simonh_ has joined #baserock12:14
*** jmacs_ has joined #baserock12:15
*** JPohlman1 has joined #baserock12:15
*** petefoth1ringham has joined #baserock12:15
*** cyndis_ has joined #baserock12:17
*** tiagogomes_ has quit IRC12:20
*** petefoth has quit IRC12:20
*** mauricemoss_ has quit IRC12:20
*** JPohlmann has quit IRC12:20
*** petefotheringham has quit IRC12:20
*** rjek has quit IRC12:20
*** jmacs has quit IRC12:20
*** cyndis has quit IRC12:20
*** petefoth_ is now known as petefoth12:20
*** tiagogomes_ has joined #baserock12:21
*** locallycompact has joined #baserock12:54
locallycompactQ, How do I make lorry controller forcibly replace branches from an upstream trove if upstream trove has changed history?12:56
jjardonhey, is in red, looking to the last log seems there is not enough space on disk?12:58
paulsherwoodouch :)12:59
paulsherwoodlocallycompact: in a trove you own, or
richard_mawlocallycompact: add a "refspecs" entry to the lorry file, listing force push refspecs, so if you want it to replace all refs you need to add `"refspecs": [ "+refs/heads", "+refs/tags" ]`12:59
paulsherwoodi think i gbo we need to consider on a case-by-case basis12:59
jjardonon the failure log of "2015-02-03 12:29:02", seems the weston image is correctly generated though (!?)13:00
paulsherwoodwould be very bad for gbo to start re-writing history without proper understanding of why13:00
locallycompactrichard_maw: You mean the conf file?13:02
richard_mawlocallycompact: the foo.lorry file. If the lorry in question is from an auto-generated .lorry file then it defaults to force pushes IIRC13:03
locallycompactrichard_maw: I don't have any lorry file, these repos are coming from an upstream trove. Is it automatically handled then?13:04
richard_mawIt *should* automatically allow the force-push then, but I can't rule out the behaviour changing since I last looked at it.13:05
locallycompactrichard_maw: Alright, I'll report back. Thanks13:06
ssam2jjardon: regarding testing your branch on ARM: the build failed because distbuild is broken13:51
jjardonssam2: oh, nevermind then13:52
ssam2is the branch blocking other stuff?13:53
*** petefoth has joined #baserock13:56
*** mdunford has quit IRC13:59
pedroalvarezok, I think that mason is alive again. Thanks jjardon for pointing that out14:05
*** mdunford has joined #baserock14:12
*** rjek_ is now known as rjek14:13
nowsterNext problem... refers to /tools/etc/, not /etc/
*** cyndis_ is now known as cyndis14:20
jjardongreen in :) , thanks pedroalvarez !14:23
pedroalvareznowster: so, it's present in /etc/
pedroalvarez(this one is not going to be easy)14:25
*** petefoth has quit IRC14:28
nowsterI've done this in stage2-fhs. It's ugly:14:30
nowster+- ln -s "$DESTDIR/etc" "$PREFIX/etc"14:30
nowsterrats! ln: /tools/etc/etc: Read-only file system14:30
pedroalvarezit should be - ln -s "$/etc" "$DESTDIR$PREFIX/etc"14:31
pedroalvarez- ln -s "/etc" "$DESTDIR$PREFIX/etc"14:32
pedroalvarezbut it is indeed an ugly hack14:34
pedroalvareznowster: is this building gcc aswell?14:34
*** petefoth has joined #baserock14:36
pedroalvareznowster: I assume that this is the context: you have created a bootstrap tarball for mips64, and you are running the ./native-bootstrap script, right?14:37
nowsternope... this is making build-essential on a system that has already done native-bootstrap14:38
nowster(having fiddled around this problem on the native-bootstrap)14:38
*** petefoth has quit IRC14:42
*** grahamfinney__ has quit IRC14:42
*** petefoth has joined #baserock14:44
*** grahamfinney__ has joined #baserock14:49
*** petefoth has quit IRC14:50
*** petefoth has joined #baserock14:51
nowsterpedroalvarez: I think it's when glibc.morph calls configure....14:51
nowster  --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]14:51
nowsteras it's PROBABLY setting PREFIX to /tools14:51
nowsterDoes that sound reasonable?14:51
pedroalvareznowster: let me check a couple of things14:52
nowsterNote that I'm not on bleeding edge at the moment.14:52
*** grahamfinney__ has quit IRC14:55
*** grahamfinney has joined #baserock14:56
*** grahamfinney_ has joined #baserock14:56
*** grahamfinney has quit IRC14:56
*** grahamfinney has joined #baserock14:56
*** grahamfinney has joined #baserock14:57
pedroalvareznowster: so just to clarify. Something is expecting the library to be in /tools/etc, but it is in /etc15:04 (linker) is expecting to be in /tools/etc but it should be looking in /etc/ as that's where ldconfig has put it.15:06
*** zoli__ has quit IRC15:07
* pedroalvarez is out of ideas15:08
pedroalvarezI wonder if makes sense to remove the stage2-glibc dependency of gcc, since it has the glibc dependency15:09
pedroalvareznowster: have you tried --sysconfdir?15:28
nowsternot yet15:28
pedroalvarezthings I'd try:15:29
pedroalvarez1) remove stage2-glibc dependency in gcc15:29
pedroalvarez2) add --sysconfdir in stage2-glibc15:29
pedroalvarez3) the same but in glibc (this should do nothing but at this point I don't know what else to try)15:29
nowsterit's following glibc that it starts failing.15:30
nowsterso I'm trying 3 first.15:30
pedroalvareznowster: '--sysconfdir=/etc' ?15:31
pedroalvarezmy assumption is that when building gcc, it's using stage2-glibc instead of using glibc (stage3). And the former is misconfigured to look for the in /tools/etc.15:33
nowsterno, it's already built glibc15:43
*** CTtpollard has quit IRC15:44
*** CTtpollard has joined #baserock15:57
tiagogomes_pedroalvarez re removing stage2-glibc, what about the stage2 binaries that dynamically link against /tools/
*** zoli__ has joined #baserock16:06
*** zoli__ has quit IRC16:12
pedroalvareztiagogomes_: true16:13
pedroalvareztiagogomes_: any suggestions to help nowster?16:13
tiagogomes_nothing comes to my mind right now, I can't look deeper at the moment16:17
nowsterHmm... The in *.chunk.glibc-libs.d is correct, but it appears that the one from stage2-glibc overrides it.16:17
pedroalvarezI wonder why we don't have  the same problem in x8616:18
pedroalvarez(or Arm)16:18
ssam2chunks should be installed in a stable order16:18
nowsteroh crap... it's probably because of the /lib64 link.16:18
ssam2from 'furthest away dependency' to 'nearest dependency'16:18
ssam2ah :)16:18
ssam2jjardon: did you look at adding tzdata yet? I will do now if not16:19
radiofreebaserock-system-config-sync looks absolutely terrifying16:22
ssam2yeah :(16:22
ssam2it does have some tests... but they're also terrifying16:23
Kinnisonradiofree: it has to do a tough job, sadly16:23
ssam2all the more reason for it not to be written in Bash16:23
ssam2or even Ash16:23
KinnisonI'm guessing it'll likely have to be written in a compiled language unless we're going to mandate python be present on all systems16:24
ssam2radiofree: for the purposes of adding a special-case hook for /etc/fstab, you'd just need to hook it in under the 'exists exists exists' case16:27
ssam2since if there's no fstab in either the old or the new system, the upgrade will work fine16:27
ssam2and if the fstab merge tool happens to be written in C, that's one less thing we'd have to convert to C later on... :)16:29
pdar_ERROR (ConnectionError): ('Connection aborted.', BadStatusLine("''",))16:42
pdar_wrong window16:42
*** anonymous1179804 has joined #baserock16:51
radiofreessam2: i'm thinking it might be easier to just let baserock-system-config-sync do whatever it wants, and then sanity check the fstab afterwards?16:51
radiofreee.g fstab-check actualuuid16:51
*** anonymous1179804 has left #baserock16:51
*** jonathanmaw has quit IRC16:51
*** anonymous1179804 has joined #baserock16:51
ssam2radiofree: fair enough16:51
ssam2especially if fstab-check gets run from the baserock-system-config-sync process16:52
radiofreeCreate a snapshot of '/tmp/tmpizO1B1/systems/new/orig' in '/tmp/tmpizO1B1/systems/new/run/orig'16:56
radiofreethat's wrong isn't it?16:56
pedroalvarezradiofree: where is that?16:57
radiofreei'm messing with system-version-manager16:57
radiofreeit's the result of doing btrfs subvolume snapshot orig_dir run_dir16:57
radiofreeorig_dir = /tmp/tmpizO1B1/systems/new/orig16:58
radiofreerun_dir = /tmp/tmpizO1B1/systems/new/run16:58
pedroalvarezand it creates .../run/orig instead?16:58
radiofreeyeah... i'm not sure how that's happening16:58
pedroalvarezradiofree: because... run already exists?16:59
radiofreei don't think it does17:00
pedroalvarezI think something similar happens with the `cp` command17:00
radiofreei'll try again17:00
*** pdar_ is now known as pdar17:00
radiofreepedroalvarez: you're right, thanks :)17:01
* pedroalvarez was testing the behaviour of `cp` to compare17:01
nowsterpedroalvarez: I think the following in stage2-glibc might fix things17:09
nowster+  mips64*)17:09
nowster+    install -d "$DESTDIR/lib64"17:09
nowster+    ln -s "$PREFIX/lib/" \17:09
nowster+          "$DESTDIR/lib64/" ;;17:09
nowsteremphasis on *might*17:09
grahamfinneyI'm working on the YBD work which involves stripping down the morph build process. I'm new to BR so I'm clueless really. Anyway the process goes so far and then hits a roadblock and I've stared at the screen long enough now so any help would be welcome. The output is at
richard_mawat a guess, you need to use `os.lstat` rather than `os.stat` because destpath is a dangling symlink?17:22
richard_mawI really can't comment about what's going on though, not knowing the codebase17:25
richard_mawwhat exactly is it trying to do at line 92 in ybd.staging?17:25
*** mariaderidder has quit IRC17:29
*** tiagogomes_ has quit IRC17:30
grahamfinneyGood question17:32
richard_mawat a guess it's trying to copy some file attributes by statting first, in which case it shouldn't be doing that because it's operating on a symlink17:33
grahamfinneyThis is something that works past this point on Paul's setup, but not on mine17:34
grahamfinneyI'll look into what you've said so far. It's  a starting point17:35
*** ssam2 has quit IRC17:38
bashrc_is there any way of running emacs in baserock?17:41
jmacs_Not yet17:42
jmacs_You'd need to build it and all its dependencies17:43
radiofreebashrc_: it has been lorried but a quick git grep of definitions suggests it hasn't been added to any stratum17:43
*** zoli__ has joined #baserock17:50
jjardonssam2: yeah, just did the patch today17:51
jjardon(for tzdata)17:51
*** zoli__ has quit IRC17:52
radiofreethe baserock emacs repo seems to be a bit.. broken?17:59
radiofreecan't clone it17:59
*** jmacs_ is now known as jmacs18:01
jjardonit actually looks broken:
*** bashrc_ has quit IRC18:04
jjardonthe url in the lorry file seems to work fine18:04
*** zoli__ has joined #baserock18:05
*** Krin has quit IRC18:05
jjardonHey! seems the GTK+ repo is not up-to-date anymore (last commit 2014-12-24)
*** sherm_ has quit IRC18:23
*** tiagogomes_ has joined #baserock18:31
*** tiagogomes_ has quit IRC18:44
*** zoli__ has quit IRC18:46
*** zoli__ has joined #baserock18:49
*** grahamfinney has quit IRC18:51
*** franred has quit IRC18:53
*** gary_perkins has quit IRC18:57
*** edcragg has quit IRC19:10
*** grahamfinney has joined #baserock19:20
*** lachlanmackenzie has quit IRC19:23
*** grahamfinney has quit IRC19:25
*** petefoth has quit IRC19:27
*** tiagogomes has joined #baserock19:53
*** zoli__ has quit IRC21:42
*** zoli__ has joined #baserock21:54
*** zoli__ has quit IRC22:06
*** zoli__ has joined #baserock22:08
*** zoli__ has quit IRC23:40
*** tiagogomes has quit IRC23:41
*** tiagogomes_ has joined #baserock23:51

Generated by 2.15.3 by Marius Gedminas - find it at!