| *** tonyb has joined #openstack-dib | 00:27 | |
| tonyb | What are the expectations of mount_mkfs_root with respect to the mount status of the buidldir (/tmp/dib_build.$random) ? I'm seeing it fail because things like /tmp/dib2_build.../mnt/dev/pts is still mounted? What shoudl be handling the unmount of them? | 00:28 |
|---|---|---|
| tonyb | COUld I be missing an element? | 00:29 |
| tonyb | I'm tryign to verify 456055 works | 00:30 |
| greghaynes | tonyb: that's likely not the actual fail. Usually I think that is cleanup after an error. Your real error is likely further up the log history | 00:37 |
| greghaynes | Could be related though | 00:37 |
| tonyb | greghaynes: Thanks I'll look further up ... | 00:43 |
| tonyb | greghaynes: Ahh okay I think I've found the root cause. No idea how to fix it but I now have something to investigate :) | 01:00 |
| greghaynes | tonyb: it'd be nice to also track down why cleanup was failing if you end up with some more info - those bugs can be particularly annoying for users | 01:01 |
| tonyb | greghaynes: I'll let you know. I'm still very much finding my feet here. | 01:04 |
| greghaynes | tonyb: np, if you get stuck and have any logs / etc I'm sure one of us can help | 01:06 |
| tonyb | greghaynes: what seems to be happening is that diskimage_builder/elements/bootloader/finalise.d/50-bootloader#126 is failing with an unbound variable and aborting that script but not triggering a failure, so no cleanup is being done | 01:09 |
| tonyb | dos that sound even remotely plausible? | 01:09 |
| greghaynes | Hrm, failure is signaled by any script exiting with non zero status, and we run everything with set -e. If that's the case it usually means something in that script errored and still had status 0 for some reason | 01:11 |
| greghaynes | Which usually boils down to us using a subshell or a clie not properly setting error status | 01:13 |
| tonyb | okay. I'll look but I don't see a subshell. | 01:16 |
| greghaynes | It also could be a red herring, grub tends to say things are bad when they still work | 01:25 |
| ianw | ohh, suse has managed to build a .qcow2 on nodepool ... first in too long! | 01:29 |
| ianw | i think the gate should be in a better state now | 01:29 |
| tonyb | greghaynes: It isn't grub failing .... I'll get to a paste snippet RSN | 01:31 |
| ianw | tonyb: you ppc-ing? | 01:32 |
| tonyb | ianw: Yeah | 01:32 |
| ianw | cool ... definitely the road less traveled there :) | 01:33 |
| ianw | is it cross-build or native? | 01:33 |
| tonyb | :) I think the centos7 review and your work disagree on $somethign but I don't knwo what yet | 01:33 |
| tonyb | ianw: what diskimag-create command did you use? | 01:34 |
| ianw | tonyb: command for what sorry? | 01:34 |
| tonyb | when you were buildin the images how did you create the image ? "diskimage-create ubuntu-minimal vm" ? or somethign else? | 01:35 |
| ianw | tonyb: I don't believe i ever have created a ppc image. i did spend some time trying to get it working with qemu emulation, but it all started segfaulting. but, "ubuntu-minimal vm" is what the ibm testing does | 01:37 |
| tonyb | Ahh okay. that may explain some of it. | 01:38 |
| tonyb | greghaynes, ianw: http://paste.openstack.org/show/613672/ that shows the local patch I have and the log snippet | 01:42 |
| ianw | /tmp/in_target.d/finalise.d/50-bootloader: line 128: DEVICES[boot]: unbound variable | 01:43 |
| ianw | right, so that should be coming out of dib-block-device | 01:43 |
| ianw | ppc is different in that it has a boot device and the root | 01:44 |
| ianw | we should be using -> http://git.openstack.org/cgit/openstack/diskimage-builder/tree/diskimage_builder/elements/vm/block-device-ppc64el.yaml | 01:45 |
| ianw | as the config on ppc | 01:45 |
| tonyb | Ahhh okay clearly we're not as the yaml I have is different to that | 01:45 |
| tonyb | so that's the root cause but the sympom is that that finalize.d/50-bootloader is gettign the unbound error but not cleaning up | 01:46 |
| ianw | well the root cause is that "50-bootloader" is looking for the path to a "boot" partition and not finding it | 01:47 |
| ianw | the cleanup path is then throwing more errors | 01:47 |
| ianw | the list of devices should be set at http://git.openstack.org/cgit/openstack/diskimage-builder/tree/diskimage_builder/lib/disk-image-create#n421 | 01:48 |
| tonyb | ianw: So I see what's going on ppc64el != ppc64le dib seems to be mixing distro architectures (ppc64el (.deb) and kernel architectures (ppc64el .deb and .rpm) | 01:48 |
| tonyb | AHh so the cleanup is throwing theerrors my bad | 01:48 |
| * tonyb really wishes there was only one architecture name .... how does DIB handle amd64 vs x86_64 ? it's basically the same issue | 01:51 | |
| ianw | wait ... el and le isn't big and little endian? | 01:52 |
| ianw | i thought "le" was little-endian, and the "el" was a play on endian-ess and swapped around indicating "big endian" | 01:53 |
| tonyb | ianw: no | 01:53 |
| tonyb | ppc64 == Big Endian, ppc64el == .deb baseds arch, ppc64le == rpm based distro arch *and* the kernel arch for both ppc64el and ppc64le | 01:54 |
| tonyb | I expect there's lots of confusion about that | 01:55 |
| tonyb | that'll be hard to untangle | 01:55 |
| ianw | !? well there you go, TIL ... | 01:56 |
| openstack | ianw: Error: "?" is not a valid command. | 01:56 |
| ianw | openstack: don't worry, we're all confused | 01:56 |
| tonyb | for testing | 01:57 |
| tonyb | I just copies the ppc64el.yaml to ppc64le.yaml | 01:57 |
| tonyb | Huzzah! that built an image I'll boot it after lunch | 01:59 |
| ianw | tonyb: it seems like it's probably right to just replace our tests with "=~ ppc64" ? | 01:59 |
| ianw | bah, although that's be as you say | 01:59 |
| tonyb | Yeah | 02:00 |
| ianw | ppc64(le|el) or whatever regex works | 02:00 |
| tonyb | I'll cook somethign up | 02:00 |
| tonyb | ianw: Can I just diskimage_builder/elements/vm/block-device-ppc64le.yaml and diskimage_builder/elements/vm/block-device-ppc64el.yaml a sumlinked pair? | 02:02 |
| ianw | tonyb: there's just a if statement in there that selects it, i think it would be better to just modify that | 02:04 |
| ianw | in diskimage_builder/lib/common-functions | 02:04 |
| ianw | at least we can put a comment in there ... | 02:04 |
| openstackgerrit | Merged openstack/diskimage-builder master: Fix mkfs use wrong label option for vfat https://review.openstack.org/476382 | 02:41 |
| tonyb | ianw: something like: http://paste.openstack.org/show/613674/ ? | 03:12 |
| ianw | tonyb: yeah, ugly ... but the whole thing is | 03:28 |
| ianw | only other thought is, maybe we "standardise" on ppc64le (or el) and just auto-convert where required for centos? | 03:28 |
| ianw | otherwise you have to build ARCH=ppc64le for centos and ARCH=ppc64el for debuntu? | 03:28 |
| ianw | is that more or less confusing? I don't know :) | 03:29 |
| tonyb | ianw: Yeah I don't know. I'm doing a small amount of auto correcting in the changes I have to get centos and rhel7 images working. Once theat's reliable I'll spin up an ubuntu and test it there | 03:33 |
| tonyb | ianw: It isn't really any different to the amd64 vs x86_64 scenario IIUC | 03:36 |
| ianw | i guess. much less cross-building of that though :) | 04:00 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: Support for Cloud Images on ppc64le for rhel7 and centos7 https://review.openstack.org/456055 | 04:08 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: [doc] Add some notes about PowerPC ARchitecture names https://review.openstack.org/477359 | 04:08 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: As far as block-device layout is concerned ppc64le == ppc64el https://review.openstack.org/477360 | 04:08 |
| tonyb | ianw: ^^ means the building cloud images for Centos and RHEL work for me. I need to double check that it still works for Ubuntu | 04:10 |
| tonyb | ianw: I'll tackle {centos,rhel}-minimal next | 04:11 |
| tonyb | ianw: Happy to take whatever re-works required | 04:11 |
| ianw | cool ... will check in a bit | 04:16 |
| prometheanfire | tonyb: starting work? :D | 05:50 |
| tonyb | prometheanfire: ? | 05:50 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: Support for Cloud Images on ppc64le for rhel7 and centos7 https://review.openstack.org/456055 | 05:58 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: [doc] Add some notes about PowerPC Architecture names https://review.openstack.org/477359 | 05:58 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: As far as block-device layout is concerned ppc64le == ppc64el https://review.openstack.org/477360 | 05:58 |
| ianw | heh, just looking | 05:59 |
| tonyb | It seems I messed up my vim config and didn't have expanttab set so I failed pep8 :( | 06:00 |
| tonyb | ... and then I messed up the rebase ... not my best day :( | 06:09 |
| *** andreas-f has quit IRC | 06:41 | |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: Support for Cloud Images on ppc64le for rhel7 and centos7 https://review.openstack.org/456055 | 07:09 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: [doc] Add some notes about PowerPC Architecture names https://review.openstack.org/477359 | 07:09 |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: As far as block-device layout is concerned ppc64le == ppc64el https://review.openstack.org/477360 | 07:09 |
| *** andreas-f has joined #openstack-dib | 07:30 | |
| openstackgerrit | Tuan Luong-Anh proposed openstack/diskimage-builder master: Replace new location of isolinux.bin for Ubuntu 16.04 https://review.openstack.org/477437 | 07:54 |
| openstackgerrit | Tuan Luong-Anh proposed openstack/diskimage-builder master: Replace new location of isolinux.bin for Ubuntu 16.04 https://review.openstack.org/477437 | 07:56 |
| *** pmannidi_2 has joined #openstack-dib | 07:56 | |
| *** pmannidi has quit IRC | 07:56 | |
| openstackgerrit | Tuan Luong-Anh proposed openstack/diskimage-builder master: Replace new location of isolinux.bin for Ubuntu 16.04 https://review.openstack.org/477437 | 07:57 |
| *** pmannidi_2 is now known as pmannidi | 07:59 | |
| *** andreas-f has quit IRC | 08:30 | |
| *** pmannidi has quit IRC | 08:40 | |
| *** pmannidi has joined #openstack-dib | 08:42 | |
| *** pmannidi has quit IRC | 08:47 | |
| *** andreas-f has joined #openstack-dib | 12:07 | |
| *** farosas has joined #openstack-dib | 12:10 | |
| *** rfolco has joined #openstack-dib | 12:34 | |
| *** isaacb has joined #openstack-dib | 13:04 | |
| *** andreas-f has quit IRC | 13:07 | |
| *** isaacb has quit IRC | 15:57 | |
| *** isaacb has joined #openstack-dib | 16:15 | |
| *** farosas_ has joined #openstack-dib | 16:26 | |
| *** farosas has quit IRC | 16:29 | |
| *** andreas-f has joined #openstack-dib | 16:49 | |
| *** isaacb has quit IRC | 17:04 | |
| openstackgerrit | Mark Hamzy proposed openstack/diskimage-builder master: Support CentOS7 in 01-override-yum-arch https://review.openstack.org/477616 | 17:46 |
| *** andreas-f has quit IRC | 17:48 | |
| *** andreas-f has joined #openstack-dib | 18:39 | |
| *** ChanServ has quit IRC | 19:30 | |
| *** ChanServ has joined #openstack-dib | 19:35 | |
| *** card.freenode.net sets mode: +o ChanServ | 19:35 | |
| *** andreas-f has quit IRC | 19:51 | |
| *** andreas-f has joined #openstack-dib | 20:51 | |
| *** andreas-f has quit IRC | 21:51 | |
| openstackgerrit | Tony Breeds proposed openstack/diskimage-builder master: dib-lint: Ignore editor temp files for linting run https://review.openstack.org/477711 | 23:44 |
| ianw | tonyb: why don't we drop most of that, and just do a "git check-ignore" and skip if it should be | 23:57 |
| tonyb | ianw: Ummm sure, I didn't knwo that was a thing | 23:58 |
| ianw | no, me either, i just was thinking "we have this in .gitignore surely" :) | 23:58 |
| tonyb | Yeah it is | 23:59 |
| tonyb | I'll respin after my meeting | 23:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!