| *** jamielennox is now known as jamielennox|away | 00:38 | |
| *** pmannidi has quit IRC | 01:44 | |
| *** pmannidi has joined #openstack-dib | 01:49 | |
| *** jamielennox|away is now known as jamielennox | 03:22 | |
| *** aparnav has joined #openstack-dib | 03:58 | |
| ianw | yolanda: i had a look at the dracut one. not totally opposed ... i wonder though if we're better served with a config file of some sort, rather that more or more elements | 07:07 |
|---|---|---|
| ianw | the existing dracut elements look like they are full of cruft, and i can't see they're used? | 07:07 |
| yolanda | hi ianw, so my dracut element has a different purpose than the existing ones. The current ones are for generating initramfs images | 07:23 |
| yolanda | the one i created is to regenerate dracut on the existing ones | 07:23 |
| yolanda | i had a conversation with greghaynes and we discussed several options, seemed that using different elements was simpler. Because it was not only needing to add the list of modules in the dracut command, but to add the depending packages | 07:24 |
| ianw | yolanda: yeah ... i noticed they had module install stuff. was just a thought really, maybe a yaml type config like the block stuff would work | 07:30 |
| yolanda | we need a way to specify the module itself, and the package that brings with him | 07:34 |
| yolanda | a yaml with module:[packages] list can make sense | 07:34 |
| *** zhugaoxiao has quit IRC | 07:57 | |
| *** zhugaoxiao has joined #openstack-dib | 07:58 | |
| *** brault has joined #openstack-dib | 08:01 | |
| ianw | i guess it's just if this is going to grow into many more elements all basically the same | 08:02 |
| yolanda | in my case i just needed the lvm and crypt, but i guess people can need some others, if we offer the possibility to enable modules on boot. Now this needs to be done post-deploy, but is useful to have it on the imaage, to avoid extra reboots | 08:18 |
| *** hashar has joined #openstack-dib | 08:31 | |
| *** andreas-f has quit IRC | 08:42 | |
| *** pmannidi has quit IRC | 08:49 | |
| *** andreas-f has joined #openstack-dib | 08:54 | |
| *** andreas-f has quit IRC | 08:55 | |
| *** openstackgerrit has quit IRC | 09:02 | |
| *** owalsh_ is now known as owalsh | 09:33 | |
| *** openstackgerrit has joined #openstack-dib | 10:14 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 10:14 |
| *** hashar is now known as hasharLunch | 10:44 | |
| *** aparnav has quit IRC | 12:06 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 12:52 |
| *** hasharLunch is now known as hashar | 14:05 | |
| openstackgerrit | Corey O'Brien proposed openstack/diskimage-builder master: Don't unmount missing paths https://review.openstack.org/446621 | 14:58 |
| *** jamielennox has quit IRC | 15:17 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 15:50 |
| *** hirewj has joined #openstack-dib | 15:51 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Always execute autorelabel https://review.openstack.org/447076 | 15:53 |
| hirewj | hi! i would like to build image for ironic with dib and after a lot of tries to build it with grub component (or better to say package, since this one fails when installing inside OS itself), i don't get out anything other then error that grub cannot be installed - it doesn't depend on which distro i try ubuntu, fedora, ... i'm building image with commands suggested in official guide and everything i get error that grub or depe | 15:54 |
| hirewj | like described here | 16:00 |
| hirewj | https://docs.openstack.org/developer/openstack-ansible-os_ironic/configure-ironic.html | 16:00 |
| greghaynes | ianw: yolanda I agree that a ton of tiny elements isnt great, although I think all we should provide is the default setup (which does maybe lvm and a couple others) and additional dracut modules we can just say need to live out of tree? | 16:02 |
| greghaynes | since its fine if users end up making a bunch of elements imo | 16:03 |
| greghaynes | I'm not super in to the idea though, if theres something else that will allow users to ask for additional modules with additional package dependencies im all for it | 16:04 |
| yolanda | greghaynes, i refactored that to use a yaml, take a look and see if that fits. it's not too complicated | 16:04 |
| yolanda | just a yaml with a list of modules and packages | 16:04 |
| greghaynes | ok | 16:05 |
| hirewj | greghaynes: any idea why ubuntu image with grub can't be build? ^ | 16:09 |
| hirewj | *built | 16:10 |
| *** jamielennox has joined #openstack-dib | 16:19 | |
| *** jamielennox is now known as jamielennox|away | 16:21 | |
| hirewj | dpkg: error processing package grub-efi-amd64 (--configure): | 16:42 |
| hirewj | subprocess installed post-installation script returned error exit status 1 | 16:42 |
| hirewj | dpkg: dependency problems prevent configuration of grub-efi-amd64-signed: | 16:43 |
| hirewj | Package grub-efi-amd64 is not configured yet. | 16:43 |
| hirewj | No DKMS packages installed: not changing Secure Boot validation state. | 16:43 |
| hirewj | Errors were encountered while processing: grub-efi-amd64 grub-efi-amd64-signed E: Sub-process /usr/bin/dpkg returned an error code (1) | 16:43 |
| *** hashar is now known as hasharAway | 16:59 | |
| hirewj | can somebody help | 17:01 |
| hirewj | does anyone know how to create ramdisk and user image for ironic with disk-image-create so that it works? | 17:05 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Apply setfiles on all mountpoints https://review.openstack.org/447076 | 17:22 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Add dracut-regenerate elements https://review.openstack.org/447047 | 17:29 |
| openstackgerrit | Corey O'Brien proposed openstack/diskimage-builder master: Ignore missing path in unmount_dir https://review.openstack.org/446621 | 17:39 |
| hirewj | disk-image-create ubuntu baremetal localboot local-config dhcp-all-interfaces grub2 -o ubuntu-xenial | 18:01 |
| hirewj | Errors were encountered while processing: grub-efi-amd64 grub-efi-amd64-signed linux-image-4.4.0-66-generic linux-image-extra-4.4.0-66-generic linux-image-generic linux-image-generic-lts-xenial | 18:01 |
| hirewj | any idea what to do for successful image building? | 18:01 |
| hirewj | ianw: ^ | 18:01 |
| hirewj | yolanda: ^ | 18:01 |
| hirewj | http://i.imgur.com/t2Xigoh.png | 18:03 |
| *** jamielennox|away is now known as jamielennox | 18:24 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Apply setfiles on all mountpoints https://review.openstack.org/447076 | 18:39 |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Apply setfiles on all mountpoints https://review.openstack.org/447076 | 18:42 |
| *** andreas-f has joined #openstack-dib | 19:12 | |
| *** hasharAway is now known as hashar | 19:22 | |
| openstackgerrit | yolanda.robla proposed openstack/diskimage-builder master: Apply setfiles on all mountpoints https://review.openstack.org/447076 | 19:41 |
| *** hirewj has quit IRC | 19:45 | |
| *** pleia2_ has joined #openstack-dib | 20:18 | |
| *** andyjones2001 has joined #openstack-dib | 20:22 | |
| *** radez` has joined #openstack-dib | 20:25 | |
| *** andyjone1 has quit IRC | 20:29 | |
| *** pleia2 has quit IRC | 20:29 | |
| *** radez has quit IRC | 20:29 | |
| ianw | hirewj: hmm, really need some more logs before that | 20:35 |
| ianw | with -x | 20:36 |
| ianw | oh, they're gone | 20:36 |
| *** pmannidi has joined #openstack-dib | 20:43 | |
| *** owalsh has quit IRC | 20:50 | |
| *** owalsh has joined #openstack-dib | 20:50 | |
| *** hashar has quit IRC | 22:01 | |
| *** mmedvede has joined #openstack-dib | 22:48 | |
| mmedvede | ianw: andreas-f: continuing from #openstack-infra, about http://lists.openstack.org/pipermail/openstack-dev/2017-March/114263.html | 22:52 |
| mmedvede | I have setup a test job on ppc64, here is one of results http://dal05.objectstorage.softlayer.net/v1/AUTH_3d8e6ecb-f597-448c-8ec2-164e9f710dd6/pkvmci/diskimage-builder/86/444586/4/check-silent/dib-dsvm-functests-devstack-xenial/0449bca/ | 22:52 |
| mmedvede | if you ctrl-f 'the chosen partition is not a PReP partition', you'll see an error I was talking about in the email | 22:53 |
| mmedvede | ianw: andreas-f: so if you let me, I would like to enable reporting of our CI on diskimage-builder | 22:54 |
| ianw | mmedvede: ok, i'd be fine with that | 22:55 |
| ianw | are you setting this up to build ppc images though, or does that just happen magically because it's on that host? | 22:55 |
| ianw | ./run_functests.sh doesn't know anything about architectures | 22:55 |
| mmedvede | ianw: it just happens magically. Well, I guess there is logic in diskimage-builder to make it happen | 22:56 |
| mmedvede | e.g. it gets correct arch cloud image | 22:56 |
| ianw | hmm, i'm not sure how intentional it is :) | 22:56 |
| ianw | so debian-minimal passed? | 22:58 |
| mmedvede | ianw: debian-minimal/stable-build-succeeds passed, but it failed at debian-minimal/stable-vm | 22:59 |
| ianw | yeah, ok | 22:59 |
| ianw | mmedvede: what platforms are you interested in? | 23:00 |
| mmedvede | ianw: there are a bunch of other tests that I am not sure would pass after, but debian-minimal/stable-vm fails with the same error that I saw building Xenial image on Xenial | 23:00 |
| mmedvede | ianw: primarily, interested in ppc64 :) | 23:00 |
| ianw | sure, but i mean debian, ubuntu, fedora, centos, suse, gentoo ... all of them? | 23:01 |
| mmedvede | ianw: or are you asking, which OS platforms? Ubuntu / Debian. But I do not mind testing more | 23:01 |
| ianw | the only thing with the functional tests is they are not booting tests | 23:01 |
| mmedvede | centos/gentoo straight up refuses to run with "only supported on x86" message | 23:01 |
| ianw | the nodepool tests are, however | 23:02 |
| ianw | does ppc use grub? | 23:02 |
| ianw | (i haven't had linux+ppc since a white iBook that I used to run linux on, so I'm out of date :) | 23:03 |
| mmedvede | ianw: judging from the code in dib element that sets grub up, yes | 23:04 |
| mmedvede | good question, maybe it installs wrong grub (ppc uses grub-ieee1275 package) | 23:05 |
| mmedvede | ianw: so can I consider your ok an official permission to report on diskimage-builder? | 23:06 |
| ianw | mmedvede: well yes I don't mind at least getting a comment with logs ... but honestly we're going to have to do some work here to get the results to be useful I think | 23:07 |
| ianw | at least making sure we're running suitable func tests, not all of them | 23:08 |
| ianw | otherwise the job will just never pass | 23:08 |
| ianw | so if it's just constantly failing, people are trained to ignore it | 23:08 |
| ianw | can we just have it comment and not vote at this point? | 23:08 |
| mmedvede | ianw: definitely. I already removed tests that I know would fail (centos/gentoo) | 23:08 |
| mmedvede | ianw: when you mentioned nodepool boot test, did you mean e.g. gate-dsvm-nodepool-debian-src-nv? | 23:09 |
| mmedvede | yeah, non-voting is fine | 23:09 |
| ianw | yes, those two nodepool tests actually build an image, and use nodepool to upload it into devstack where it gets booted and registered | 23:10 |
| ianw | it might be interesting, but obviously this is also a tacit test of the entire devstack + nodepool infrastructure too :) | 23:11 |
| mmedvede | nice. I'll consider adding that job too. functional passing seems to be prerequisite for nodepool one passing though | 23:11 |
| ianw | yes ... if we can't build the image we can't boot it :) | 23:13 |
| ianw | mmedvede: and we are sure this worked before right? i think you said it did in your mail. was it completely booting? | 23:14 |
| mmedvede | ianw: yes, it was working before and building, I did put snippet from good build in the bug https://bugs.launchpad.net/diskimage-builder/+bug/1674402 | 23:16 |
| openstack | Launchpad bug 1674402 in diskimage-builder "bootloader element fails to install grub on ppc64 platform" [Undecided,New] | 23:16 |
| mmedvede | ianw: and it was booting - we use the images built with dib in our CI | 23:16 |
| ianw | mmedvede: can you attach the full log of the prior working run? | 23:17 |
| ianw | i guess clearly the difference is in the trying to install in loop0 v loop0p1 ... presumably p1 is a /boot partition | 23:18 |
| ianw | just wondering if we actually create it | 23:18 |
| mmedvede | ianw: that is the first thing I've noticed, but there could be something else going on, e.g. why would grub try using loop0p1 while it is told explicitly to use loop0p | 23:19 |
| mmedvede | s/grub/install-grub/ | 23:19 |
| ianw | surely that --modules=part_msdos is at least unnecessary, maybe even wrong ... | 23:20 |
| ianw | we're running grub-install on IMAGE_BLOCK_DEVICE_WITHOUT_PART | 23:22 |
| ianw | ok, elements/vm/block-device.d/10-partition does have all this ppc setup | 23:28 |
| ianw | in 1.28 | 23:29 |
| mmedvede | sanitizing good log a bit, would upload shortly | 23:29 |
| mmedvede | ianw: I'll push a test patch removing part_msdos option, so would also test if reporting is working ok | 23:30 |
| ianw | mmedvede: yeah, that's not it | 23:30 |
| mmedvede | no need to try? | 23:30 |
| ianw | i don't think so, that bit hasn't changed | 23:30 |
| ianw | i do think we've regressed here in diskimage_builder/block_device/level1/partitioning.py | 23:31 |
| *** jamielennox is now known as jamielennox|away | 23:36 | |
| *** jamielennox|away is now known as jamielennox | 23:40 | |
| ianw | mmedvede: ok, commented in the bug, i think i'm starting to piece together this. | 23:42 |
| ianw | mmedvede: are you ok on 1.28.0 for the time being? | 23:42 |
| mmedvede | ianw: thanks. Yeah, I can build on 1.28.0 for now. Although it is not a smooth experience any more :) | 23:43 |
| mmedvede | ianw: trying to wrap my head around partitioning.py patch https://review.openstack.org/#/c/375261/. It is one big patch | 23:44 |
| ianw | mmedvede: yes, it was a big change, that's why we bumped to 2.0.0 :) but it lays the foundation for us to do fancy things like lvm and crypt partitions etc etc | 23:46 |
| mmedvede | Thanks for the help. There should be reports from IBM PowerKVM CI on diskimage-builder starting now | 23:50 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
