| *** jamielennox is now known as jamielennox|away | 00:50 | |
| *** jamielennox|away is now known as jamielennox | 01:21 | |
| *** sudipto has joined #openstack-dib | 04:29 | |
| *** sudipto_ has joined #openstack-dib | 04:29 | |
| *** sudipto has quit IRC | 05:14 | |
| *** sudipto_ has quit IRC | 05:14 | |
| *** sudipto has joined #openstack-dib | 05:58 | |
| *** sudipto_ has joined #openstack-dib | 05:58 | |
| *** sudipto has quit IRC | 07:36 | |
| *** sudipto_ has quit IRC | 07:36 | |
| *** strigazi_AFK is now known as strigazi | 08:28 | |
| *** sudipto_ has joined #openstack-dib | 08:31 | |
| *** sudipto has joined #openstack-dib | 08:31 | |
| *** hashar has joined #openstack-dib | 08:36 | |
| *** pmannidi has quit IRC | 08:49 | |
| *** prometheanfire has quit IRC | 09:42 | |
| *** prometheanfire has joined #openstack-dib | 09:43 | |
| *** owalsh-afk has joined #openstack-dib | 09:47 | |
| *** owalsh-afk is now known as owalsh | 09:49 | |
| *** prometheanfire has quit IRC | 09:50 | |
| *** prometheanfire has joined #openstack-dib | 09:51 | |
| *** owalsh is now known as owalsh-brb | 10:24 | |
| *** owalsh-brb is now known as owalsh | 10:37 | |
| *** sudipto has quit IRC | 11:22 | |
| *** sudipto_ has quit IRC | 11:22 | |
| *** owalsh is now known as owalsh-lunch | 12:25 | |
| *** tommylikehu has quit IRC | 12:39 | |
| *** tommylikehu has joined #openstack-dib | 12:40 | |
| *** yolanda has quit IRC | 13:18 | |
| *** yolanda has joined #openstack-dib | 13:18 | |
| *** sudipto_ has joined #openstack-dib | 13:37 | |
| *** sudipto has joined #openstack-dib | 13:37 | |
| *** owalsh-lunch is now known as owalsh | 13:46 | |
| *** aparnav has joined #openstack-dib | 14:18 | |
| openstackgerrit | Bob Fournier proposed openstack/diskimage-builder: Handle failure of carrier check in dhcp-all-interfaces.sh https://review.openstack.org/419527 | 15:38 |
|---|---|---|
| openstackgerrit | Merged openstack/diskimage-builder: Run dhcp-interface@.service after network.target https://review.openstack.org/416664 | 15:47 |
| *** sudipto has quit IRC | 16:05 | |
| *** sudipto_ has quit IRC | 16:05 | |
| *** sudipto_ has joined #openstack-dib | 16:05 | |
| *** sudipto has joined #openstack-dib | 16:05 | |
| *** aparnav has quit IRC | 16:08 | |
| *** aparnav has joined #openstack-dib | 16:38 | |
| *** aparna has joined #openstack-dib | 16:40 | |
| *** aparnav has quit IRC | 16:43 | |
| *** aparna has quit IRC | 16:49 | |
| *** hashar has quit IRC | 16:51 | |
| openstackgerrit | Paul Belanger proposed openstack/diskimage-builder: Bump fedora-minimal release to 25 https://review.openstack.org/419559 | 16:56 |
| *** sudswas__ has joined #openstack-dib | 17:07 | |
| *** sudipto_ has quit IRC | 17:07 | |
| *** sudipto has quit IRC | 17:08 | |
| *** sudipto has joined #openstack-dib | 17:08 | |
| *** pabelanger has quit IRC | 17:41 | |
| *** pabelanger has joined #openstack-dib | 17:41 | |
| *** owalsh has quit IRC | 17:51 | |
| *** openstackgerrit has quit IRC | 18:33 | |
| pabelanger | ianw: mind reviewing: https://review.openstack.org/#/c/414634/ removes some hardcoded settings from ubuntu-minimal. | 19:03 |
| pabelanger | ianw: so we can use AFS mirrors for nodepool dsvm job: https://review.openstack.org/#/c/414273/ | 19:03 |
| *** sudipto has quit IRC | 19:05 | |
| *** sudswas__ has quit IRC | 19:05 | |
| *** pabelanger has quit IRC | 19:39 | |
| *** pabelanger has joined #openstack-dib | 20:01 | |
| *** openstackgerrit has joined #openstack-dib | 21:18 | |
| openstackgerrit | Andreas Florath proposed openstack/diskimage-builder: Easy diskimage-builder development environment setup https://review.openstack.org/419655 | 21:18 |
| openstackgerrit | Paul Belanger proposed openstack/diskimage-builder: Add spec for debootstrap-minimal element https://review.openstack.org/419679 | 22:41 |
| pabelanger | ianw: re: 413868, this is what I want http://paste.openstack.org/show/594782/ | 22:52 |
| pabelanger | but know no other way of doing it | 22:52 |
| pabelanger | if there is a better way, I am happy to try that | 22:52 |
| ianw | pabelanger: why not just write BAR (no $!) | 22:53 |
| pabelanger | ianw: because I didn't think of that | 22:54 |
| pabelanger | ianw: what syntax do you want to see for ${!DIB_*} ? | 22:54 |
| ianw | no that bit's fine | 22:54 |
| ianw | it's $!BAR that doesn't seem to do anything | 22:54 |
| ianw | (and the other vars following) | 22:55 |
| ianw | or i mean the other vars | 22:55 |
| ianw | ${!DIB_*} just means "all vars starting with DIB_ right? | 22:55 |
| openstackgerrit | Paul Belanger proposed openstack/diskimage-builder: Start with an empty environment for chroot https://review.openstack.org/413868 | 22:56 |
| pabelanger | ianw: yes | 22:56 |
| ianw | right, names matching prefix | 22:56 |
| pabelanger | ${!DIB_@} is valid too | 22:56 |
| ianw | while ${!DIB_VAR} is indirect expansion | 22:56 |
| ianw | but $!DIB_VAR is actually "$!""DIB_VAR" | 22:57 |
| ianw | even though the man page says the { is optional, at least for that case it's not | 22:57 |
| pabelanger | ianw: not sure I follow on ${!DIB_VAR} | 22:57 |
| ianw | ${!DIB_VAR} means "use the value of DIB_VAR as the name of the variable" | 22:58 |
| pabelanger | Oh, yes. | 22:58 |
| pabelanger | I read it as _VAR was some bash magic string | 22:58 |
| ianw | yeah. the manual seems to suggest $!DIB_VAR would be the same thing | 22:59 |
| ianw | but reality seems to not agree | 22:59 |
| pabelanger | ianw: let me look up the xtrace trick you mentioned | 23:00 |
| *** owalsh has joined #openstack-dib | 23:00 | |
| ianw | pabelanger: see -> function kill_chroot_processes () { | 23:01 |
| ianw | just silences the spam from the loop over all the vars | 23:01 |
| openstackgerrit | Paul Belanger proposed openstack/diskimage-builder: Start with an empty environment for chroot https://review.openstack.org/413868 | 23:02 |
| pabelanger | k, lets see what tests say | 23:03 |
| ianw | cool, yeah it's easy to spot in the logs | 23:04 |
| ianw | getting that balance between useful and spammy has been an ongoing challenge | 23:05 |
| ianw | i think the logs are way better than they used to be in terms of signal:noise | 23:05 |
| pabelanger | ianw: okay, on to fstrim. So, the issue is, I cannot run diskimage-builder inside docker today, because fstrim fails. | 23:05 |
| pabelanger | everything I google says, fstrim won't work in docker | 23:05 |
| pabelanger | so, that was the reason for the flag | 23:06 |
| ianw | right, but if you're trying to make a .qcow2, we want the output trimmed? | 23:06 |
| ianw | otherwise making an image under docker means you just get an unnecessarily large output image? | 23:06 |
| openstackgerrit | Merged openstack/diskimage-builder: Fix Gentoo builds on Ubuntu 16.04 Xenial hosts https://review.openstack.org/413395 | 23:07 |
| ianw | and also, the only way you can get it right under docker is to use that flag, which presumably you don't release until it fails, which isn't a great interface | 23:07 |
| ianw | is this because you're not really root under docker? | 23:08 |
| pabelanger | ianw: not sure about trimmed actually, I haven't compared the sizes. But I guess that is the question, how do you trim a mount, inside a container | 23:11 |
| pabelanger | ianw: no, this was a privileged container (root docker user) | 23:11 |
| pabelanger | and using sudo inside container | 23:12 |
| ianw | so, i'm not sure why it wouldn't make sense to trim a loopback type device even from within a docker container | 23:14 |
| pabelanger | give me a second, getting error | 23:15 |
| ianw | fstrim: /: FITRIM ioctl failed: Operation not supported | 23:15 |
| pabelanger | ya, that looks right | 23:15 |
| ianw | but that should be running inside the chroot | 23:15 |
| ianw | which should be within a mounted loopback image at that point? | 23:16 |
| pabelanger | http://serverfault.com/questions/820169/should-i-be-running-fstrim-in-a-container | 23:16 |
| pabelanger | was that I found | 23:16 |
| pabelanger | give me a sec | 23:17 |
| ianw | right ... it makes no sense to run fstrim inside a container, where say your / isn't a block device, but really some random directory on a block device | 23:17 |
| ianw | but -- inception | 23:17 |
| ianw | we're in a container, but then chrooted into a mounted block device | 23:18 |
| ianw | so "/" should, in theory, be a device that could be trimmed | 23:18 |
| pabelanger | $ sudo docker run --privileged=true -t -i nodepool-builder /bin/bash | 23:19 |
| pabelanger | root@8ab8d0952caa:/# disk-image-create --no-tmpfs -n -o ubuntu-minimal ubuntu-minimal | 23:19 |
| pabelanger | is the command I am currently running | 23:19 |
| pabelanger | I wonder if I also need to add --cap-add | 23:20 |
| pabelanger | will try that next | 23:20 |
| ianw | yes, that or apparmor type permissions would get in the way probably | 23:20 |
| ianw | it would be good to see the mounts just before fstrim | 23:21 |
| pabelanger | doh, forgot -t | 23:23 |
| pabelanger | Oh, actually | 23:23 |
| pabelanger | it failed | 23:23 |
| pabelanger | okay, let me try add-cap | 23:23 |
| ianw | yeah, maybe SYS_ADMIN needs to be added? | 23:27 |
| ianw | yeah, i think --cap-add=SYS_ADMIN | 23:28 |
| pabelanger | same failure | 23:29 |
| pabelanger | --cap-add=ALL | 23:29 |
| pabelanger | let me get some debug | 23:29 |
| ianw | hmm, operation not supported i guess, rather than perms denied | 23:30 |
| openstackgerrit | Andreas Florath proposed openstack/diskimage-builder: Docker matrix build https://review.openstack.org/414347 | 23:45 |
| pabelanger | /dev/loop2 on /tmp/dib_build.3K700JIq/mnt type ext4 (rw,relatime,data=ordered) | 23:48 |
| pabelanger | that is the loop device that fstrim fails on | 23:48 |
| ianw | does it work outside docker? | 23:49 |
| pabelanger | yes | 23:49 |
| ianw | can you strace the fstrim inside docker? | 23:50 |
| *** pmannidi has joined #openstack-dib | 23:50 | |
| pabelanger | let me manually do it | 23:51 |
| *** owalsh has quit IRC | 23:51 | |
| ianw | was that above from "mount" inside docker, or outside it? | 23:52 |
| pabelanger | ianw: say again | 23:52 |
| ianw | "/tmp/dib_build.3K700JIq/mnt" ... that's docker's "/tmp" right? | 23:54 |
| pabelanger | yes | 23:54 |
| ianw | what does "mount" outside docker show? | 23:54 |
| pabelanger | 1 sec | 23:55 |
| pabelanger | http://paste.openstack.org/show/594789/ | 23:57 |
| pabelanger | ianw: ^ | 23:57 |
| ianw | so it has no idea about the mounts inside the container? | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!