ianw | hmm, fdisk does an fsync on the file and a sync() | 00:00 |
---|---|---|
ianw | sync(3) = 0 | 00:00 |
ianw | close(3) = 0 | 00:01 |
ianw | write(1, "Syncing disks.\n", 15) = 15 | 00:01 |
ianw | sync() = 0 | 00:01 |
ianw | i think we'd do well to replicate that | 00:01 |
*** andreas-f has quit IRC | 00:32 | |
*** openstackgerrit has joined #openstack-dib | 00:37 | |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Sync after writing partition table https://review.openstack.org/475203 | 00:37 |
ianw | rm_work / johnsom: ^ with the benefit of hindsight, this does seem like a more likely cause ... | 00:37 |
johnsom | The other thought I had was to add a small retry loop here. | 00:40 |
ianw | yeah, that seems a bit non-deterministic like a sleep() ... i'm more convinced about the sync() but still odd afaik we've never seen it once in any of our functional or nodepool tests, nor the infra builders hit it | 00:46 |
ianw | oh dear ... debian has released | 00:48 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Install systemd earlier for strech https://review.openstack.org/475206 | 01:41 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Install systemd earlier for Stretch https://review.openstack.org/475206 | 01:42 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Install systemd earlier for Stretch https://review.openstack.org/475206 | 03:27 |
ianw | ok, next problem for today, nb03 is full | 04:48 |
ianw | $ mount | wc -l | 04:49 |
ianw | 1390 | 04:49 |
ianw | oh dear ... it's a mess | 04:49 |
ianw | i'm rebooting it ... it's too far gone with 50 loop devices | 04:52 |
ianw | ok, it's building again with free space | 05:02 |
rm_work | :) | 05:05 |
rm_work | thanks for continuing to look at this issue | 05:05 |
rm_work | that seems sane, just need the gate issue to resolve and then recheck i guess ;P | 05:07 |
ianw | rm_work: the gate should be fixed. i tried using a depends-on but it didn't seem to work with the octavia job? | 05:15 |
ianw | i couldn't see the sync call being made | 05:16 |
ianw | the global gate should be working i mean | 05:16 |
ianw | dib is going to require the debian fix https://review.openstack.org/#/c/475206/ | 05:16 |
ianw | not sure if anyone has an opinion on that | 05:16 |
*** andreas-f has joined #openstack-dib | 05:18 | |
*** andreas-f has left #openstack-dib | 06:00 | |
*** andreas-f has joined #openstack-dib | 06:00 | |
ianw | andreas-f: hmm, i'm not sure it's that different. it pulls in sysv-init via package-installs anyway | 06:19 |
rm_work | yeah i meant for https://review.openstack.org/#/c/475203/ to merge, it's going to need the gate issue resolved and then a recheck ;P | 06:22 |
ianw | rm_work: yeah, the nodepool builder hit that problem too | 06:42 |
*** aparnav has joined #openstack-dib | 07:09 | |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Sync after writing partition table https://review.openstack.org/475203 | 07:13 |
openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Sync after writing partition table https://review.openstack.org/475203 | 07:14 |
ianw | yolanda: ^ | 07:14 |
yolanda | ianw, just one change there? | 07:15 |
ianw | oh, sorry, i just rebased that on 475206 which hasn't changed | 07:16 |
ianw | just so they can go into the gate in order | 07:16 |
yolanda | ok looking | 07:16 |
yolanda | how is 475203 related to 475206? | 07:23 |
ianw | it's not, just that it needs to go in to un break the gate | 07:31 |
*** openstackgerrit has quit IRC | 07:48 | |
*** andreas-f has quit IRC | 09:31 | |
*** owalsh_ is now known as owalsh | 10:06 | |
*** hwoarang has quit IRC | 10:28 | |
*** hwoarang has joined #openstack-dib | 10:29 | |
*** amarao has joined #openstack-dib | 11:26 | |
amarao | Hello. I'm trying to migrate from dib 1.x to dib 2.6.0 and I found that after repackaging it as a deb-file (with same scripts) diskimage-create no longer works: disk-image-create | 11:28 |
amarao | bash: /usr/lib/python2.7/dist-packages/diskimage_builder/lib/disk-image-create: No such file or directory. Where this file should come from? | 11:28 |
ianw | amarao: the file layout has changed somewhat, deb packages would need to be updated | 11:49 |
amarao | Yes, and I'm trying to understand what need to be changed. | 11:50 |
ianw | however, it should all install correctly from setup.py with entry points, so really the package doesn't need to do much | 11:50 |
ianw | amarao: see some of the comments in http://lists.openstack.org/pipermail/openstack-dev/2017-February/111833.html | 11:51 |
*** chhavi has joined #openstack-dib | 11:52 | |
amarao | ok, thanks, I'd try to sort out it. Virtualenv as 'must do' is not look promising though... | 11:53 |
ianw | amarao: it should actually be a lot easier. "disk-image-create" should be in lib/ relative to where "import diskimage_builder" is ... you can see we just exec it in http://git.openstack.org/cgit/openstack/diskimage-builder/tree/diskimage_builder/disk_image_create.py#n72 | 11:57 |
ianw | ergo, "setup.py install" just works | 11:57 |
ianw | i would reccommend a virtualenv approach, really, but some people don't ... infra install directly for example on nodepool builders | 11:58 |
amarao | We are very oldschool and software should be packaged, and packaging should be scripted in CI server. | 11:58 |
amarao | apt-get install python-diskimage-builder. | 11:58 |
*** rfolco has joined #openstack-dib | 11:58 | |
ianw | that's fine ... dh_python should "just work" ... nothing fancy going on | 11:59 |
amarao | Unfortunately it doesn't. That why I'm trying to figure out what happens. | 12:00 |
amarao | %: | 12:00 |
amarao | dh $@ --buildsystem=python_distutils --with python | 12:00 |
amarao | That's the main part. Plus few quirks to remove .eggs from source tree (after clean) and removal of *.pyc files. That's all. | 12:01 |
ianw | only thing i can think is that diskimage_builder/lib doesn't have an __init__.py | 12:03 |
ianw | have you got a package i can see? | 12:03 |
ianw | amarao: oh, i assume you're using something based on https://git.openstack.org/cgit/openstack/deb-python-diskimage-builder/tree/debian/rules?h=debian/newton | 12:14 |
*** farosas has joined #openstack-dib | 12:15 | |
ianw | i don't have time right now, but i can take a look later, i'd think using pybuild as a first step, it can also build 2&3 packages (python3 pretty well tested) and i wouldn't be surprised if distutils gets something wrong | 12:15 |
*** openstackgerrit has joined #openstack-dib | 13:22 | |
openstackgerrit | Merged openstack/diskimage-builder master: Install systemd earlier for Stretch https://review.openstack.org/475206 | 13:22 |
*** aparnav has quit IRC | 13:24 | |
*** chhavi has quit IRC | 16:36 | |
openstackgerrit | Merged openstack/diskimage-builder master: Sync after writing partition table https://review.openstack.org/475203 | 16:49 |
*** amarao has quit IRC | 17:22 | |
*** zigo has quit IRC | 18:05 | |
*** zigo has joined #openstack-dib | 18:09 | |
prometheanfire | and amarao is gone :| | 18:12 |
*** jamielennox has quit IRC | 18:19 | |
*** lennyb has quit IRC | 18:19 | |
*** fungi has quit IRC | 18:19 | |
*** jamielennox has joined #openstack-dib | 18:21 | |
*** fungi has joined #openstack-dib | 18:25 | |
*** lennyb has joined #openstack-dib | 18:27 | |
*** dirk has joined #openstack-dib | 20:33 | |
dirk | Hi | 20:33 |
openstackgerrit | Mikhail S Medvedev proposed openstack/diskimage-builder master: DNM, Revert "Sync after writing partition table" https://review.openstack.org/475524 | 20:34 |
*** farosas has quit IRC | 21:02 | |
ianw | mmedvede: is it the stretch release? | 21:06 |
ianw | rm_work / johnsom : ^^ things looking better with the sync calls in? | 21:06 |
johnsom | Too early to tell I think. We haven't been running a lot of jobs yet today, but some have gone in recently | 21:08 |
ianw | ok, fingers crossed | 21:08 |
ianw | because i'm pretty low on other ideas of what it is ATM :) | 21:09 |
mmedvede | ianw: not sure what is causing the failure I am seeing. Wasn't the latest patch. It looks like there is no /etc/modprobe.d dir now when the 15-cleanup-debootstrap is called | 21:14 |
mmedvede | it being the stretch release is a good guess | 21:15 |
*** rfolco has quit IRC | 21:21 | |
*** mgagne has quit IRC | 21:21 | |
*** mgagne has joined #openstack-dib | 21:22 | |
*** mmedvede has left #openstack-dib | 21:30 | |
*** mmedvede has joined #openstack-dib | 21:30 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!