| ianw | you can git log, and be happy about it :) | 00:00 |
|---|---|---|
| jamielennox | lol, if only | 00:00 |
| ianw | but pretty sure state gets updated when it goes somewhere | 00:00 |
| ianw | i've only submitted a few for netdev and dave miller usually updates it | 00:01 |
| jamielennox | i should probably not checkout kernel.git on this coffee shop wifi | 00:06 |
| jamielennox | it's not fast | 00:06 |
| ianw | i'll just sync, i doubt it's it's linus' tree yet | 00:07 |
| jamielennox | yea, i'm pretty sure it's not | 00:08 |
| jamielennox | and even if it is, that would mean i need to backport it all onto the xenail kernel that's actually running | 00:09 |
| ianw | i very surprised nobody noticed before may 1, 2017 ... | 00:09 |
| jamielennox | i guess no one uses -x often? | 00:09 |
| jamielennox | so i guess the solution for us is a dib fix that does | 00:10 |
| jamielennox | if [ `df -t overlay /tmp` ]; then du -a -c; else du -a -x -c; fi | 00:11 |
| ianw | one question is, do we really need the "-x" | 00:12 |
| ianw | or is it just there for saftey? it does seem everything should be unmounted | 00:12 |
| jamielennox | so there are things in dev in my test | 00:14 |
| jamielennox | but they're 0 size | 00:14 |
| ianw | presumably it's unmounted? | 00:15 |
| jamielennox | if it's showing files in /tmpXX/dev/YYYY i assume it's mounted | 00:17 |
| jamielennox | i have an du -a -c log, but it's like 2mb of file strucutre | 00:17 |
| jamielennox | not sure it's useful | 00:18 |
| jamielennox | but eg | 00:18 |
| jamielennox | 0/tmp/dib_build.uuziyOKj/built/dev/tty0 | 00:18 |
| jamielennox | 0/tmp/dib_build.uuziyOKj/built/dev/smpte0 | 00:18 |
| jamielennox | 0/tmp/dib_build.uuziyOKj/built/dev/loop0 | 00:18 |
| ianw | that's it? that seems like stray files, and /dev/ isn't mounted | 00:19 |
| jamielennox | no there's a lot | 00:19 |
| ianw | just a mount, then grep for dib_build.uuziyOKj would tell us | 00:19 |
| jamielennox | i can put a mount output in at that point as well | 00:19 |
| ianw | yeah ... sorry i don't know in detail what's going on there, so building from first principles as it were | 00:20 |
| ianw | i don't think dib will ever be accused of being over-commented ... | 00:20 |
| jamielennox | i can put a bash statement in there that'll hang and let me check what's happening, but i then basically rebuild the pod to clean it up | 00:21 |
| jamielennox | i guess i could try building it on a normal machine and testing the difference between -x and not | 00:23 |
| ianw | or pop in a gate test and see if the devstack jobs explode | 00:23 |
| jamielennox | but really all we care about here is the image size right, so if we removed -x it might grow a bit? | 00:23 |
| jamielennox | heh, so you test via gate as well :) | 00:24 |
| jamielennox | but yea, i don't think it will explode, it'd just be a test of how much image size changes | 00:24 |
| ianw | i think the problem would be if something is totally wrong, like it starts traversing the wrong place due to links etc | 00:25 |
| jamielennox | any idea how that would happen? it seems to be just trying to find the size of a directory dib build | 00:27 |
| jamielennox | ianw: this is the output of mount at the du point: http://paste.openstack.org/show/613453/ | 00:29 |
| ianw | so everything unmounted, as we thought ... i'm tending towards removing the "-x" as unnecessary | 00:32 |
| ianw | i haven't deep dived to figure out, has it been there forever? | 00:32 |
| jamielennox | i've traced it back to at least 2013 | 00:35 |
| ianw | https://review.openstack.org/#/c/25866/ is where it came in i think | 00:35 |
| jamielennox | yea, basically from the beginning | 00:36 |
| ianw | yeah, and it's been carried forward ever since | 00:36 |
| jamielennox | yep, that's the commit i'm looking at | 00:37 |
| jamielennox | gah, i wish kernel worked like other places, it would be nice if that commit had a bug associated | 00:38 |
| *** zhugaoxiao has quit IRC | 00:39 | |
| *** Administrator__ has quit IRC | 00:39 | |
| openstackgerrit | Jamie Lennox proposed openstack/diskimage-builder master: Don't use du -x to calculate image size https://review.openstack.org/476732 | 00:43 |
| ianw | see how it goes! | 00:48 |
| jamielennox | is the gate reporting correctly at the moment anyway? | 00:48 |
| jamielennox | i put up https://review.openstack.org/#/c/476302/ yesterday that should be a trivial fix and it's not passed | 00:48 |
| ianw | largely ... i'm just working on moving the download jobs out | 00:48 |
| ianw | there's a couple of known issues | 00:49 |
| ianw | 1) is that the mirrors can be really really slow after a "vos release", so jobs timeout | 00:49 |
| ianw | 2) is a known but unfixed issue with neutron/shade and floating ip's which has randomly hit us a couple of times | 00:49 |
| jamielennox | ok, i'll just keep rechecking that one till it passes, i don't know what i'm looking for in the 2mb console.log job output | 00:50 |
| ianw | scroll up about 200 lines, that's usually where the real error is :) | 00:51 |
| jamielennox | oh, right, mirrors | 00:52 |
| jamielennox | 2017-06-22 22:23:20.973362 | [centos7/build-succeeds] Error downloading packages: | 00:53 |
| jamielennox | 2017-06-22 22:23:20.973387 | [centos7/build-succeeds] sudo-1.8.6p7-23.el7_3.x86_64: [Errno 256] No more mirrors to try. | 00:53 |
| ianw | yep, i'm kicking those jobs out into a -nv job | 00:55 |
| ianw | the "-minimal" jobs should be using infra mirrors now | 00:55 |
| openstackgerrit | Ian Wienand proposed openstack/diskimage-builder master: Move image download tests to default skip https://review.openstack.org/476734 | 01:01 |
| *** rfolco has joined #openstack-dib | 01:20 | |
| prometheanfire | ianw: looks like I might be able to install with '--target /usr/lib64/python2.7/site-packages' | 01:29 |
| ianw | prometheanfire: is this just because gentoo has modified pip to not install globally? | 01:30 |
| prometheanfire | yep | 01:30 |
| prometheanfire | we've had many users break their system by doing so | 01:30 |
| prometheanfire | I've thought about properly packaging glean, but haven't yet | 01:31 |
| ianw | i thought gentoo was about the freedom to break your system horribly | 01:32 |
| ianw | </joking> | 01:32 |
| prometheanfire | so did I :P | 01:32 |
| *** rfolco has quit IRC | 01:41 | |
| *** rfolco has joined #openstack-dib | 01:44 | |
| *** rfolco has quit IRC | 02:00 | |
| *** aparnav has joined #openstack-dib | 03:26 | |
| *** chhavi has joined #openstack-dib | 04:15 | |
| *** andreas-f has joined #openstack-dib | 05:03 | |
| openstackgerrit | Merged openstack/diskimage-builder master: Move image download tests to default skip https://review.openstack.org/476734 | 05:14 |
| mordred | ianw, jamielennox: I'm still poking at the neutron/shade issue - it's driving me absolutely batty | 05:39 |
| mordred | ianw: sorry it's taken so long - had ansiblefest in london yesterday which has disrupted my ability to go OCD on the problem | 05:40 |
| ianw | mordred: that's ok :) it hasn't been a huge issue, especially compared to the other stuff like ubuntu breaking their sha256 hashes for images | 05:41 |
| mordred | ianw: oy | 05:43 |
| *** aparnav has quit IRC | 11:14 | |
| *** openstackgerrit has quit IRC | 11:33 | |
| *** farosas has joined #openstack-dib | 13:12 | |
| *** rfolco has joined #openstack-dib | 13:22 | |
| *** rfolco has quit IRC | 13:23 | |
| *** rfolco has joined #openstack-dib | 13:23 | |
| *** hashar_ has joined #openstack-dib | 15:12 | |
| *** chhavi has quit IRC | 15:30 | |
| *** chhavi has joined #openstack-dib | 15:41 | |
| *** hashar_ has quit IRC | 16:04 | |
| *** chhavi has quit IRC | 16:36 | |
| *** openstackgerrit has joined #openstack-dib | 17:08 | |
| openstackgerrit | Dirk Mueller proposed openstack/diskimage-builder master: Avoid hanging endlessly on unreachable cache urls https://review.openstack.org/476981 | 17:08 |
| *** andreas-f has quit IRC | 18:44 | |
| *** rfolco has quit IRC | 21:09 | |
| *** rfolco has joined #openstack-dib | 21:25 | |
| *** farosas has quit IRC | 21:34 | |
| *** andreas-f has joined #openstack-dib | 21:36 | |
| *** andreas-f has quit IRC | 22:36 | |
| *** rfolco has quit IRC | 22:39 | |
| *** rfolco has joined #openstack-dib | 23:10 | |
| *** rfolco has quit IRC | 23:10 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!