*** ivve has quit IRC | 00:16 | |
openstackgerrit | Colleen Murphy proposed openstack/diskimage-builder master: Fix login.defs config for tumbleweed https://review.opendev.org/696989 | 00:21 |
---|---|---|
openstackgerrit | Colleen Murphy proposed openstack/diskimage-builder master: Allow zypper repos to be overrideable https://review.opendev.org/696990 | 00:31 |
*** yolanda has quit IRC | 01:25 | |
*** jangutter has joined #openstack-dib | 03:59 | |
*** jangutter has quit IRC | 04:03 | |
*** jangutter has joined #openstack-dib | 04:46 | |
*** jangutter has quit IRC | 04:51 | |
*** jangutter has joined #openstack-dib | 06:47 | |
*** jangutter has quit IRC | 06:52 | |
*** brault has joined #openstack-dib | 07:05 | |
*** yolanda has joined #openstack-dib | 07:22 | |
openstackgerrit | Michele Baldessari proposed openstack/diskimage-builder master: Make sure DIB_DEBUG_TRACE has a default value https://review.opendev.org/697008 | 07:26 |
*** jangutter has joined #openstack-dib | 08:04 | |
*** ivve has joined #openstack-dib | 08:51 | |
*** dasp has quit IRC | 09:42 | |
*** dasp has joined #openstack-dib | 09:43 | |
*** dasp has quit IRC | 13:12 | |
*** dasp has joined #openstack-dib | 13:12 | |
openstackgerrit | Harald Jensås proposed openstack/diskimage-builder master: Install ndisc6 package in element script https://review.opendev.org/696532 | 13:59 |
*** ivve has quit IRC | 16:15 | |
openstackgerrit | Carlos Goncalves proposed openstack/diskimage-builder master: Fix regex for mirror URL substitution https://review.opendev.org/697147 | 18:13 |
*** ivve has joined #openstack-dib | 19:28 | |
openstackgerrit | Carlos Goncalves proposed openstack/diskimage-builder master: Install rng-tools in Red Hat family distro images https://review.opendev.org/697183 | 20:44 |
openstackgerrit | Merged openstack/diskimage-builder master: Make sure DIB_DEBUG_TRACE has a default value https://review.opendev.org/697008 | 21:37 |
openstackgerrit | Merged openstack/diskimage-builder master: Fix login.defs config for tumbleweed https://review.opendev.org/696989 | 21:56 |
openstackgerrit | Merged openstack/diskimage-builder master: Break retry loop on success in dhcp-all-interfaces https://review.opendev.org/696873 | 21:56 |
openstackgerrit | Merged openstack/diskimage-builder master: Fix regex for mirror URL substitution https://review.opendev.org/697147 | 21:56 |
openstackgerrit | Merged openstack/diskimage-builder master: Install ndisc6 package in element script https://review.opendev.org/696532 | 22:41 |
johnsom | ianw I have a DIB element bug I have a question for you about if you are round and have a minute. | 23:13 |
ianw | johnsom: sure | 23:13 |
johnsom | This lovely file in the svc-map element: https://opendev.org/openstack/diskimage-builder/src/branch/master/diskimage_builder/elements/svc-map/extra-data.d/10-merge-svc-map-files | 23:13 |
johnsom | 1. runs the rando-python | 23:13 |
johnsom | 2. Uses yaml, but doesn't require it installed. | 23:14 |
johnsom | Since this is extra-data.d it's outside the chroot. | 23:14 |
johnsom | There is a package-install there, but it's phase is pre-install which is probably correct, but doesn't help this extra-data.d script. | 23:15 |
johnsom | I think I just need to switch this over to dib-python and make it install PyYaml somehow, but outside the chroot I'm not sure the best practice there. | 23:16 |
johnsom | This is the failed job log for reference: https://storage.bhs1.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_8cd/695291/12/check/octavia-v2-act-stdby-dsvm-scenario-two-node/8cdbc7d/controller/logs/dib-build/amphora-x64-haproxy.qcow2_log.txt.gz | 23:17 |
ianw | johnsom: ok ... it looks particularly unloved. from a grep it's only referenced from ironic-agent, which is also deprecated -- is it really necessary to fix as a first thought? | 23:17 |
johnsom | ianw https://github.com/openstack/octavia/blob/master/elements/amphora-agent/svc-map | 23:18 |
ianw | hrm, ok so octavia calls it too | 23:18 |
johnsom | Sadly RHEL named it different | 23:18 |
ianw | ok, so firstly in terms of yaml, that should be in requirements? | 23:19 |
ianw | https://opendev.org/openstack/diskimage-builder/src/branch/master/requirements.txt#L9 | 23:19 |
johnsom | Hmm, yes... Good point. | 23:20 |
ianw | i don't know if we have a solution to the random python thing | 23:22 |
johnsom | dib-python is only valid inside the chroot? | 23:22 |
ianw | dib-run-parts will just run diskimage_builder/elements/svc-map/extra-data.d/10-merge-svc-map-files | 23:22 |
ianw | yeah, dib-python is installed into /usr/local/bin in the chroot, but not outside it | 23:23 |
johnsom | Ah, so it won't really invoke that bang path then? | 23:23 |
ianw | i think the problem is it *does*. dib-run-parts just sees it as an executable file and runs it | 23:23 |
ianw | we have DIB_PYTHON_EXEC, which we use to run scripts that dib library functions are calling | 23:24 |
ianw | but dib-run-parts doesn't have any logic to say "oh, that looks like a python file, i should run that with DIB_PYTHON_EXEC" | 23:25 |
ianw | i'm not sure how to tackle that exactly ... | 23:26 |
ianw | it might be time to bite the bullet and go python3 only | 23:26 |
johnsom | Yeah, sadly DIB is branchless so that could get interesting. | 23:27 |
ianw | that would mean dropping trusty and centos7 as build hosts | 23:27 |
ianw | yeah, i mean we could freeze a python2 branch (and take important backports) but ... it's not like we have a ton of people who will maintain that | 23:29 |
johnsom | Right. | 23:29 |
johnsom | We can probably get "creative" and figure out another way.... | 23:29 |
johnsom | How would you feel if I updated dib-run-parts to look at the first line of the file and check for python, then run it with the right version based on DIB_PYTHON_EXEC or such? | 23:38 |
ianw | johnsom: it feels terrible, but i'm not having any better ideas. we did bring dib-run-parts inside dib (it used to come from a generic util) so such changes at least are isolated | 23:40 |
ianw | it does use "/usr/bin/env python"; so if we somehow had a link to the right python, and just ran all scripts with that prepended to PATH maybe? | 23:42 |
johnsom | Yeah, it's not ideal for me either. The only other option would be to install our own wrapper interpreter early and change the script to use it in the bang path | 23:42 |
ianw | i wonder, if disk-image-create made a temp directory and symlinked "python" to DIB_PTYHON_EXEC, and then exported PATH=/tmp/dir:${PATH} *maybe* that would work, for everything using "env python"? | 23:43 |
ianw | (also, assuming all scripts at 2+3 safe) | 23:43 |
johnsom | It should actually. | 23:44 |
johnsom | env should walk that path down, so yeah, that would work | 23:44 |
johnsom | A bit of "magic" but would get us through this whole python->python3->python headache | 23:45 |
*** ivve has quit IRC | 23:47 | |
ianw | ++ | 23:47 |
johnsom | Do you want to push something up? | 23:47 |
* johnsom thinks "famous silly question" | 23:49 | |
ianw | johnsom: umm, i'm just looking at some other stuff right now, but could do later if it's not urgent. otherwise i'm happy to review :) | 23:50 |
johnsom | Ok, I will try to take a pass at it, but feel free to review/revise | 23:51 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!