| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-01-05 | |||
| 10:03:43 | sean-k-mooney | stephenfin: o/ happy new year | |
| 10:03:51 | stephenfin | sean-k-mooney: o/ ditto | |
| 10:04:05 | sean-k-mooney | see my comment on https://review.openstack.org/#/c/530961/2 | |
| 10:04:35 | stephenfin | sean-k-mooney: Hahaha | |
| 10:04:41 | stephenfin | This is what I'm currently reading | |
| 10:04:46 | stephenfin | https://docs.python.org/2/library/subprocess.html#subprocess.Popen.communicate | |
| 10:05:04 | stephenfin | I was trying to capture stderr/stdout to see if that was indeed the issue (I suspected it was) | |
| 10:06:53 | stephenfin | sean-k-mooney: I'm thinking we just configure that to 'python -m privsep.[path_to_module]' | |
| 10:07:33 | sean-k-mooney | stephenfin: the test rodoflo is adding will need at least cap_net_admin to run so it will need to launch privesp with sudo to get that | |
| 10:09:19 | stephenfin | Right, I meant 'sudo python -m' as I thought we'd keep the virtualenv-configured Python, but we obviously won't actually | |
| 10:09:22 | sean-k-mooney | stephenfin: i think we can alter the config fuction here https://review.openstack.org/#/c/484386/30/os_vif/tests/functional/base.py | |
| 10:10:40 | sean-k-mooney | stephenfin: if we addin the a override for the oslo privsep helper_command form sudo privsep-helper to sudo -EH privsep-helper its shoudl resolve the issue | |
| 10:11:38 | sean-k-mooney | technicall i theink just the -E is need to keep the virtual env intact but i always use -EH just to be safe | |
| 10:12:08 | sean-k-mooney | my spelling is on point as always... | |
| 10:21:21 | stephenfin | sean-k-mooney: Turns out PATH isn't passed through via '-E'. Something about secure_path. Who knew! | |
| 10:32:16 | sean-k-mooney | stephenfin: really ok... im pulling down the patch in a clean vm and ill see if i can reporduce the gate issue | |
| 10:35:24 | sean-k-mooney | yep same issue | |
| 10:36:16 | sean-k-mooney | and the privsep-helper is in .tox/functional/bin/privsep-helper as i taught but not in /bin | |
| 10:38:05 | stephenfin | sean-k-mooney: This is as far as I've got http://paste.openstack.org/show/638995/ | |
| 10:40:14 | stephenfin | sean-k-mooney: But that's yielding this, ostensibly because we're using absolute paths to a package that is only installed in the virtualenv http://paste.openstack.org/show/638997/ | |
| 10:40:29 | sean-k-mooney | that worked for me | |
| 10:40:39 | stephenfin | It did? | |
| 10:40:45 | stephenfin | Failing for me on CentOS :( | |
| 10:40:48 | sean-k-mooney | well it passed all the tests | |
| 10:41:40 | sean-k-mooney | http://paste.openstack.org/show/638998/ | |
| 10:42:45 | stephenfin | How weird | |
| 10:43:07 | openstackgerrit | sahid proposed openstack/nova master: libvirt: disconnect volume from host during detach https://review.openstack.org/515008 | |
| 10:43:08 | stephenfin | You've other modifications though, to say that you have logging | |
| 10:43:32 | sean-k-mooney | no i just applied what you gave | |
| 10:43:52 | sean-k-mooney | http://paste.openstack.org/show/638999/ | |
| 10:44:16 | sean-k-mooney | that was applied directly on top of your zuul patch | |
| 10:44:20 | sean-k-mooney | 15264957c475a6ecc1648a9ca2f04c9beb95f98d | |
| 10:45:10 | sean-k-mooney | i can spin up a centos vm and test it there again if that helps | |
| 10:45:35 | stephenfin | If you wouldn't mind | |
| 10:45:42 | stephenfin | I'll push this now for review/rework | |
| 10:45:48 | stephenfin | See what the gate thinks too | |
| 10:47:06 | openstackgerrit | Stephen Finucane proposed openstack/os-vif master: zuul: Enable functional tests in gate https://review.openstack.org/530961 | |
| 10:47:07 | openstackgerrit | Stephen Finucane proposed openstack/os-vif master: Configure privsep binary https://review.openstack.org/531358 | |
| 10:47:10 | sean-k-mooney | no prob it will only take like 5min to get the env setup i mainly the proxies...# | |
| 10:57:35 | openstackgerrit | Matthew Booth proposed openstack/nova master: Don't generate fake disk_info in swap_volume https://review.openstack.org/530787 | |
| 10:57:35 | openstackgerrit | Matthew Booth proposed openstack/nova master: Remove redundant swap_volume tests https://review.openstack.org/531179 | |
| 10:57:36 | openstackgerrit | Matthew Booth proposed openstack/nova master: Local disk serial numbers for the libvirt driver https://review.openstack.org/529380 | |
| 11:04:37 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Refactor encryptor attach and detach calls https://review.openstack.org/460243 | |
| 11:04:38 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Attach and detach encryptors during swap_volume https://review.openstack.org/531233 | |
| 11:05:50 | sean-k-mooney | stephenfin: v2 fails v3 gives me the No module named internal.command.ip.test_impl_pyroute2 issue but only on centos. on ubunutu v3 works | |
| 11:06:45 | stephenfin | Try this on Ubuntu -> sudo -E su -c 'echo $PATH' | |
| 11:07:43 | stephenfin | CentOS has 'Defaults secure_path = /sbin:/bin:/usr/sbin:/usr/bin'. I imagine Ubuntu might not | |
| 11:07:44 | sean-k-mooney | /usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games vs /sbin:/bin:/usr/sbin:/usr/bin on centos | |
| 11:08:25 | stephenfin | Hmm, still doesn't pass through the virtualenv directory. I'd expected it would | |
| 11:08:40 | sean-k-mooney | well i was not in a venves ill try that instead | |
| 11:09:05 | stephenfin | Oh, yeah :D source .tox/functional/bin/activate first :) | |
| 11:09:42 | sean-k-mooney | i think " tox -e venv -- sudo -E su -c 'echo $PATH'" will also work | |
| 11:11:46 | sean-k-mooney | stephenfin: not much is jumping out at me http://paste.openstack.org/show/639010/ | |
| 11:16:31 | stephenfin | sean-k-mooney: Is that any help to us though? | |
| 11:23:13 | sean-k-mooney | i dont think its related to the system path. | |
| 11:34:37 | sean-k-mooney | stephenfin: i think this has someithing to do with the python path. specifically i do not think the privsep deamon spawnded via sudo -E on cetos is using the python path that with the virtual_env | |
| 11:35:13 | stephenfin | Yeah, I've tried setting 'PYTHONPATH' but haven't had any luck so far | |
| 11:35:23 | stephenfin | I wonder if it's using a different Python interpreter too? | |
| 11:35:50 | sean-k-mooney | slightly 2.7.12 on ubuntu and 2.7.5 on centos | |
| 11:36:21 | stephenfin | No, I meant the one in the virtualenv vs the system one | |
| 11:36:47 | sean-k-mooney | oh you can force that but no not by default | |
| 11:37:07 | stephenfin | I'm not actually sure how virtualenv configures PYTHONPATH. I was assuming it was done by way of the environment variable but apparently not | |
| 11:37:24 | sean-k-mooney | if i activeate teh virutal enve and run python i can run "from os_vif.internal.command.ip import impl_pyroute2" but it dose not work if i do sudo -E python | |
| 11:38:17 | stephenfin | Yeah, I tried 'sudo PYTHONPATH=.tox/functional/lib/python2.7/site-packages' too, but to no avail | |
| 11:38:20 | sean-k-mooney | hum that is the same on ubuntho though | |
| 11:43:13 | sean-k-mooney | stephenfin: well python -c "import sys; print('\n'.join(sys.path))" is pretty clear. with sudo -E teh venv disapears | |
| 11:46:08 | stephenfin | cdent: Finished https://review.openstack.org/#/c/513526/. One potential follow-up in there | |
| 11:48:11 | cdent | stephenfin: roger that. I think that block being where it is is a holdover from an earlier version where the logic was somewhat different and a rebase required shifting the logic a bit | |
| 11:48:46 | cdent | actually... | |
| 11:50:29 | cdent | no, that statement's right | |
| 11:54:24 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Attach and detach encryptors during swap_volume https://review.openstack.org/531233 | |
| 12:35:24 | sean-k-mooney | efried: o/ | |
| 12:39:07 | sean-k-mooney | efried: is https://review.openstack.org/#/c/531260/ the top patch on the nested resouce provides series? | |
| 13:40:18 | mhenkel | hi All | |
| 13:41:00 | mhenkel | I am trying to use the haproxy template similar to this one: https://review.openstack.org/#/c/474107/8/puppet/services/keystone.yaml | |
| 13:42:07 | mhenkel | however, when I define server_names: "%{hiera('contrail_config_node_names')}" haproxy puppet fails with: | |
| 13:42:30 | mhenkel | character '[' is not permitted in server name '[\"overcloud-contrailcontroller-0.internalapi.localdomain\"]'." | |
| 13:42:30 | mhenkel | "Error: /Stage[main]/Haproxy/Haproxy::Instance[haproxy]/Haproxy::Config[haproxy]/Concat[/etc/haproxy/haproxy.cfg]/File[/etc/haproxy/haproxy.cfg]/content: change from {md5}1f337186b0e1ba5ee82760cb437fb810 to {md5}d12e9a54e85e6bb63391d61cf9eb1da6 failed: Execution of '/usr/sbin/haproxy -f /etc/haproxy/haproxy.cfg20180105-12-13t72r6 -c' returned 1: [ALERT] 004/024209 (275) : parsing [/etc/haproxy/haproxy.cfg20180105-12-13t72r6:53] : | |
| 13:46:39 | mhenkel | any idea why %{hiera('contrail_config_node_names')} returns [\"overcloud-contrailcontroller-0.internalapi.localdomain\"] instead of overcloud-contrailcontroller-0.internalapi.localdomain ? | |
| 13:48:20 | stephenfin | mhenkel: Sounds like something for #tripleo or #puppet | |
| 13:53:12 | artom | mdbooth, I did have an idea of "meta-reviews" a while ago | |
| 13:53:39 | artom | The (very legitimate) response was "make it happen" | |
| 13:53:50 | artom | To which my also very legitimate reaction was "ugh" | |
| 13:53:54 | artom | And nothing came of it :/ | |
| 13:54:17 | mdbooth | artom: My concern's more that I suspect people have review lists like 'open reviews with +1 from Zuul and no -1s' | |
| 13:54:43 | mdbooth | And my patch would now fall foul of that for a completely dumb reason | |
| 13:55:07 | mdbooth | An alternate idea would be the ability to mark a -1 as 'I looked at that, and have no intention of addressing it'. | |
| 13:55:30 | mdbooth | Then people might filter on unaddressed -1s instead. | |
| 13:56:05 | mhenkel | stephenfin: sorry, wrong channel | |
| 13:57:53 | mdbooth | artom: Meta reviews would be good, though. | |
| 13:58:00 | artom | mdbooth, also, I suspect just showing who the -1 is from would help that | |
| 13:58:05 | mdbooth | Although, who would do them? | |
| 14:00:07 | stephenfin | mdbooth, artom: If you're using the dashboards from gerrit-dash-creator, they shouldn't matter. The only -1s used to filter stuff there are core ones https://github.com/openstack/gerrit-dash-creator/tree/master/dashboards | |
| 14:00:14 | stephenfin | Just as an aside | |
| 14:00:27 | mdbooth | stephenfin: Ah, that's awesome | |
| 14:00:59 | mdbooth | Well, somewhat elitist, but better that than exclude based on drive-bys | |
| 14:01:02 | mdbooth | ;) | |
| 14:01:25 | stephenfin | Elitist, but the best we've got ;) | |
| 14:03:58 | openstackgerrit | Lajos Katona proposed openstack/nova master: Deduplicate aggregate notification samples https://review.openstack.org/531162 | |
| 14:06:30 | hrw | speaking of reviews... anyone with free time to take a look at two simple aarch64 related patches? https://review.openstack.org/#/c/530965/ takes care of cpu_mode and https://review.openstack.org/#/c/489951/ makes aarch64 bootable out-of-the-box | |