| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-11-05 | |||
| 21:59:55 | efried | well, I didn't change 5.8 to 5.6 in my env | |
| 21:59:59 | sean-k-mooney | well it basicaly is right now | |
| 21:59:59 | efried | swhy I'm asking. | |
| 22:00:22 | efried | and if I reboot I have to restack, right? | |
| 22:00:48 | sean-k-mooney | there is another thing you coudl try first | |
| 22:00:53 | sean-k-mooney | but i did it the reboot way | |
| 22:01:22 | sean-k-mooney | you can add security_driver="none" | |
| 22:01:40 | sean-k-mooney | to /etc/libvirt/qemu.conf and restart libvirt-bin | |
| 22:01:55 | sean-k-mooney | but if you rerun stack.sh it will overwrite it | |
| 22:02:07 | sean-k-mooney | ill fix this up and automated in the plugin | |
| 22:02:12 | sean-k-mooney | its ment to do this for you | |
| 22:02:43 | sean-k-mooney | but as i said it has not been updated for ubuntu 18.04 | |
| 22:04:13 | efried | sean-k-mooney: and tbc, is this the same issue as | |
| 22:04:14 | efried | Connection to libvirt failed: authentication unavailable: no polkit agent available to authenticate action 'org.libvirt.unix.manage': libvirtError: authentication unavailable: no polkit agent available to authenticate action 'org.libvirt.unix.manage' | |
| 22:04:14 | efried | ? | |
| 22:04:30 | sean-k-mooney | yes | |
| 22:04:32 | efried | k | |
| 22:05:15 | sean-k-mooney | polkit is used to interact with selinux via dbus on fedora/centos/rhel | |
| 22:05:24 | sean-k-mooney | is apparently the libvirt default | |
| 22:05:43 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Update keypairs in saving an instance object https://review.opendev.org/683043 | |
| 22:05:44 | sean-k-mooney | im just not passing the correct flags when building it to have it auto install the apparmor files | |
| 22:06:06 | sean-k-mooney | i should prably just ask danpb how to do it correctly | |
| 22:11:18 | melwitt | mriedem: do you remember any func test change from the past made to avoid "{"forbidden": {"code": 403, "message": "Maximum number of ports exceeded"}}" | |
| 22:11:55 | mriedem | are you backporting something to queens or newton or something and hitting that? | |
| 22:11:57 | melwitt | I'm backporting stuff and hitting this because I'm presumably missing some underlying func test improvement | |
| 22:11:58 | melwitt | yes | |
| 22:12:24 | melwitt | I thought if anyone might remember something related to that, it would be you | |
| 22:12:39 | mriedem | not specifically, but i'd look for port quota stuff on the NeutronFixture | |
| 22:12:54 | melwitt | thanks, any direction helps a lot :) | |
| 22:13:11 | mriedem | https://review.opendev.org/#/q/I1dbccc2be6ba79bf267edac9208c80e187e6256a ? | |
| 22:13:36 | mriedem | https://review.opendev.org/#/c/587412/4/nova/tests/fixtures.py@1347 is the thing i'd think you want | |
| 22:14:23 | melwitt | looks promising | |
| 22:25:08 | sean-k-mooney | efried: ok i think i have fixed the issue locally ill work on a patch. but basically i need to tweak the /etc/libvirt/libvirtd.conf sligtly and nova will be happy | |
| 22:25:31 | sean-k-mooney | efried: http://paste.openstack.org/show/785827/ | |
| 22:27:02 | openstackgerrit | Merged openstack/nova master: Follow up to I3e28c0163dc14dacf847c5a69730ba2e29650370 https://review.opendev.org/692856 | |
| 22:27:04 | efried | sean-k-mooney: okay. I tried doing that conf change and now it's doing this | |
| 22:27:04 | efried | Connection to libvirt failed: Failed to connect socket to '/var/run/libvirt/libvirt-sock': No such file or directory: libvirtError: Failed to connect socket to '/var/run/libvirt/libvirt-sock': No such file or directory | |
| 22:28:12 | sean-k-mooney | where is that error? nova? virsh? | |
| 22:28:47 | efried | nova compute startup (journalctl) | |
| 22:29:15 | sean-k-mooney | ok what do you get if you do "sudo systemctl status libvirt-bin" | |
| 22:30:34 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Improve metadata server performance with large security groups https://review.opendev.org/656084 | |
| 22:30:37 | mriedem | this is a simple performance improvement fix, been sitting a long time waiting for the originaly author to touch it up ^ | |
| 22:30:46 | mriedem | and when i went to do it, i realized there was a much simpler fix | |
| 22:31:30 | efried | d'oh, sorry, already started unstacking. | |
| 22:31:56 | sean-k-mooney | no worries | |
| 22:32:07 | efried | sean-k-mooney: I'm adding those extra libvirtd.conf lines (from within the plugin) and will restack. | |
| 22:32:32 | sean-k-mooney | by the way i have code in the pluging to fix the app armor stuff | |
| 22:32:34 | sean-k-mooney | sudo apparmor_parser -R /etc/apparmor.d/usr.sbin.libvirtd || /bin/true | |
| 22:32:39 | sean-k-mooney | im guessing that is failing | |
| 22:32:48 | sean-k-mooney | because i thinkg the package name chagned | |
| 22:32:55 | sean-k-mooney | so im fining that too | |
| 22:33:00 | efried | are we not set -e here? | |
| 22:33:27 | efried | oh, || true would ignore :( | |
| 22:33:34 | sean-k-mooney | yes | |
| 22:34:06 | sean-k-mooney | the /bin/ture was to ignore it on subsequent runs but its really just a hack | |
| 22:35:03 | sean-k-mooney | install_build_deps for libvirt-bin used to pull in the apparmor utils but i apparently does not anymore so ill install them direcly | |
| 22:35:16 | sean-k-mooney | im guessing you dont have apparmor_parser right | |
| 22:36:34 | sean-k-mooney | i should not really depend on the transitive depencies like that | |
| 22:37:58 | efried | sean-k-mooney: I have an apparmor_parser in /sbin | |
| 22:38:44 | sean-k-mooney | hum ok it should have remvoed the proifle then but ill document the other way in anycase | |
| 22:39:41 | efried | the file it's referring to doesn't exist fwiw | |
| 22:41:07 | melwitt | mriedem: you were right, that solved the 403. now it's on to 409 "Multiple possible networks found, use a Network ID to be more specific". I'm sifting through NeutronFixture changes | |
| 22:41:54 | efried | sean-k-mooney: with those extra lines, now stack fails when libvirt-bin won't start. | |
| 22:41:55 | sean-k-mooney | efried: there is still a profile in /etc/apparmor.d/libvirt/ in my case | |
| 22:42:32 | efried | that line is skipping | |
| 22:42:32 | efried | +/opt/stack/devstack-plugin-libvirt-qemu/devstack/libs/libvirt:install_libvirt_qemu_src:210 sudo apparmor_parser -R /etc/apparmor.d/usr.sbin.libvirtd | |
| 22:42:32 | efried | File /etc/apparmor.d/usr.sbin.libvirtd not found, skipping... | |
| 22:43:05 | efried | the libvirt-bin service is failing thusly: | |
| 22:43:05 | efried | Nov 05 16:40:14 nucle systemd[1]: libvirtd.service: Service hold-off time over, scheduling restart. | |
| 22:43:05 | efried | Nov 05 16:40:14 nucle systemd[1]: libvirtd.service: Scheduled restart job, restart counter is at 5. | |
| 22:43:05 | efried | Nov 05 16:40:14 nucle systemd[1]: Stopped Virtualization daemon. | |
| 22:43:05 | efried | Nov 05 16:40:14 nucle systemd[1]: libvirtd.service: Start request repeated too quickly. | |
| 22:43:05 | efried | Nov 05 16:40:14 nucle systemd[1]: libvirtd.service: Failed with result 'exit-code'. | |
| 22:43:06 | efried | Nov 05 16:40:14 nucle systemd[1]: Failed to start Virtualization daemon. | |
| 22:43:08 | sean-k-mooney | ya it has obviously moved | |
| 22:43:45 | efried | I see these | |
| 22:43:45 | efried | /etc/apparmor.d/cache/usr.sbin.libvirtd | |
| 22:43:45 | efried | /opt/stack/libvirt/src/security/apparmor/usr.sbin.libvirtd | |
| 22:43:46 | sean-k-mooney | do you have anything in the journal | |
| 22:44:18 | efried | Assuming I would find it with journalctl -u libvirt-bin, no | |
| 22:44:35 | sean-k-mooney | it would be libvirtd-bin | |
| 22:44:43 | sean-k-mooney | but i dont think that is where its logging | |
| 22:44:48 | efried | either way | |
| 22:44:49 | efried | no | |
| 22:45:03 | sean-k-mooney | ok give me a sec to push this up and test it | |
| 22:45:14 | efried | crap, I f'ed up the libvirtd.conf hack, sec. | |
| 22:47:50 | melwitt | mriedem: ok, looks like I just needed more of that change and now things are working. gonna try backporting the whole thing | |
| 22:48:57 | mriedem | melwitt: which target branch? is this upstream? | |
| 22:49:06 | mriedem | because i don't think we want to backport that whole thing... | |
| 22:50:09 | melwitt | mriedem: this is downstream queens. really? ok | |
| 22:50:19 | mriedem | well, it's big | |
| 22:50:26 | mriedem | note you already have an abandoned backport of that upstream on stable/queens | |
| 22:50:38 | mriedem | https://review.opendev.org/#/c/599764/ | |
| 22:51:05 | melwitt | cause it looks like I need the bits removing the validate_networks stub, add the shared network stuff, and add stuff from this change too https://review.opendev.org/#/c/585385/15/nova/tests/fixtures.py | |
| 22:51:18 | melwitt | lol welp | |
| 22:51:25 | mriedem | you can't just do fake_network.set_stub_network_methods again? | |
| 22:51:42 | melwitt | maybe I can. I have zero memory of doing that before | |
| 22:51:45 | melwitt | I'll try it | |
| 22:52:05 | melwitt | that would be way easier | |
| 22:54:09 | sean-k-mooney | efried: im testing https://review.opendev.org/#/c/693094/1 now | |