*** goldyfruit_ has quit IRC | 00:09 | |
johnsom | rm_work Ok, paid my penance. I have reviewed the whole AZ chain. | 00:18 |
---|---|---|
johnsom | The gerrit bot is broken BTW, so we won't see patches announced | 00:21 |
*** armax has quit IRC | 00:22 | |
*** spatel has joined #openstack-lbaas | 00:41 | |
*** sapd1 has quit IRC | 00:47 | |
*** abaindur has joined #openstack-lbaas | 01:02 | |
*** abaindur has quit IRC | 01:03 | |
*** openstackstatus has joined #openstack-lbaas | 01:03 | |
*** ChanServ sets mode: +v openstackstatus | 01:03 | |
*** abaindur has joined #openstack-lbaas | 01:03 | |
*** abaindur has quit IRC | 01:19 | |
*** abaindur has joined #openstack-lbaas | 01:48 | |
*** abaindur has quit IRC | 01:48 | |
*** abaindur has joined #openstack-lbaas | 01:49 | |
*** abaindur has quit IRC | 02:13 | |
*** spatel has quit IRC | 02:37 | |
*** ricolin has joined #openstack-lbaas | 02:50 | |
*** armax has joined #openstack-lbaas | 03:14 | |
*** Trevor_V has quit IRC | 03:18 | |
rm_work | Ah lol k | 04:13 |
*** ramishra has joined #openstack-lbaas | 05:49 | |
rm_work | now just need cgoncalves to get back to reviewing so we can get stuff moving :D | 06:05 |
rm_work | ugh still have that bug or whatever where on py37 the test_driver_agent functional tests take FOREEEVER | 06:07 |
rm_work | i forget exactly what caused it, but annoying | 06:08 |
*** openstackgerrit has joined #openstack-lbaas | 06:09 | |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Availability Zone admin API https://review.opendev.org/693765 | 06:09 |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Availability Zone admin API https://review.opendev.org/693765 | 06:10 |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Support creating a LB in a specified availability-zone https://review.opendev.org/693762 | 06:10 |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Support creating an LB in a specified availability-zone https://review.opendev.org/693762 | 06:38 |
rm_work | cgoncalves: https://review.opendev.org/#/c/694057/ :D | 06:40 |
rm_work | dayou: ^^ :P | 06:41 |
openstackgerrit | Merged openstack/octavia-lib master: Availability zone / metadata validation https://review.opendev.org/694057 | 06:59 |
rm_work | wooo thanks dayou :D | 07:13 |
dayou | rm_work, my pleasure :D | 07:13 |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Support creating an LB in a specified availability-zone https://review.opendev.org/693762 | 08:10 |
*** ccamposr__ has joined #openstack-lbaas | 08:17 | |
*** ccamposr has quit IRC | 08:20 | |
*** pcaruana has joined #openstack-lbaas | 08:21 | |
*** tesseract has joined #openstack-lbaas | 08:35 | |
*** rpittau|afk is now known as rpittau | 08:58 | |
*** gcheresh has joined #openstack-lbaas | 09:07 | |
*** ricolin has quit IRC | 09:43 | |
*** gcheresh has quit IRC | 09:47 | |
*** luksky has joined #openstack-lbaas | 11:34 | |
*** rcernin has quit IRC | 12:01 | |
*** goldyfruit_ has joined #openstack-lbaas | 13:20 | |
openstackgerrit | Merged openstack/octavia-dashboard master: Imported Translations from Zanata https://review.opendev.org/694477 | 13:38 |
*** TrevorV has joined #openstack-lbaas | 15:00 | |
openstackgerrit | Gregory Thiemonge proposed openstack/octavia-tempest-plugin master: Add UDP test scenario https://review.opendev.org/656515 | 15:08 |
*** armax has joined #openstack-lbaas | 15:42 | |
*** luksky has quit IRC | 16:04 | |
johnsom | gthiemonge https://review.opendev.org/#/q/status:open+project:openstack/octavia+branch:master+topic:udp_support | 16:07 |
rm_work | johnsom: https://review.opendev.org/#/c/693765/ re-review? ignore the post-failure and grenade being dumb | 16:13 |
rm_work | I think we could merge :D | 16:13 |
johnsom | rm_work That file comment was that you are missing the equivalent to https://github.com/openstack/octavia/blob/master/octavia/api/v2/controllers/provider.py#L70 | 16:15 |
gthiemonge | johnsom: thanks | 16:18 |
rm_work | ah would that go in the same controller? | 16:19 |
rm_work | in providers.py i mean | 16:19 |
johnsom | gthiemonge Sorry those patches fell off the radar and are still open. We were not sure if Zhao was going to finish them or not but at this point it is pretty clear that he is no longer working on them. | 16:19 |
johnsom | Yeah, I think so | 16:19 |
rm_work | k | 16:20 |
rm_work | yeah ok i see | 16:20 |
rm_work | otherwise good? | 16:20 |
johnsom | I think so. I did a pretty in depth review yesterday. I will scan your changes now but I don't expect any issues | 16:21 |
rm_work | K | 16:24 |
johnsom | I missed one extra indent, but otherwise, yeah, I would +2 it with the capabilities API in place. You fixed all of the other stuff I found. | 16:26 |
*** ricolin has joined #openstack-lbaas | 16:32 | |
colin- | is ComputeWaitTimeoutException in octavia-worker something i can override in config? | 16:46 |
colin- | seeing output related to it that i don't remember before stein just want to understand it better | 16:47 |
rm_work | johnsom: uhh is this wrong? https://github.com/openstack/octavia/blob/master/octavia/policies/provider_flavor.py#L25 | 16:48 |
johnsom | colin- The retries and timeout are CONF.controller_worker.amp_active_retries and CONF.controller_worker.amp_active_wait_sec | 16:48 |
rm_work | https://github.com/openstack/octavia/blob/master/octavia/tests/functional/api/v2/base.py#L83-L84 | 16:49 |
johnsom | colin- We added a new timeout that is connectivity to help clarify the problem and to help with failovers working better when both amps of a act/stdby pair failed at the same time. Maybe that is what is different? | 16:49 |
colin- | perhaps, will take a look thx | 16:50 |
colin- | that is plenty to go on | 16:50 |
johnsom | rm_work Yes, that is wrong. Lucky it is just for documentation and not actual policy enforcement.... Though the functional test would have caught that. | 16:51 |
rm_work | hmm | 16:51 |
rm_work | would have? | 16:51 |
rm_work | why didn't it? | 16:51 |
rm_work | do we not have functional tests? O_o | 16:51 |
johnsom | Because that line is only for docs, not functionality | 16:51 |
rm_work | ah k | 16:51 |
rm_work | i'll just fix it in this patch | 16:51 |
johnsom | Thanks. Yeah, that dict in the list is only used for the "sample file generation" which we don't really use, but an operator could run. | 16:52 |
johnsom | rm_work This is the test that covers the functionality: https://github.com/openstack/octavia/blob/master/octavia/tests/functional/api/v2/test_provider.py#L128 | 16:53 |
openstackgerrit | Merged openstack/octavia-lib master: Switch to Ussuri jobs https://review.opendev.org/690836 | 16:56 |
johnsom | haleyb Would you mind adding Ajay as a co-author on https://review.opendev.org/#/c/693132/ to resolve the https://review.opendev.org/#/c/692077/ issue. Technically his patch was first. | 16:56 |
*** rpittau is now known as rpittau|afk | 17:01 | |
TrevorV | Hey johnsom , when you get a few seconds, could you answer a quick question for me? In Triple-O, do you know if any of the different octavia service require being run on a compute node? | 17:14 |
johnsom | TrevorV No, we have to code that runs on the compute nodes directly. We of course spin up service VMs inside the compute nodes, but nothing goes on the node itself. | 17:17 |
johnsom | to->no | 17:17 |
TrevorV | thanks for confirming that, I thought as much from a conversation I had with rm_work a while ago. Could you humor me with an example or two of what service VMs you mean? | 17:19 |
johnsom | The amphora | 17:19 |
*** ricolin has quit IRC | 17:21 | |
TrevorV | johnsom alright, thanks. So amphora are created and run on the compute nodes, is that worded properly? | 17:21 |
johnsom | No | 17:21 |
johnsom | They are not run on the compute nodes, they are run inside the compute environment, i.e. inside nova. | 17:22 |
johnsom | There is zero Octavia code installed on compute nodes. | 17:22 |
TrevorV | Oh okay, so, after the amphora image is uploaded to glance, octavia just spawns nova servers with that image as the amphora? | 17:23 |
johnsom | Right | 17:23 |
TrevorV | which of course, lives on the compute nodes. Got it! thanks a bunch, that helps me immensely | 17:23 |
rm_work | yep | 17:23 |
johnsom | Pretty much your quote is "There is zero Octavia code installed on the compute nodes" | 17:23 |
rm_work | think of octavia as just another cloud-user | 17:23 |
rm_work | we can be installed off to the side | 17:23 |
johnsom | A really spiffy cloud user. lol | 17:24 |
rm_work | and we do need admin access, but otherwise we just use standard APIs | 17:24 |
rm_work | :D | 17:24 |
rm_work | yes | 17:24 |
johnsom | No, we do not need admin access. | 17:24 |
johnsom | That is not true. | 17:24 |
rm_work | we di | 17:24 |
rm_work | *we do | 17:24 |
rm_work | to create ports on user networks | 17:24 |
rm_work | and plug them to VMs in different projects | 17:24 |
rm_work | that requires admin | 17:24 |
johnsom | No, you can run Octavia under non-admin projects/users. And you should | 17:24 |
rm_work | errr | 17:24 |
rm_work | we definitely need admin to neutron | 17:24 |
johnsom | It would require RBAC rules, but not admin level | 17:25 |
rm_work | last i checked it was literally an admin check to do the stuff we do with cross-binding ports | 17:25 |
johnsom | Not even full "service" project level, but that is a good compromise | 17:25 |
johnsom | Nope. This is a common mis-conception. We do not need to run as full "admin" in the cloud. | 17:25 |
rm_work | hmm | 17:26 |
johnsom | Many people do it simply because it's "easy" | 17:26 |
johnsom | But it is definitely not the right answer | 17:26 |
rm_work | i actually wrote one of the neutron patches to make one of those things work... hold on | 17:27 |
johnsom | It has been done in previous roles/products I was a part of.... | 17:28 |
rm_work | ahhhh | 17:28 |
johnsom | Really at some point I should take a week and write that up as another chapter in our install guide, but.... time happens | 17:28 |
rm_work | it was for my L3 driver T_T | 17:28 |
rm_work | https://review.opendev.org/#/c/463888/ | 17:28 |
rm_work | literally "is_admin" check lol | 17:29 |
johnsom | I would argue that is implemented wrong..... | 17:29 |
johnsom | If fact, is_admin is deprecated | 17:29 |
rm_work | lol whelp | 17:29 |
rm_work | i mean that was 2.5 years ago | 17:29 |
rm_work | so was it deprecated THEN? :D | 17:29 |
johnsom | Well, it was still wrong then at least | 17:30 |
rm_work | lol | 17:30 |
rm_work | kevin benton helped me with it :D | 17:30 |
rm_work | and kevin and miguel +2'd heh | 17:31 |
rm_work | oh hey and brian reviewed too lol | 17:31 |
rm_work | johnsom calling out ALL the cores :D | 17:31 |
rm_work | and two PTLs | 17:31 |
johnsom | Yes, yes I am | 17:32 |
rm_work | I don't disagree with you I guess :D just funny no one else seemed to care | 17:32 |
johnsom | It's putting policy logic in a strange spot in the code and it probably should be the context_is_advsvc role that neutron uses | 17:34 |
rm_work | yeah that sounds like it'd be more right | 17:34 |
rm_work | i wonder if this line is even still there | 17:35 |
johnsom | It is | 17:35 |
*** tesseract has quit IRC | 17:35 | |
johnsom | But they still haven't met the policy-in-code goal either, so maybe it will get cleaned up when they finish that work. | 17:35 |
johnsom | Anyway, Octavia *should* not require full cloud ADMIN rights to run. | 17:36 |
johnsom | Aside from your hack, I don't think it does today | 17:36 |
johnsom | rm_work https://bugs.launchpad.net/neutron/+bug/1853637 | 17:41 |
openstack | Launchpad bug 1853637 in neutron "Assign floating IP to port owned by another tenant is not override-able with RBAC policy" [Undecided,New] | 17:41 |
johnsom | See something, say something. lol | 17:42 |
rm_work | lol | 17:48 |
rm_work | hmm i think i'm done with this patch | 17:49 |
openstackgerrit | Adam Harwell proposed openstack/octavia master: Availability Zone admin API https://review.opendev.org/693765 | 17:49 |
johnsom | Yeah, assuming the tests pass, that version looks good to me | 18:02 |
*** irclogbot_1 has quit IRC | 18:27 | |
*** irclogbot_3 has joined #openstack-lbaas | 18:30 | |
*** beisner has quit IRC | 18:35 | |
*** beisner has joined #openstack-lbaas | 18:36 | |
rm_work | need to update the tempest side | 18:43 |
rm_work | cgoncalves: https://review.opendev.org/#/c/691135/ | 18:43 |
rm_work | trying to cut down the number of outstanding patches I have waiting to merge | 18:43 |
rm_work | also i promised that feature to someone internally >_< | 18:44 |
rm_work | ok, updated tempest tests to include new tests for that API | 19:09 |
rm_work | vvv | 19:09 |
openstackgerrit | Adam Harwell proposed openstack/octavia-tempest-plugin master: [WIP] Test the Availability Zone [Profile] admin API https://review.opendev.org/695291 | 19:09 |
rm_work | i really really hate our grenade right now | 19:09 |
rm_work | i hate it so so much | 19:10 |
johnsom | Yeah, annoying. I mentioned some of the neutron issues I have seen to the neutron team. Not sure if the nova stuff is really neutron related or not | 19:14 |
rm_work | the last one i saw seemed to fail on keystone upgrade? :/ | 19:14 |
johnsom | Yeah, I think I have seen one of those too | 19:14 |
rm_work | same this time | 19:15 |
rm_work | ahhhh | 19:16 |
rm_work | 2019-11-22 19:00:47.992 | Found existing installation: PyYAML 3.12 | 19:16 |
rm_work | 2019-11-22 19:00:48.412 | ERROR: Cannot uninstall 'PyYAML'. It is a distutils installed project and thus we cannot accurately determine which files belong to it which would lead to only a partial uninstall. | 19:16 |
rm_work | whelp | 19:16 |
rm_work | another one of those | 19:16 |
johnsom | Hmm, I thought all of those were fixed a long time ago.... | 19:16 |
johnsom | We found one issue in the nova scripts for grenade yesterday. It is using the py2 OSC when devstack is py3, but that shouldn't impact us yet as we are still running that under py2 | 19:20 |
johnsom | My guess is it is fallout from this: https://review.opendev.org/#/c/649097/ | 19:30 |
johnsom | purely a guess though | 19:31 |
*** ccamposr has joined #openstack-lbaas | 19:32 | |
johnsom | Might be one of those packages keystone is installing for tests also raised their minimum | 19:33 |
johnsom | It is definitely installed in the OS however: ii python3-yaml 3.12-1build2 amd64 YAML parser and emitter for Python3 | 19:34 |
*** ccamposr__ has quit IRC | 19:34 | |
*** gcheresh has joined #openstack-lbaas | 19:48 | |
rm_work | eugh well ... i guess our gates are broken until this is resolved? joy | 19:59 |
rm_work | i won't be able to track this today | 19:59 |
openstackgerrit | Gregory Thiemonge proposed openstack/octavia-tempest-plugin master: Use VM as client in traffic operation scenario tests https://review.opendev.org/687562 | 20:06 |
*** abaindur has joined #openstack-lbaas | 20:23 | |
*** abaindur has quit IRC | 20:28 | |
*** luksky has joined #openstack-lbaas | 20:32 | |
johnsom | I can try the disable n-vnc advice infra gave. | 20:35 |
johnsom | Funny enough, I already have a fixgrenade folder. I guess it is a recurring problem | 20:36 |
johnsom | We certainly have zero need for that vnc stuff | 20:37 |
*** ccamposr__ has joined #openstack-lbaas | 20:38 | |
*** ccamposr has quit IRC | 20:40 | |
*** gcheresh has quit IRC | 20:44 | |
*** rcernin has joined #openstack-lbaas | 20:49 | |
openstackgerrit | Michael Johnson proposed openstack/octavia master: Disable nova console services in grenade job https://review.opendev.org/695763 | 20:52 |
*** abaindur has joined #openstack-lbaas | 21:11 | |
*** servagem has quit IRC | 21:26 | |
*** spatel has joined #openstack-lbaas | 21:38 | |
*** spatel has quit IRC | 21:39 | |
*** abaindur has quit IRC | 22:05 | |
*** spatel has joined #openstack-lbaas | 22:08 | |
*** spatel has quit IRC | 22:13 | |
*** spatel has joined #openstack-lbaas | 22:15 | |
*** rcernin has quit IRC | 22:24 | |
*** KeithMnemonic has quit IRC | 22:44 | |
*** spatel has quit IRC | 22:58 | |
*** luksky has quit IRC | 23:04 | |
*** TrevorV has quit IRC | 23:30 | |
openstackgerrit | Brian Haley proposed openstack/octavia-lib master: Stop testing python 2 https://review.opendev.org/693132 | 23:39 |
openstackgerrit | Michael Johnson proposed openstack/octavia master: Disable nova console services in grenade job https://review.opendev.org/695763 | 23:52 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!