| ralonsoh | hberaud[m], hello! before the meeting, can I ask you one question? | 14:38 |
|---|---|---|
| ralonsoh | this is about migrating the Neutron metadata proxy we current have | 14:38 |
| hberaud[m] | sure | 14:38 |
| ralonsoh | we are currently using a local implementation, neutron.api.wsgi.Server | 14:39 |
| ralonsoh | that uses ProcessLauncher | 14:39 |
| ralonsoh | will it be there after the eventlet removal of oslo.services? | 14:40 |
| hberaud[m] | Yes it will remain available. Only the wsgi module (and things like the backdoor) will be removed, the processlauncher is part of the service sub-module so we won't remove it | 14:42 |
| ralonsoh | hberaud[m], thanks! | 14:42 |
| hberaud[m] | ralonsoh: you are welcome | 14:42 |
| hberaud[m] | #startmeeting openstack-eventlet-removal | 15:00 |
| opendevmeet | Meeting started Tue Dec 3 15:00:01 2024 UTC and is due to finish in 60 minutes. The chair is hberaud[m]. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_eventlet_removal' | 15:00 |
| hberaud[m] | Ping list: amorin, JayF, gouthamr | 15:00 |
| amorin | hello! | 15:00 |
| gouthamr | o/ | 15:00 |
| hberaud[m] | #link https://etherpad.opendev.org/p/epoxy-eventlet-tracking | 15:00 |
| hberaud[m] | We are around line 68 | 15:00 |
| ralonsoh | hi | 15:01 |
| hberaud[m] | (of our etherpad) | 15:01 |
| JayF | o/ I am here highly distracted | 15:02 |
| hberaud[m] | I'll have to shorten a bit this meeting, because I chair an other meeting just after | 15:02 |
| hberaud[m] | so we have ~ 55 min | 15:02 |
| hberaud[m] | here we go | 15:03 |
| hberaud[m] | #topic Review task completion | 15:03 |
| hberaud[m] | keep track of the mistral failure on Eventlet requirements bump https://review.opendev.org/c/openstack/requirements/+/933257 (@amorin) | 15:03 |
| amorin | I am the first in that list, I did some check locally, I can reproduce the issue. Still I dont know how to fix | 15:03 |
| amorin | I havnt been able to spend enough time on it | 15:03 |
| hberaud[m] | failures are related to neutron? | 15:04 |
| amorin | I believe mistral is not the only one in that situation, so maybe if other members have a clue about the issue, that would help a lot :) | 15:04 |
| hberaud[m] | because I see that Jens replied to Rodolfo | 15:04 |
| ralonsoh | which ones? I don't see that in the CI | 15:04 |
| hberaud[m] | https://review.opendev.org/c/openstack/requirements/+/933257/comments/ecf51f53_21487174 | 15:04 |
| ralonsoh | this is something we are fixing right now | 15:05 |
| hberaud[m] | apparently it fail here https://zuul.opendev.org/t/openstack/build/4c991243369c47dd8d655be343d4faf8 | 15:05 |
| ralonsoh | we have some problem in the migration from eventlet to wsgi in Neutron | 15:05 |
| hberaud[m] | ack, thanks ralonsoh | 15:05 |
| ralonsoh | but we are more than aware | 15:05 |
| hberaud[m] | good, so I think we can consider it as under control | 15:06 |
| amorin | we should still keep an action on my side for the next couple of weeks | 15:06 |
| hberaud[m] | the second task was: to raise warning about masakari failures on the ML https://review.opendev.org/c/openstack/requirements/+/933257 (assigned to myself) | 15:06 |
| hberaud[m] | amorin: yes | 15:07 |
| hberaud[m] | I duplicated it to next meeting | 15:07 |
| hberaud[m] | so concerning the masakari failure, I still have to contact them | 15:07 |
| hberaud[m] | so I also duplicate it | 15:08 |
| hberaud[m] | next task was: ML about the SGI working group and various sub aspects (assigned to myself) | 15:08 |
| hberaud[m] | and the mail was sent https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/T5DBJFNU65JDL5QSNO5B3DMHPNTDVETN/ | 15:08 |
| hberaud[m] | we got no response | 15:09 |
| hberaud[m] | and the last task was: revive the pipemutex removal discussion | 15:09 |
| hberaud[m] | and we got some related discussions with Itamar on IRC | 15:10 |
| hberaud[m] | #topic update from HTTP/SGI working group | 15:10 |
| hberaud[m] | ralonsoh: do you need some help with your wsgi topic? | 15:11 |
| ralonsoh | with the removal of the metadata proxy in Neutron | 15:11 |
| ralonsoh | that is now using eventlet.wsgi | 15:12 |
| ralonsoh | so I' | 15:12 |
| hberaud[m] | do you already have patches? | 15:12 |
| ralonsoh | so I'm deciding what framework should I use | 15:12 |
| ralonsoh | not at all, I'm just deciding how to proceed | 15:12 |
| hberaud[m] | and what is your trend? | 15:13 |
| ralonsoh | but, to be honest, I don't see any valid example in the openstakc repos | 15:13 |
| ralonsoh | maybe using flask, as in https://opendev.org/openstack/keystone/commit/4ec6bc5a44a85295395a3567fe90d1acc0dd7eff | 15:13 |
| hberaud[m] | what do you think about "routes"? | 15:13 |
| hberaud[m] | flask seems a good option | 15:13 |
| ralonsoh | we can sync at the end of this meeting, thanks! | 15:14 |
| hberaud[m] | ok no problem | 15:14 |
| hberaud[m] | @all anything else about WSGI? | 15:15 |
| hberaud[m] | ok, move on | 15:16 |
| hberaud[m] | #topic update from task working group | 15:16 |
| hberaud[m] | so the new oslo.service backend implementation has started https://review.opendev.org/c/openstack/oslo.service/+/935783 | 15:17 |
| hberaud[m] | we will soon seen follow up patches related to the implementation of the various layers of this topic | 15:17 |
| hberaud[m] | for now the existing patch is related to the implemention of the backend option's internal mechanisms | 15:18 |
| hberaud[m] | I don't think we have much to discuss related to this topic | 15:19 |
| hberaud[m] | #topic update from interoperability working group | 15:19 |
| hberaud[m] | the oslo.db new engine facade implementation faced some repetitive timeouts, I think they are legit and Mike told us that he will give another check on it https://review.opendev.org/c/openstack/oslo.db/+/922976?usp=search | 15:21 |
| hberaud[m] | this patch will be merged once the gates will be green | 15:21 |
| hberaud[m] | #topic Open Discussion | 15:21 |
| hberaud[m] | Anything else that you want to discuss today? | 15:22 |
| hberaud[m] | I think we can mention that we observed some failure within the recent support of python 3.13 https://github.com/eventlet/eventlet/issues/998 | 15:22 |
| hberaud[m] | I think the support of Python 3.13 needs some adjustments | 15:23 |
| croelandt | To be fair, all Glance tests pass except for one, so it's pretty good :) | 15:24 |
| hberaud[m] | Also, I think that we can mention that we are close from the end of the year, so our next meeting in 2 weeks will be the last of the years, apart if someone else volunteer to chair it during my absence (I won't be available during 2 weeks) | 15:25 |
| hberaud[m] | good news, thank you croelandt | 15:25 |
| hberaud[m] | Well, if you have no other topics I think we've covered it all. | 15:28 |
| hberaud[m] | thanks everyone | 15:28 |
| hberaud[m] | #endmeeting | 15:28 |
| opendevmeet | Meeting ended Tue Dec 3 15:28:41 2024 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:28 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_eventlet_removal/2024/openstack_eventlet_removal.2024-12-03-15.00.html | 15:28 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_eventlet_removal/2024/openstack_eventlet_removal.2024-12-03-15.00.txt | 15:28 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_eventlet_removal/2024/openstack_eventlet_removal.2024-12-03-15.00.log.html | 15:28 |
| amorin | thanks | 15:28 |
| ralonsoh | bye | 15:28 |
| ralonsoh | hberaud[m], do you have a couple of mins? | 15:28 |
| hberaud[m] | yes | 15:29 |
| ralonsoh | I'm trying to replace https://github.com/openstack/neutron/blob/master/neutron/agent/linux/utils.py#L463 | 15:29 |
| ralonsoh | that inherits from https://github.com/openstack/neutron/blob/master/neutron/api/wsgi.py#L102 | 15:29 |
| ralonsoh | we are using it as a wsgi proxy server for the metadata service | 15:29 |
| ralonsoh | so we can spawn, if needed, several metadata processes attending the requests | 15:30 |
| hberaud[m] | ack | 15:30 |
| ralonsoh | so I'm trying to find the best replacement for this module | 15:31 |
| ralonsoh | and sean sent me this: https://opendev.org/openstack/keystone/commit/4ec6bc5a44a85295395a3567fe90d1acc0dd7eff | 15:31 |
| ralonsoh | that is using flask | 15:31 |
| hberaud[m] | ack | 15:31 |
| ralonsoh | but O | 15:32 |
| ralonsoh | but I'm not sure if this is the best/easiest implementation, to be honest | 15:32 |
| ralonsoh | if I only need a wsgi proxy here | 15:32 |
| ralonsoh | do you have any recommendation? | 15:32 |
| hberaud[m] | in all the case you need a framework piece to manage your routing, isn't? | 15:33 |
| ralonsoh | not at all | 15:33 |
| hberaud[m] | let me double check your code | 15:34 |
| hberaud[m] | so if I correctly understood, your metadata process metadata execute requests and wait for? | 15:35 |
| ralonsoh | wait for | 15:36 |
| ralonsoh | the metadata process (or processes) is a server | 15:36 |
| hberaud[m] | ok | 15:36 |
| ralonsoh | https://github.com/openstack/neutron/blob/1249afb2ae6e5a44da3c4d4b563dd46eabc4e2c8/neutron/agent/metadata/proxy_base.py#L42 | 15:36 |
| ralonsoh | this is the base class (will depend on OVN or not OVN) | 15:37 |
| hberaud[m] | and I see that you already use webob | 15:38 |
| hberaud[m] | Where your Server part looks like more to something similar to unicorn or a server | 15:39 |
| hberaud[m] | is your solution is not simply to keep the metadata proxy_base more or less as it is, and to refactor the Server class with something like gunicorn? | 15:40 |
| ralonsoh | that could be an option, yes | 15:41 |
| hberaud[m] | I don't think I'll be able to give you an answer or good advice without digging a bit more on it | 15:41 |
| ralonsoh | I was thinking about refactoring the server class | 15:42 |
| hberaud[m] | I can double check it in the coming days and I can try to bring with something for the next meeting | 15:42 |
| ralonsoh | what means the question " something similar to unicorn or a server"? | 15:42 |
| ralonsoh | no rush, just if you have time, of course | 15:42 |
| hberaud[m] | Yes the server class seems to be your corner case | 15:42 |
| hberaud[m] | but I'm not sure to see where both class (MetadataProxyHandlerBase, and Server) are interconnected | 15:44 |
| ralonsoh | the first is the application of the second | 15:44 |
| hberaud[m] | at first glance there is no inheritance | 15:44 |
| hberaud[m] | and so you pass MetadataProxyHandlerBase to Server? | 15:45 |
| ralonsoh | yes, as the app | 15:46 |
| ralonsoh | one sec | 15:46 |
| hberaud[m] | and then Server serve it, as it would serve any other passed app | 15:46 |
| ralonsoh | https://github.com/openstack/neutron/blob/1249afb2ae6e5a44da3c4d4b563dd46eabc4e2c8/neutron/agent/ovn/metadata/server.py#L108 | 15:47 |
| hberaud[m] | so yes, in this case, at first glance, I do not see reason to touch to MetadataProxyHandlerBase | 15:47 |
| ralonsoh | yeah, that was my impression, that I need only a wsgi replacement | 15:47 |
| hberaud[m] | indeed | 15:49 |
| hberaud[m] | while you do not use eventlet.wsgi for routing, there is no reasons to refactor the app | 15:49 |
| hberaud[m] | (side) | 15:50 |
| ralonsoh | I'm using eventlet.wsgi.server to spawn the app | 15:50 |
| hberaud[m] | yeah but not the routes | 15:51 |
| hberaud[m] | so lets focus only on Server | 15:51 |
| hberaud[m] | sorry I need to setup my other meeting | 15:51 |
| ralonsoh | for sure, thanks! | 15:52 |
| hberaud[m] | in all the case I'll give it an other check | 15:53 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!