| *** dasm is now known as dasm|off | 01:24 | |
| opendevreview | Miro Tomaska proposed openstack/neutron stable/xena: Improve agent provision performance for large networks https://review.opendev.org/c/openstack/neutron/+/870785 | 03:13 |
|---|---|---|
| opendevreview | Miro Tomaska proposed openstack/neutron stable/wallaby: Improve agent provision performance for large networks https://review.opendev.org/c/openstack/neutron/+/870786 | 03:14 |
| opendevreview | Miro Tomaska proposed openstack/neutron stable/victoria: Improve agent provision performance for large networks https://review.opendev.org/c/openstack/neutron/+/870787 | 03:14 |
| opendevreview | Dmitrii Shcherbakov proposed openstack/neutron-lib master: ext-gw-multihoming: api-def and api-ref https://review.opendev.org/c/openstack/neutron-lib/+/870887 | 08:21 |
| *** froyo_ is now known as froyo | 08:29 | |
| opendevreview | Vasyl Saienko proposed openstack/neutron-vpnaas master: Fix pep6 and doc jobs https://review.opendev.org/c/openstack/neutron-vpnaas/+/870715 | 10:24 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron-vpnaas master: Fix pep8 and doc jobs for tox4 https://review.opendev.org/c/openstack/neutron-vpnaas/+/870715 | 10:45 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Remove backwards compatibility with OVN < v20.09 https://review.opendev.org/c/openstack/neutron/+/870621 | 11:31 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Add vnic_type and binding profile capabilities to LSP info https://review.opendev.org/c/openstack/neutron/+/867359 | 12:24 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Add vnic_type and binding profile capabilities to LSP info https://review.opendev.org/c/openstack/neutron/+/867359 | 12:46 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Introduce the new OVN Neutron Agent https://review.opendev.org/c/openstack/neutron/+/869909 | 12:47 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Implementation of OVN Neutron Agent https://review.opendev.org/c/openstack/neutron/+/870024 | 12:47 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] New OVN Neutron Agent extension: QoS for HWOL https://review.opendev.org/c/openstack/neutron/+/870115 | 12:47 |
| *** gthiemon1e is now known as gthiemonge | 13:09 | |
| frickler | ralonsoh: I forgot to mention that yesterday in the meeting, you should've made some advertising for the openinfra live session tomorrow. the audience here would likely be a good target | 13:23 |
| *** dasm|off is now known as dasm | 13:34 | |
| ges | Hello! We are having a little issue with segments: DB is taking a hit from queries to the segmenthostmappings. We noticed that for us, since we have exactly 1 segment per network, accessible from each host, segmenthostmappings contains N*M entries where N is the number of networks and M is the number of hosts. Potentially millions of entries. | 13:35 |
| ges | The expensive DB query cannot leverage indexes because it is built around a subquery, so we end-up fullscanning segmenthostmappings. Has someone else exeprienced something similar? | 13:36 |
| ralonsoh | frickler, during the next meeting, as we usually do, we announce the recorded session in https://openinfra.dev/live/#all-episodes | 13:40 |
| ralonsoh | but I was not aware of the next session | 13:40 |
| opendevreview | Bence Romsics proposed openstack/neutron-lib master: port-hints: api definition https://review.opendev.org/c/openstack/neutron-lib/+/870080 | 13:40 |
| ralonsoh | ges, can you point out what query is affected? or what CLI command or request? | 13:42 |
| opendevreview | Lajos Katona proposed openstack/neutron master: Fullstack: Wait placement process fixtrue to really stop https://review.opendev.org/c/openstack/neutron/+/870956 | 13:46 |
| ges | ralonsoh: here's the query. https://paste.openstack.org/show/bkJurCGDr4vHZlxT8bjA/ I am not sure 100% of when it runs though | 13:56 |
| ralonsoh | ges, and what is not indexed in this query? | 13:57 |
| opendevreview | Bence Romsics proposed openstack/neutron master: WIP port hints: api extension https://review.opendev.org/c/openstack/neutron/+/870081 | 14:00 |
| opendevreview | Bence Romsics proposed openstack/neutron master: WIP port hints: db model, db migration, ovo, extension driver, policies https://review.opendev.org/c/openstack/neutron/+/870082 | 14:00 |
| ges | segmenthostmapping is where we dont use the indexes. here's the analyze format=json for an example query : https://paste.openstack.org/show/bYGzLaBUvXOlxkKma8ld/ | 14:04 |
| ges | Maybe "not using index" is not super accurate, but what I understand of this query is we try to find all the hostmappings where a certain tenant could have access to, and at the moment it is using a temporary table because the way the query is crafted uses a subquery. But I am also wondering whether it is expected that, for a region with 1000 computes and 5000 networks, we would | 14:09 |
| ges | end-up with 5_000_000 rows in segmenthostmappings. | 14:09 |
| ralonsoh | ges, from this log, this is "networkrbacs" the one not indexed | 14:09 |
| ralonsoh | ges, what version of Neutron are you using? | 14:09 |
| ges | ralonsoh: stein | 14:10 |
| ralonsoh | ges, ok, quite old | 14:10 |
| ralonsoh | please check https://review.opendev.org/c/openstack/neutron/+/810072 | 14:10 |
| ralonsoh | this is the patch that adds the missing indexes to *rbacs tables | 14:10 |
| ralonsoh | that improves the rbac related queries a lot | 14:11 |
| ralonsoh | apart from this, there is a pending bug https://bugs.launchpad.net/neutron/+bug/1918145 | 14:11 |
| ralonsoh | and a n-lib patch to improve the rbac subquery modeling | 14:12 |
| ralonsoh | but this is still under development | 14:12 |
| ges | thanks! I could try with the index see if things get better, but I am not sure : we currently only have 222 rows in networkrbacs anyway | 14:19 |
| *** dkehn__ is now known as dkehn | 14:26 | |
| ges | I am also curious about the end-state of segmenthostmappings, when every network maps to every host. That sounds... bad. | 14:26 |
| opendevreview | Lajos Katona proposed openstack/neutron master: Avoid failure of get_device_details if net from cache is empty https://review.opendev.org/c/openstack/neutron/+/869678 | 14:28 |
| ralonsoh | It doesn't if it is done properly. In any case, why would you need that? | 14:28 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Bump the minimum pyroute2 version to 0.7.3 https://review.opendev.org/c/openstack/neutron/+/870963 | 14:30 |
| ges | ralonsoh: we enabled segments because we needed deffered IP allocation on some networks, but all segments can be reached by all hosts, hence the M*N in segmenthostmappings. | 14:43 |
| ges | I was able to reduce the cost of the query by having mariadb perform the query with straight_join. I think the query planner changes the plan once the table reaches a certain size or something. | 14:44 |
| ralonsoh | froyo, hi! | 15:15 |
| ralonsoh | can you please check https://review.opendev.org/c/openstack/releases/+/870639 comment? | 15:15 |
| froyo | ralonsoh: sure | 15:15 |
| ralonsoh | well, elodilles' comment there | 15:15 |
| ralonsoh | froyo, thanks! | 15:15 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [FT] Change the logic to check the IPv6 address state https://review.opendev.org/c/openstack/neutron/+/870967 | 15:33 |
| ralonsoh | mlavalle, slaweq lajoskatona please, can you check elodilles comments here? | 15:39 |
| ralonsoh | https://review.opendev.org/c/openstack/releases/+/870620 | 15:39 |
| ralonsoh | I've replied too. We should consider if we need to remove some merged backports | 15:39 |
| ralonsoh | (and maybe be more careful backporting patches) | 15:40 |
| lajoskatona | ralonsoh: ack, I check it (to tell the truth I checked this morning but I wasnt strong enough for it :)) | 15:47 |
| ralonsoh | thanks a lot | 15:47 |
| mlavalle | ralonsoh: will check it today | 16:15 |
| ralonsoh | thanks! | 16:18 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Bump the minimum pyroute2 version to 0.7.3 https://review.opendev.org/c/openstack/neutron/+/870963 | 16:44 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Ignore reno artefacts (RELEASENOTES.rst and reno.cache) https://review.opendev.org/c/openstack/neutron/+/870977 | 16:52 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [FT] Change the logic to check the IPv6 address state https://review.opendev.org/c/openstack/neutron/+/870967 | 16:56 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [OVN] Add vnic_type and binding profile capabilities to LSP info https://review.opendev.org/c/openstack/neutron/+/867359 | 17:02 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [FT] Change the logic to check the IPv6 address state https://review.opendev.org/c/openstack/neutron/+/870967 | 17:14 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: DNM - TESTING PATCH: neutron-functional-with-uwsgi https://review.opendev.org/c/openstack/neutron/+/870979 | 17:14 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Add tox Python3.11 job to the testing queues https://review.opendev.org/c/openstack/neutron/+/869196 | 17:24 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: [FT] Change the logic to check the IPv6 address state https://review.opendev.org/c/openstack/neutron/+/870967 | 17:29 |
| *** gmann is now known as gmann_afk | 17:29 | |
| *** gmann_afk is now known as gmann | 17:41 | |
| *** gmann is now known as gmann_afk | 18:06 | |
| opendevreview | Merged openstack/neutron master: Allow shared resources between physical and tunnelled networks https://review.opendev.org/c/openstack/neutron/+/866127 | 18:11 |
| lajoskatona | mlavalle: I started the etherpad for neutronclient: https://etherpad.opendev.org/p/python-neutronclient_deprecation | 18:14 |
| lajoskatona | mlavalle: tomorrow I try to add more details for the missing things, but now I think I listed there all things that are started or checked at least | 18:15 |
| *** gmann_afk is now known as gmann | 18:56 | |
| *** prometheanfire is now known as Guest1769 | 20:12 | |
| opendevreview | Merged openstack/neutron stable/wallaby: [ovn]neutron agent show real heartbeat_timestamp https://review.opendev.org/c/openstack/neutron/+/869352 | 21:01 |
| opendevreview | Merged openstack/neutron stable/wallaby: Fix handling the restart of ovn-controllers https://review.opendev.org/c/openstack/neutron/+/868264 | 21:01 |
| *** Guest1769 is now known as prometheanfire | 21:24 | |
| opendevreview | Jakub Libosvar proposed openstack/neutron master: ovn-migration: Stop neutron server while running db sync https://review.opendev.org/c/openstack/neutron/+/871004 | 22:08 |
| *** dmellado_ is now known as dmellado | 22:31 | |
| opendevreview | Dmitrii Shcherbakov proposed openstack/neutron-lib master: ext-gw-multihoming: api-def and api-ref https://review.opendev.org/c/openstack/neutron-lib/+/870887 | 22:44 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!