| opendevreview | Miro Tomaska proposed openstack/neutron master: FRR driver for a EVPN router https://review.opendev.org/c/openstack/neutron/+/988158 | 01:34 |
|---|---|---|
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add compute1 node to BGP multinode tempest job https://review.opendev.org/c/openstack/neutron/+/986075 | 07:10 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add compute1 node to BGP multinode tempest job https://review.opendev.org/c/openstack/neutron/+/986075 | 07:10 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add compute1 node to BGP multinode tempest job https://review.opendev.org/c/openstack/neutron/+/986075 | 07:13 |
| opendevreview | Bharath M V proposed openstack/neutron-tempest-plugin master: [Tap-as-a-Service] Adding 2 Tap as a service test cases. https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/989931 | 07:18 |
| opendevreview | Elvira García Ruiz proposed openstack/neutron master: Add PVLAN service plugin https://review.opendev.org/c/openstack/neutron/+/978798 | 07:47 |
| opendevreview | Elvira García Ruiz proposed openstack/neutron master: Add PVLAN OVN Driver https://review.opendev.org/c/openstack/neutron/+/985150 | 07:47 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron stable/2025.1: [Fullstack] Kill processes during cleanup phase https://review.opendev.org/c/openstack/neutron/+/989116 | 07:58 |
| slaweq | hi lajoskatona [ykarel_](https://matrix.to/#/@_oftc_ykarel_:matrix.org) - can you check https://review.opendev.org/c/openstack/neutron-fwaas/+/988588 when you will have few minutes? Thx in advance | 08:25 |
| opendevreview | Merged openstack/neutron-lib master: ml2: document LIFO delete order and binding level topology semantics https://review.opendev.org/c/openstack/neutron-lib/+/989150 | 08:38 |
| opendevreview | Bharath M V proposed openstack/neutron-tempest-plugin master: [Tap-as-a-Service] Adding 2 Tap as a service test cases. https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/989931 | 08:47 |
| opendevreview | Bharath M V proposed openstack/neutron-tempest-plugin master: [Tap-as-a-Service] Adding 2 Tap as a service test cases. https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/989931 | 09:08 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: WIP == Use tmpfs-backed MySQL template database for functional tests https://review.opendev.org/c/openstack/neutron/+/989945 | 10:57 |
| *** iurygregory_ is now known as iurygregory | 12:07 | |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Add ``l3-agent-scheduler-ha-priority`` extension for ML2/OVN https://review.opendev.org/c/openstack/neutron/+/982792 | 12:50 |
| opendevreview | Bharath M V proposed openstack/neutron-tempest-plugin master: [Tap-as-a-Service] Adding 2 Tap as a service test cases. https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/989931 | 12:53 |
| *** iurygregory_ is now known as iurygregory | 13:00 | |
| opendevreview | Bharath M V proposed openstack/neutron-tempest-plugin master: [Tap-as-a-Service] Adding 2 Tap as a service test cases. https://review.opendev.org/c/openstack/neutron-tempest-plugin/+/989931 | 13:31 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Revert "Group periodic thread workers together" https://review.opendev.org/c/openstack/neutron/+/989970 | 13:50 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Add compute1 node to BGP multinode tempest job https://review.opendev.org/c/openstack/neutron/+/986075 | 13:51 |
| sean-k-mooney | i could be missign somthing but https://review.opendev.org/c/openstack/neutron/+/989970 | 17:52 |
| sean-k-mooney | or rahter https://review.opendev.org/c/openstack/neutron/+/989970/comments/ffbb62f4_8fd3e239 | 17:52 |
| sean-k-mooney | the current code does thread_workers = [worker for worker in periodic_workers | 17:52 |
| sean-k-mooney | if worker.worker_process_count < 1] | 17:52 |
| sean-k-mooney | so the way a wroker ased to run it its own process is to set worker_process_count > 1 | 17:53 |
| sean-k-mooney | for the relevnet plugin | 17:53 |
| sean-k-mooney | the change i was making was only for the plugins that requested threaded workers | 17:53 |
| sean-k-mooney | by having that set to 0 | 17:53 |
| sean-k-mooney | ralonsoh: ^ | 17:53 |
| sean-k-mooney | so that shoudl have already been configurable | 17:54 |
| sean-k-mooney | or the cod in nuetonr has a bug where it does not supprot mixed workers at all | 17:54 |
| sean-k-mooney | is neutron missing config option to choso the worker count? | 17:57 |
| sean-k-mooney | it looks liek ye are hardcoding it https://github.com/openstack/neutron/blob/b1a6575e5bc27d44ec5e28f796fa86e37f94538c/neutron/services/bgp/worker.py#L23 | 17:57 |
| sean-k-mooney | so if bgp want to alreay run in a process shoudl it not pass 1 | 17:58 |
| sean-k-mooney | and idally have a config option to contol this for each worker | 17:58 |
| sean-k-mooney | the default in https://github.com/openstack/neutron/blob/master/neutron/worker.py#L18 is 1 | 17:58 |
| sean-k-mooney | so the bgp worker is expclity opting out of using process workers | 17:58 |
| sean-k-mooney | and opting in to usign a thread worker | 17:59 |
| sean-k-mooney | so reverting https://review.opendev.org/c/openstack/neutron/+/988810 does not look correct to me | 17:59 |
| ralonsoh | sean-k-mooney, I'm proposing a new patch, one sec | 18:04 |
| ralonsoh | sean-k-mooney, in any case, I would need to push a documentation patch to inform about what methods are executed | 18:05 |
| sean-k-mooney | ack | 18:05 |
| ralonsoh | the bug Eduardo opened is very descriptive | 18:05 |
| sean-k-mooney | yes but its wrong | 18:05 |
| sean-k-mooney | its true that its broken | 18:06 |
| sean-k-mooney | but as i noted https://github.com/openstack/neutron/blob/b1a6575e5bc27d44ec5e28f796fa86e37f94538c/neutron/services/bgp/worker.py#L23 | 18:06 |
| sean-k-mooney | the root cause is not my change | 18:06 |
| sean-k-mooney | its that the bgp worker is exicplty requetin to now run in a process | 18:06 |
| ralonsoh | sean-k-mooney, it is: the workers are wrapped inside the AllServicesNeutronWorker class | 18:06 |
| ralonsoh | and the methods called (wait in particular) are not the same | 18:07 |
| sean-k-mooney | only if worker_process_count is 0 | 18:07 |
| sean-k-mooney | well < 1 | 18:07 |
| sean-k-mooney | if you changed https://github.com/openstack/neutron/blob/b1a6575e5bc27d44ec5e28f796fa86e37f94538c/neutron/services/bgp/worker.py#L23 | 18:07 |
| sean-k-mooney | to 1 | 18:07 |
| ralonsoh | sean-k-mooney, if worker_process_count>0, the workers won't be executed in the periodic service | 18:07 |
| sean-k-mooney | ok yes but isnt that what you want here? | 18:08 |
| sean-k-mooney | you either need to dispatch the work ont a thread/rpocess pool form the perodic or you need to run it on its own process permently with its own perodic loop | 18:09 |
| ralonsoh | no, that limitation added in this patch (one sec) is not accurate | 18:09 |
| ralonsoh | Ic87ec5cc5037cab6eb0e24746a95ab4c8447870b | 18:09 |
| sean-k-mooney | i see | 18:11 |
| opendevreview | Eduardo Olivares proposed openstack/neutron master: Revert "Group periodic thread workers together" https://review.opendev.org/c/openstack/neutron/+/989970 | 18:12 |
| sean-k-mooney | ralonsoh: i dont think you propose the patch you intended? | 18:13 |
| sean-k-mooney | that is identicl to eh prior one | 18:13 |
| ralonsoh | sean-k-mooney, gerrit is not allowing me to propose a patch on top of ^^^ | 18:13 |
| ralonsoh | remote: commit 2f17248: warning: subject >50 characters; use shorter first paragraph | 18:14 |
| ralonsoh | To ssh://review.opendev.org:29418/openstack/neutron.git | 18:14 |
| ralonsoh | ! [remote rejected] HEAD -> refs/for/master%topic=periodic-worker (change https://review.opendev.org/c/openstack/neutron/+/923500 closed) | 18:14 |
| ralonsoh | and I need my patch on top of this one | 18:14 |
| sean-k-mooney | weried the subject lent warnign can be ignored | 18:14 |
| sean-k-mooney | have you tried -R | 18:14 |
| ralonsoh | yeah, let me try this | 18:15 |
| sean-k-mooney | you might also need to add your signed off by to the botome patch btu that shoudl nto be needed in general | 18:15 |
| sean-k-mooney | wait | 18:15 |
| sean-k-mooney | its complianing about https://review.opendev.org/c/openstack/neutron/+/923500 | 18:15 |
| ralonsoh | yes | 18:15 |
| ralonsoh | but this patch is already merged and closed | 18:16 |
| sean-k-mooney | right | 18:16 |
| sean-k-mooney | what are you tring to do in your patch? | 18:16 |
| sean-k-mooney | are you tryign to revert https://review.opendev.org/c/openstack/neutron/+/923500 | 18:16 |
| ralonsoh | ok ok, one sec | 18:16 |
| ralonsoh | I have a related-bug thing there | 18:16 |
| sean-k-mooney | im wonderign if you acendienally resued the same change id | 18:16 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: WIP == Group periodic thread workers with selective wrapping https://review.opendev.org/c/openstack/neutron/+/990002 | 18:17 |
| ralonsoh | aaaaand there we are | 18:17 |
| ralonsoh | so RPC will be always in a segregated process | 18:17 |
| ralonsoh | same as any worker with worker_process_count>0 | 18:17 |
| ralonsoh | and all worker_process_count=0, will be wrapped inside the allworkers class | 18:18 |
| sean-k-mooney | hum ok | 18:18 |
| ralonsoh | I need to document the API calls differences | 18:18 |
| sean-k-mooney | you do realse we end up with more then 1 rpc work today | 18:18 |
| ralonsoh | I need to go now, sorry | 18:18 |
| sean-k-mooney | no worries | 18:18 |
| sean-k-mooney | just an fyi in the os-vif jobs we had 7 process the parent and 6 chile process and two of them were rpc workers | 18:19 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!