| manuvakery1 | Query repost: | 04:34 |
|---|---|---|
| manuvakery1 | Hi, I have a 2 node cluster and I am running a rally scenario against it to create and delete servers. When i set the server count to 2 I see the response time as follows | 04:34 |
| manuvakery1 | nova.boot_servers : 11.941 sec (Avg) | 04:34 |
| manuvakery1 | nova.delete_servers: 14.675 sec (Avg) | 04:34 |
| manuvakery1 | make the server count to 5 added a huge difference in delete server | 04:34 |
| manuvakery1 | nova.boot_servers: 18.803 sec (Avg) | 04:34 |
| manuvakery1 | nova.delete_servers: 42.988 sec (Avg) | 04:34 |
| manuvakery1 | is this expected? | 04:34 |
| manuvakery1 | the compute nodes are not under any load | 04:34 |
| opendevreview | Jorge San Emeterio proposed openstack/nova master: WIP: Testing whether tests on bug#1998148 still fail. https://review.opendev.org/c/openstack/nova/+/880135 | 11:06 |
| opendevreview | Jorge San Emeterio proposed openstack/nova master: WIP: Testing whether tests on bug#1998148 still fail. https://review.opendev.org/c/openstack/nova/+/880135 | 13:34 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 13:37 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 13:38 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 13:39 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 13:40 |
| sean-k-mooney | bauzas: fyi that should now be ready to go ^ | 13:41 |
| bauzas | sean-k-mooney: yup, I saw the updates :) | 13:55 |
| sean-k-mooney | the ping was more to say im done updatign it if you want to review | 13:56 |
| sean-k-mooney | bauzas: once you are done with that i would also liek to land https://review.opendev.org/q/topic:sqlalchemy-20+project:openstack/nova+status:open if we can form stephenfin | 13:57 |
| sean-k-mooney | gibi: ^ melwitt: ^ if either of ye can review those that would also be good | 13:58 |
| stephenfin | sean-k-mooney: Speaking of, just replied on https://review.opendev.org/c/openstack/nova/+/860850 | 14:20 |
| kashyap | gibi: bauzas: When you can, can you have a look at this from Sean: https://review.opendev.org/c/openstack/nova/+/880231 | 14:22 |
| bauzas | kashyap: already know it :) | 14:23 |
| kashyap | Thank you; I know, I'm slow in catching up :) | 14:23 |
| opendevreview | Merged openstack/os-traits master: Update python testing as per zed cycle testing runtime https://review.opendev.org/c/openstack/os-traits/+/841682 | 15:16 |
| dansmith | bauzas: AFAICT, we don't ever hit this trap in any of our CI jobs: https://review.opendev.org/c/openstack/nova/+/879687/5/nova/compute/resource_tracker.py | 16:47 |
| sean-k-mooney | i think the only way we could try and create a claim on a disabled node is if there was a race | 17:09 |
| sean-k-mooney | specificaly if between the time we selected the node in the schdluer and before the node reached the compute we updated the service to disabled | 17:09 |
| sean-k-mooney | assuming self.diabled is the compute service disabled value? | 17:10 |
| dansmith | sean-k-mooney: no | 17:24 |
| dansmith | sean-k-mooney: we had a long conversation about it yesterday | 17:24 |
| dansmith | I replied to your comments on the review | 17:27 |
| sean-k-mooney | oh ok this is not related to the compute service state. reading | 18:20 |
| sean-k-mooney | oh ok so for non-ironic | 18:21 |
| sean-k-mooney | for use to get here we woudl need two compute-agents that the saem config.host value so that they both shared teh same queue in rabbit or somethign equaly bad/wrong | 18:22 |
| sean-k-mooney | im jsut tryign to think how we would recive a request fo another compute node that this agent does not know about | 18:23 |
| sean-k-mooney | im am wonderign if i can think of a way to repoduce this but mostly coming up blank | 18:25 |
| sean-k-mooney | maybe if you set Conf.host=X and then changed /etc/hostname with preantelope | 18:26 |
| sean-k-mooney | a pendign request in the queue for the old comptue node would end up here? | 18:26 |
| sean-k-mooney | that cant happen now in either case since the agent wont start if you did this | 18:27 |
| sean-k-mooney | so ya i agree this shold not jsut be ignored | 18:27 |
| sean-k-mooney | i just dont knwo if its possibel to cause today | 18:28 |
| sean-k-mooney | is this code path used in the ironic case with the hashring stuff? | 18:28 |
| sean-k-mooney | i would expect it to be common | 18:29 |
| sean-k-mooney | it was not a behvior i was aware of in any case | 18:30 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 18:39 |
| opendevreview | sean mooney proposed openstack/nova master: add hypervisor version weigher https://review.opendev.org/c/openstack/nova/+/880231 | 18:43 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!