| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2023-07-19 | |||
| 13:43:26 | dansmith | kashyap: I saw one of that test failing just now too | |
| 13:43:53 | dansmith | the test failure maybe kinda looked like it didn't actually rebuild the guest, because the host key didn't match (or something) | |
| 13:45:19 | kashyap | dansmith: Aah, I see. The host key mismatch means hopeffully it's a transient issue | |
| 13:45:33 | dansmith | well, I dunno, if we just saw two of them... | |
| 13:57:37 | kashyap | Heh | |
| 17:14:01 | mnaser | sean-k-mooney: i know you play a fair bit with these things .. besides asap2, is there another way to get hw accelerated vxlan.. i'm wondering if ovs-dpdk can work for tenant networks (vxlan) too... but havent seen much reference in docs | |
| 18:05:25 | sean-k-mooney[m] | we do not have support for hardware offloads with dpdk yet | |
| 18:05:39 | sean-k-mooney[m] | in openstack | |
| 18:07:31 | sean-k-mooney[m] | mnaser: https://specs.openstack.org/openstack/nova-specs/specs/2023.2/approved/support-napatech-linkvirtualization-smartnic.html is the spec you want | |
| 18:07:56 | sean-k-mooney[m] | that will add support for hardwar offloads with ovs-dpdk in general | |
| 18:08:16 | sean-k-mooney[m] | you do not need asap2 by the way | |
| 18:08:30 | sean-k-mooney[m] | upstream kernel ovs support hardware offloads | |
| 18:08:45 | sean-k-mooney[m] | but you do need the right nic driver | |
| 18:09:19 | sean-k-mooney[m] | you do not need the forked ovs however | |
| 18:10:13 | sean-k-mooney[m] | mnaser: https://docs.openvswitch.org/en/latest/topics/dpdk/phy/#representors is the upstream ovs doc for dpdk hardware offloads | |
| 18:15:31 | sean-k-mooney[m] | https://github.com/openvswitch/ovs/blob/master/Documentation/howto/dpdk.rst#flow-hardware-offload-experimental | |
| 18:35:01 | mnaser | sean-k-mooney[m]: so from an openstack pov, only offload we can do right now is asap2 | |
| 18:35:40 | sean-k-mooney[m] | technially we only supprot kernel ovs hardware offload with upstream ovs | |
| 18:36:49 | sean-k-mooney[m] | the in tree melanox kernel nic driver should be sufficient | |
| 18:37:09 | sean-k-mooney[m] | but asap is what nvidia/mellonx brand it as | |
| 18:37:20 | mnaser | do you know if intel has similar support? | |
| 18:37:46 | sean-k-mooney[m] | not currently. they had poc but no product in the market | |
| 18:40:00 | mnaser | sean-k-mooney[m]: i see, cause i ran into this https://networkbuilders.intel.com/docs/networkbuilders/intel-ethernet-controller-700-series-open-vswitch-hardware-acceleration-application-note.pdf | |
| 18:40:56 | sean-k-mooney[m] | yep with fortvial and ovs-dpdk | |
| 18:41:25 | mnaser | sean-k-mooney[m]: so only possible if that spec gets implemented? | |
| 18:42:06 | sean-k-mooney[m] | yep. dpdk hardware offloads in ovs is experimental | |
| 18:42:49 | sean-k-mooney[m] | but ovs-dpdk has had support for using rte_flow with some offload using fortviale/700 series and later | |
| 18:43:10 | sean-k-mooney[m] | for years | |
| 18:43:28 | sean-k-mooney[m] | but an end to end solution has not really matiarlised | |
| 18:43:35 | mnaser | i see, so really dont have a lot of choices in this section unfortunatel | |
| 18:43:36 | mnaser | y | |
| 18:43:41 | sean-k-mooney[m] | the spec i linked should enable that | |
| 18:44:03 | mnaser | bummer, ive been trying to get these palo alto vm series | |
| 18:44:07 | mnaser | and they refuse to work with mlx nics | |
| 18:44:21 | mnaser | and dpdk cant be offloaded for tenant networks | |
| 18:45:21 | sean-k-mooney[m] | for what its worth i joined redhat form intel 5 years ago and i was working on intel hardware offload support before i left | |
| 18:45:41 | sean-k-mooney[m] | and in that time they have not really made any forward progress until recently | |
| 18:46:02 | sean-k-mooney[m] | napatech are partly doing the enabling in that spec on behalf of intel | |
| 18:46:49 | sean-k-mooney[m] | which i found kind of odd when there PM said that at the PTG as intel used to have peole that worked on openstack | |
| 18:51:52 | mnaser | yeah its been frustrating since PA only supports a few sriov vf's which doesnt include mlx ones | |
| #openstack-nova - 2023-07-20 | |||
| 07:30:44 | opendevreview | Amit Uniyal proposed openstack/nova master: Delete dangling bdms https://review.opendev.org/c/openstack/nova/+/882284 | |
| 08:09:29 | opendevreview | Amit Uniyal proposed openstack/nova master: Reproducer for dangling bdms https://review.opendev.org/c/openstack/nova/+/881457 | |
| 08:09:30 | opendevreview | Amit Uniyal proposed openstack/nova master: Delete dangling bdms https://review.opendev.org/c/openstack/nova/+/882284 | |
| 08:51:30 | opendevreview | Artom Lifshitz proposed openstack/nova master: Call Neutron immediately upon _post_live_migration() start https://review.opendev.org/c/openstack/nova/+/883682 | |
| 08:51:30 | opendevreview | Artom Lifshitz proposed openstack/nova master: Parallelize calls to Neutron and Cinder in post_live_migration https://review.opendev.org/c/openstack/nova/+/883678 | |
| 10:47:29 | sean-k-mooney | dansmith: when you are around could you review https://review.opendev.org/c/openstack/devstack/+/888648 to unblock https://review.opendev.org/c/openstack/nova/+/886779 | |
| 10:55:30 | stephenfin | bauzas: There's a student from Oregon State University that has been working on OSC for the last few months and she's now eager to try her hand at one of the services themselves. How reasonable would it be to suggest a low-hanging nova API-related RFE for her to work on? | |
| 10:57:13 | stephenfin | I'm happy to keep covering the mentoring side of things but the ask would be for (a) suggestions of something relatively useful to do and (b) reviews if she works on it. She's around until June 2024 so it'll be relatively long-term (i.e. the C release cycle and a chunk of D) | |
| 10:58:56 | stephenfin | Easier things I can think of are changing the 'addresses' field in 'GET /servers/{serverID}' so we use an ID for the top-level key, to prevent munging of similarly named networks, and updating a few of the HTTP response codes, but I'm sure there's much more useful stuff to do | |
| 11:34:36 | opendevreview | Stephen Finucane proposed openstack/nova master: db: Remove legacy placement tables from API DB https://review.opendev.org/c/openstack/nova/+/889086 | |
| 11:34:36 | opendevreview | Stephen Finucane proposed openstack/nova master: db: Remove legacy tables from main DB https://review.opendev.org/c/openstack/nova/+/889087 | |
| 11:34:37 | opendevreview | Stephen Finucane proposed openstack/nova master: db: Remove legacy columns from main DB https://review.opendev.org/c/openstack/nova/+/889088 | |
| 11:55:13 | bauzas | stephenfin: sorry was at lunch | |
| 11:56:03 | bauzas | stephenfin: we don't have a lot of low-hanging-fruits RFEs for the moment, but sure we can help | |
| 13:23:30 | dansmith | sean-k-mooney: yep | |
| 13:25:35 | kashyap | dansmith: Morning, is this the same thing that you saw the other day too? the SSL subject alt name not matching host name: https://zuul.opendev.org/t/openstack/build/f0102bbbbb854e7cb6f9d99f582f8c4c | |
| 13:26:58 | sean-k-mooney | dansmith: if you have time to add another small devstack patch to the list https://review.opendev.org/c/openstack/devstack/+/878718 fixes reboots on fedora. | |
| 13:27:24 | sean-k-mooney | i.e. makeing rebooting the devstack vm work if you use fedora/centos for the vm like it does on ubuntu/debiain | |
| 13:27:49 | sean-k-mooney | i fixed that based on the ptg feedback but havnt looked at it since | |
| 13:43:34 | dansmith | kashyap: that should be fixed now | |
| 13:43:44 | dansmith | it was a bug in the ubuntu curl package which was reverted | |
| 13:44:25 | dansmith | sean-k-mooney: got it | |
| 13:44:41 | kashyap | dansmith: Aaah, thank you! Lemme recheck | |
| 15:02:06 | opendevreview | Sylvain Bauza proposed openstack/nova master: Add a new NumInstancesWeigher https://review.opendev.org/c/openstack/nova/+/886232 | |
| 16:13:31 | opendevreview | Maxim Monin proposed openstack/nova master: Server Rescue leads to Server ERROR state if original image is deleted https://review.opendev.org/c/openstack/nova/+/872385 | |
| 19:32:32 | opendevreview | Jay Faulkner proposed openstack/nova master: [ironic] Optionally set lessee on deployment https://review.opendev.org/c/openstack/nova/+/888290 | |
| 20:43:24 | opendevreview | Dan Smith proposed openstack/nova master: DNM: Test tempest worker re-distribution https://review.opendev.org/c/openstack/nova/+/889130 | |
| 20:56:06 | opendevreview | Ghanshyam proposed openstack/nova master: DNM: testing Tempest run increased concurrency https://review.opendev.org/c/openstack/nova/+/889135 | |
| 20:56:55 | opendevreview | Felipe Reyes proposed openstack/nova master: Register riscv64 architecture. https://review.opendev.org/c/openstack/nova/+/889137 | |
| #openstack-nova - 2023-07-21 | |||
| 06:22:45 | opendevreview | Maxim Monin proposed openstack/nova master: Server Rescue leads to Server ERROR state if original image is deleted https://review.opendev.org/c/openstack/nova/+/872385 | |
| 10:50:15 | opendevreview | Amit Uniyal proposed openstack/nova master: Added context manager for instance lock https://review.opendev.org/c/openstack/nova/+/873648 | |
| 10:50:15 | opendevreview | Amit Uniyal proposed openstack/nova master: Disconnecting volume from the compute host https://review.opendev.org/c/openstack/nova/+/877446 | |
| 12:00:07 | opendevreview | Amit Uniyal proposed openstack/nova master: Disconnecting volume from the compute host https://review.opendev.org/c/openstack/nova/+/877446 | |
| 13:32:55 | opendevreview | Takashi Natsume proposed openstack/nova-specs master: Create specs directory for 2024.1 Caracal https://review.opendev.org/c/openstack/nova-specs/+/889177 | |
| 13:36:57 | opendevreview | Takashi Natsume proposed openstack/nova master: Update contributor guide for 2023.2 Bobcat https://review.opendev.org/c/openstack/nova/+/876447 | |
| 13:41:33 | dansmith | bauzas: can you fix these nits real quick and I'll +2+A? https://review.opendev.org/c/openstack/nova/+/886232 | |
| 13:41:35 | dansmith | or I can do it | |
| 13:41:44 | bauzas | dansmith: sure, lemme look | |
| 13:42:18 | bauzas | dansmith: ack, thanks for reviewing, will do | |
| 13:51:07 | opendevreview | Sylvain Bauza proposed openstack/nova master: Add a new NumInstancesWeigher https://review.opendev.org/c/openstack/nova/+/886232 | |
| 13:52:08 | auniyal | dansmith and all continuing to lock discussion - what type of action can be prevented on instance if instance is locked. | |
| 13:54:12 | dansmith | auniyal: anything with this, AFAIK: https://github.com/openstack/nova/blob/master/nova/compute/api.py#L2701 | |
| 14:00:22 | auniyal | th deorator is with stop and reboot as well, but I was able to reboot and stop locked instance | |
| 14:00:31 | dansmith | as admin or user? | |
| 14:00:37 | auniyal | yes admin | |
| 14:00:51 | dansmith | auniyal: https://github.com/openstack/nova/blob/master/nova/compute/api.py#L240 | |
| 14:00:58 | dansmith | admins are not bothered by the lock | |
| 14:01:18 | auniyal | ack, | |
| 14:01:30 | auniyal | nova-manage can be run by user right | |
| 14:02:21 | dansmith | no | |
| 14:02:37 | dansmith | nova-manage requires physical access to the backend infrastructure | |
| 14:02:54 | dansmith | it's almost entirely direct-to-db, bypassing all of this sort of stuff in almost all cases | |
| 14:03:07 | dansmith | https://docs.openstack.org/nova/rocky/cli/nova-manage.html | |
| 14:03:14 | dansmith | "nova-manage controls cloud computing instances by managing various admin-only aspects of Nova." | |
| 14:12:23 | auniyal | thanks dansmith, | |
| 14:25:45 | auniyal | dansmith, I am facing this issue right now - https://review.opendev.org/c/openstack/nova/+/873648/comment/af765efc_a7161f39/ | |
| 14:25:46 | auniyal | so this is when the instance is not locked already. | |
| 14:25:46 | auniyal | however when I tested it by manually locking before running nova-manage cmd, db query went fine and there were no err msg. | |
| 14:25:46 | auniyal | even though in both cases while running instance_action.finish, instance was locked, query context and query values were same as required. | |
| 14:29:51 | dansmith | sorry, I'm not sure what context manager you mean you've removed, but action.finish failing means action_start never actually hit the DB, and none of that should have anything to do with lock as noted above | |