| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-08-29 | |||
| 18:23:28 | sean-k-mooney | https://review.opendev.org/#/c/672595/21 | |
| 18:23:30 | artom | sean-k-mooney, the test results? | |
| 18:23:34 | sean-k-mooney | yep | |
| 18:23:46 | artom | The executive summary would be "woot", correct? | |
| 18:23:54 | sean-k-mooney | yes | |
| 18:24:00 | artom | Woot! | |
| 18:24:06 | sean-k-mooney | it seamed to work as intended | |
| 18:24:35 | sean-k-mooney | i am currently redeployting ot test a different patch | |
| 18:24:54 | sean-k-mooney | should i redeploy with v21 after or is it more or less the same | |
| 18:25:06 | mriedem | https://review.opendev.org/#/c/678887/ failed again | |
| 18:25:34 | artom | mriedem, yeah, I'm done faffing with the tests (for now, I suppose), looking into why virt_type still isn't being set correctly | |
| 18:25:39 | mriedem | https://zuul.opendev.org/t/openstack/build/49bea2053e864a9c818840b8c452606a/log/logs/subnode-2/etc/nova/nova-cpu.conf.txt.gz | |
| 18:25:39 | mriedem | the subnode is still wrong | |
| 18:25:40 | sean-k-mooney | woudl it be simpler if we jsut wrone a non legacy job to test it | |
| 18:26:03 | mriedem | i'm just going to write a devstack patch to set LIBVIRT_TYPE=kvm and depend on it | |
| 18:26:04 | sean-k-mooney | https://review.opendev.org/#/c/652197/ more or less is that non legacy job | |
| 18:26:11 | sean-k-mooney | but its failing because of fedora | |
| 18:26:46 | sean-k-mooney | new version of qemu is breaking oslo utils | |
| 18:27:11 | sean-k-mooney | but its fedora with the nightly build of qemu so we wont see that in distors for quite some time yet | |
| 18:27:39 | sean-k-mooney | mriedem: and ya that should also work | |
| 18:28:10 | sean-k-mooney | im going to try and get that nfv job passing again before RC1 | |
| 18:28:22 | sean-k-mooney | but can you wighin in on one aspect | |
| 18:28:59 | artom | mriedem, hah, that's one way of doing it | |
| 18:29:02 | sean-k-mooney | would you prefer a fedora based job that uses nightly qemu or a ubuntul 18.04 based job that only ran on vexhost/used neted virt | |
| 18:29:10 | openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: Run LM integration tests with NUMA flavor https://review.opendev.org/678887 | |
| 18:29:29 | sean-k-mooney | artom: well we already tried 4 others that didnt work | |
| 18:29:35 | sean-k-mooney | so its worth a shot | |
| 18:29:50 | artom | I'm not complaining :) | |
| 18:30:14 | sean-k-mooney | any opipion on ^ | |
| 18:30:44 | sean-k-mooney | e.g. should i piviot https://review.opendev.org/#/c/652197/ to just use nested vert on vexhost or try to get teh fedora based job working | |
| 18:30:53 | artom | I don't understand the differences between the options well enough to make a call | |
| 18:31:23 | sean-k-mooney | one ues fedroa an the virt preview repo to get a new enought qemu to support pinning and numa wiht out nested vert | |
| 18:31:31 | sean-k-mooney | so it would run on all the providers | |
| 18:31:47 | artom | Except that's now how it's done in real life, right? | |
| 18:31:54 | sean-k-mooney | the other option is to say screwit and just make it use nested vert and only run it on vexhost | |
| 18:32:00 | artom | So the second option would be better for a "permanent" CI thing | |
| 18:32:07 | artom | that's not | |
| 18:32:56 | sean-k-mooney | it depense if you are emulating other architefure like when rackspace used to run x86 on power its realisitc | |
| 18:33:21 | sean-k-mooney | but most people will use kvm or xen or hardware accleration | |
| 18:35:08 | artom | So I think yeah, go with the most realistic scenario | |
| 18:36:08 | artom | Also, keep in mind Fort Nebula was willing to provide multi-node NUMA flavors, and though infra wasn't thrilled because they want to be cloud-agnostic, they were OK with an experimental manually-triggered job to run on only that nodeset. | |
| 18:36:26 | mriedem | efried: belay that | |
| 18:37:08 | sean-k-mooney | ok in that case ill rebase it on top of your stuff to test that too but i might wait a bit before doing that | |
| 18:37:44 | artom | donnyd, actually, since we're on the topic, is ^^ still a thing you're planning to do, and do you need any assistance in doing it? | |
| 18:38:12 | donnyd | artom: I surely can | |
| 18:38:39 | sean-k-mooney | it would be nice to get a nightly job running. i can proably propose something if there is a lable wew can use | |
| 18:38:49 | donnyd | I wanted to make sure FN was stable and would run jobs without a bunch of FN related issues... seems to me like its going well enough | |
| 18:39:09 | donnyd | and I may need some assistance in getting it running | |
| 18:39:54 | sean-k-mooney | if you can create a flavor and we can add a lable to upstream node pool to provide a limited number of that flavor say 2-4 instances | |
| 18:40:02 | sean-k-mooney | i can proably do the rest. | |
| 18:40:09 | sean-k-mooney | but not for a week or two | |
| 18:40:55 | donnyd | I already have a custom label for FN nobody uses | |
| 18:41:33 | sean-k-mooney | yes bute we would need FN-multi_numa on that mapped to a specil flavor with hw:numa_nodes=2 set | |
| 18:42:25 | donnyd | oh ok | |
| 18:42:53 | donnyd | Well I am up for anything and I can do up to 4 numa nodes because my hypervisors have 4 procs | |
| 18:43:16 | sean-k-mooney | are they ivybridge or newer? | |
| 18:43:33 | donnyd | unlikley - checking | |
| 18:43:37 | artom | mriedem, FWIW, I've added next wednesday's stx meeting to my calendar, and installed that 'Zoom' thing they yse | |
| 18:43:39 | artom | *use | |
| 18:43:44 | sean-k-mooney | cause you could do 8 if they have more then 10 cores with a bios tweek but we only need 2 | |
| 18:44:04 | sean-k-mooney | with 2 numa nodes we can test everything | |
| 18:44:38 | sean-k-mooney | stx is starlingX | |
| 18:44:39 | donnyd | https://ark.intel.com/content/www/us/en/ark/products/64607/intel-xeon-processor-e5-4620-16m-cache-2-20-ghz-7-20-gt-s-intel-qpi.html | |
| 18:44:41 | donnyd | sandy bridge | |
| 18:44:47 | donnyd | no may not be helpful | |
| 18:44:56 | openstackgerrit | Merged openstack/nova master: Ensure online migrations have a unique name https://review.opendev.org/679096 | |
| 18:45:00 | artom | sean-k-mooney, yeah, they wanted to test NUMA LM as well | |
| 18:45:08 | openstackgerrit | Merged openstack/nova master: Allow strict_proxies for sdk Connection https://review.opendev.org/676926 | |
| 18:45:13 | sean-k-mooney | donnyd its the generation before intel added cluser on die | |
| 18:45:14 | artom | But... it might just be faster to get FN running in CI :) | |
| 18:45:26 | sean-k-mooney | but as i said 2 numa nodes is all that is needed | |
| 18:46:06 | sean-k-mooney | intel is also ment to be fixing the nfv ci at some point | |
| 18:46:14 | sean-k-mooney | maybe | |
| 18:46:19 | donnyd | so will mine work? | |
| 18:46:23 | sean-k-mooney | yep | |
| 18:46:46 | artom | sean-k-mooney, let's make a bet on who gets there first, FN in CI, Intel, or stx :D | |
| 18:47:03 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Rename the nova-grenade-live-migration job to nova-grenade-multinode https://review.opendev.org/678841 | |
| 18:47:03 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Move live_migration test hooks under gate/ https://review.opendev.org/679251 | |
| 18:47:04 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add cold migrate and resize to nova-grenade-multinode https://review.opendev.org/679210 | |
| 18:47:37 | mriedem | artom: you can ask them to use bluejeans during their meeting | |
| 18:48:04 | artom | It's OK, I'm not dogmatic about my video conferencing solutions :D | |
| 18:48:36 | sean-k-mooney | i think me testing stuff manully will be fater tehn all 3. i should start commint with my sean-mooney-ci gerrit account to confuse people | |
| 18:49:09 | donnyd | I can work on it later tonight (after my real job) or tomorrow morning (before my real job) | |
| 18:49:20 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Introduce live_migration_claim() https://review.opendev.org/635669 | |
| 18:49:21 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: LM: Use Claims to update numa-related XML on the source https://review.opendev.org/635229 | |
| 18:49:21 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: New objects for NUMA live migration https://review.opendev.org/634827 | |
| 18:49:22 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Deprecate CONF.workarounds.enable_numa_live_migration https://review.opendev.org/640021 | |
| 18:49:22 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: NUMA live migration support https://review.opendev.org/634606 | |
| 18:49:23 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: DNM: Run LM integration tests with NUMA flavor https://review.opendev.org/678887 | |
| 18:49:23 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Functional tests for NUMA live migration https://review.opendev.org/672595 | |
| 18:49:34 | donnyd | any good docs to point me to for setup? | |
| 18:50:23 | sean-k-mooney | donnyd: i can send you the command or document them in an etherpad if you like | |
| 18:50:46 | sean-k-mooney | over all there is not too much that you need too do | |
| 18:50:54 | sean-k-mooney | we jus tneed to know the name of the flavor | |
| 18:50:55 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Rename the nova-grenade-live-migration job to nova-grenade-multinode https://review.opendev.org/678841 | |
| 18:50:56 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add cold migrate and resize to nova-grenade-multinode https://review.opendev.org/679210 | |
| 18:51:00 | sean-k-mooney | and you need to add a few extra specs | |
| 18:51:06 | mriedem | artom: ha just did ours | |
| 18:51:34 | sean-k-mooney | and then 2 patches 1 for the project-config repo to add the new lable to nodepool and 1 to nova to add the perodic job | |
| 18:51:35 | mriedem | unbelievably, the teacher didn't comment on my new fresh haircut | |