| hungnt1 | i have 1 openstack cluster run on production. | 04:14 |
|---|---|---|
| hungnt1 | i have 1 openstack cluster run on production. | 04:14 |
| hungnt1 | all compute run CPU Intel Xeon Gold 6126 and fixed CPU model for VM is Skylake-Server | 04:14 |
| hungnt1 | now i have 10 new server run CPU Intel Xeon Gold 6330 | 04:14 |
| hungnt1 | How can I check the new CPU compatibility with SkyLake-Server for migration VM ? | 04:14 |
| hungnt1 | can anyone help me ? | 04:14 |
| sean-k-mooney[m] | the 6126 is skylake and 6330 are icelake | 04:23 |
| sean-k-mooney[m] | if you use a custom cpu model of skulake-server on all host they should be compatible | 04:24 |
| sean-k-mooney[m] | you will need to set cpu_mode=custom cpu_model=skylake-server in the libvirt section of the nova.conf | 04:25 |
| sean-k-mooney[m] | rather then relying on the default host-model | 04:25 |
| hungnt1_ | now, my config is cpu_model=skylake-server | 04:25 |
| sean-k-mooney[m] | you can check compatibality using libvirt | 04:25 |
| sean-k-mooney[m] | yep that should work | 04:25 |
| sean-k-mooney[m] | the simplest way yo check really is to just deploy the nodes and try the migration honestly but it should work | 04:26 |
| hungnt1_ | i need check compatibility before order server | 04:27 |
| hungnt1_ | need confirm 6330 model contain all feature flags of skylake-server | 04:29 |
| hungnt1_ | model ? | 04:29 |
| hungnt1_ | where can get list feature flag of SkyLake and IceLake | 04:33 |
| sean-k-mooney[m] | normally i would suggest downloading the datasheet form intel https://ark.intel.com/content/www/us/en/ark/products/212458/intel-xeon-gold-6330-processor-42m-cache-2-00-ghz.html | 04:33 |
| sean-k-mooney[m] | but it does not seam to be avaialable for that model | 04:33 |
| hungnt1_ | i cann't find it too | 04:35 |
| hungnt1_ | makes me not 100% sure about compatibility | 04:36 |
| sean-k-mooney[m] | you could compare the cpu models in libvirt/qemu | 04:36 |
| sean-k-mooney[m] | i.e. assert the icelake model is a superset of the skylake one | 04:37 |
| sean-k-mooney[m] | the abstract cpu models are intended too allow this type of compatiblity | 04:38 |
| sean-k-mooney[m] | but other then that im not sure there is much else we can suggest | 04:38 |
| hungnt1_ | maybe i need contact Intel to get list feature | 04:39 |
| hungnt1_ | and make compareration | 04:39 |
| sean-k-mooney[m] | https://paste.opendev.org/show/811678/ | 04:41 |
| hungnt1_ | oh | 04:42 |
| hungnt1_ | tks u | 04:42 |
| hungnt1_ | tôi sẽ kiểm tra, nó hữu ích | 04:42 |
| hungnt1_ | i will check, it helpful | 04:42 |
| sean-k-mooney[m] | no worries those xmls basicaly define the abstrack cpu models used or at least a semi human readable version of them | 04:43 |
| sean-k-mooney[m] | the only think im not sure about is <feature name='intel-pt' removed='yes'/> everything else looks like a superset | 04:45 |
| sean-k-mooney[m] | ah it processor tracing | 04:47 |
| sean-k-mooney[m] | https://bugzilla.redhat.com/show_bug.cgi?id=1853972 | 04:47 |
| sean-k-mooney[m] | looks like it was never implemented in qemu/kvm | 04:47 |
| hungnt1_ | in libvirt v4.5.0, it have <feature name='intel-pt'/> in Icelake map | 06:33 |
| hungnt1_ | https://www.diffchecker.com/XXMDm5kL | 06:35 |
| hungnt1_ | i had diffcheck in ver 4.5.0 | 06:35 |
| *** hemna0 is now known as hemna | 07:37 | |
| *** redrobot6 is now known as redrobot | 10:34 | |
| gibi | hi! I need a second core (other that bauzas) on the bugfix https://review.opendev.org/c/openstack/nova/+/820549 | 10:39 |
| stephenfin | gibi: Done. Nice work | 11:26 |
| gibi | stephenfin: thanks a lot! | 11:26 |
| gibi | stephenfin: this might be intresting to you as well https://review.opendev.org/c/openstack/nova/+/821423 but it is not urgent as I have to fix tempest first | 11:27 |
| stephenfin | gibi: Looking. I'm good on https://review.opendev.org/c/openstack/nova/+/766771 too. I know I originally wrote it, but that was ages ago and I'd forgotten about it so it's basically someone else's code now, heh | 11:29 |
| gibi | yeah, it is a long story. funnily we got hit by it downstream so suddenly it got prioritized up on my list | 11:31 |
| stephenfin | I think it was a downstream bug that caused me to work on that originally too, but I can't really be sure at this point | 11:31 |
| gibi | but the parallel tempest execution actually triggers the bug and as soon as I started rejecting it tempest started to fail randomly | 11:31 |
| gibi | so now I have a nice task to fix up tempest | 11:32 |
| gibi | hence https://review.opendev.org/c/openstack/tempest/+/821732 | 11:32 |
| *** tkajinam is now known as Guest8517 | 11:58 | |
| sean-k-mooney | gibi: it kind of looks like it deadlocked | 12:29 |
| gibi | sean-k-mooney: it has issues with the serial decorator itself as well. I'm troubleshooting locally | 12:29 |
| sean-k-mooney | gibi you know looking at the decorator you could justr set __serial=True in the defition of the subclasses that should be serial too | 12:44 |
| gibi | sean-k-mooney: sure, that is an alternative to the decorator | 12:44 |
| opendevreview | Erlon R. Cruz proposed openstack/nova master: Fix pre_live_migration rollback https://review.opendev.org/c/openstack/nova/+/815324 | 12:49 |
| opendevreview | Erlon R. Cruz proposed openstack/nova master: Adds regression test for bug LP#1897483 https://review.opendev.org/c/openstack/nova/+/821840 | 12:49 |
| gibi | sean-k-mooney: it appears that I should not rely on __serial as the __ makes is special from attribute lookup perspective | 12:56 |
| sean-k-mooney | oh yes __ at the start alters name mangaling | 12:56 |
| gibi | I knew I hit this before but I alway forget | 12:58 |
| sean-k-mooney | gibi: i woudl just make it all UPPERCASE and remvoe the undersocres | 12:59 |
| sean-k-mooney | like the class constants we have in the func test for api version and such | 12:59 |
| sean-k-mooney | bauzas: elodilles: this was rebased https://review.opendev.org/c/openstack/nova/+/816883 can you +w to get it moving again | 13:26 |
| opendevreview | Merged openstack/nova master: [rt] Apply migration context for incoming migrations https://review.opendev.org/c/openstack/nova/+/820549 | 13:33 |
| elodilles | sean-k-mooney: done | 13:36 |
| sean-k-mooney | elodilles: thanks | 13:36 |
| elodilles | no problem :) | 13:36 |
| sean-k-mooney | we had some downstream interest in it so just wanted to relay it upstream | 13:37 |
| *** kopecmartin_ is now known as kopecmartin | 13:41 | |
| *** whoami-rajat__ is now known as whoami-rajat | 14:00 | |
| erlon | sean-k-mooney: hey, can you get some time to review https://review.opendev.org/c/openstack/nova/+/821840? Its the regression test for https://review.opendev.org/c/openstack/nova/+/815324 | 14:00 |
| erlon | sean-k-mooney: please let me know if this is what you were thinking | 14:00 |
| erlon | also, I need a piece of advice. I cant make it work, and it seems that the mock I'm using is just not mocking. I tried several ways, one is commented | 14:01 |
| *** lbragstad8 is now known as lbragstad | 14:09 | |
| sean-k-mooney | erlon: not quite. you seam to be trying to create the reguression test more or less form scratch | 14:37 |
| sean-k-mooney | erlon: i would have started by using one of our libvirt funtal test case base classes | 14:38 |
| sean-k-mooney | also functional test should not really be mocking the way this is | 14:39 |
| sean-k-mooney | this is closer to a unit test | 14:39 |
| sean-k-mooney | erlon: so i proably should have started form 1897483 | 14:40 |
| sean-k-mooney | * https://github.com/openstack/nova/blob/master/nova/tests/functional/libvirt/base.py#L31 | 14:40 |
| sean-k-mooney | or from integrated_helpers._IntegratedTestBase | 14:41 |
| sean-k-mooney | if it was not a libvirt specific issue | 14:41 |
| sean-k-mooney | erlon: this is an example live migration functional test | 14:42 |
| sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/tests/functional/regressions/test_bug_1939545.py | 14:43 |
| *** simondodsley_ is now known as simondodsley | 14:58 | |
| *** mnaser_ is now known as mnaser | 14:58 | |
| *** johnsom_ is now known as johnsom | 15:00 | |
| *** viks___ is now known as viks__ | 15:00 | |
| *** bbezak_ is now known as bbezak | 15:00 | |
| *** erbarr_ is now known as erbarr | 15:01 | |
| *** NobodyCam_ is now known as NobodyCam | 15:02 | |
| *** clarkb is now known as Guest8536 | 15:11 | |
| opendevreview | Merged openstack/nova stable/wallaby: Ensure MAC addresses characters are in the same case https://review.opendev.org/c/openstack/nova/+/816883 | 15:21 |
| *** Guest8536 is now known as clarkb | 16:18 | |
| opendevreview | Alexey Stupnikov proposed openstack/nova stable/victoria: Ensure MAC addresses characters are in the same case https://review.opendev.org/c/openstack/nova/+/816927 | 19:26 |
| *** tbachman_ is now known as tbachman | 23:34 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!