Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-05-23
06:38:35 ileixe Hi forks.
06:40:08 ileixe I got an simple test using cpu pinning lately, and the result is somewhat unexpected. So if you are experince in the same environment, please let me know what's wrong.
06:41:11 ileixe My test is that VMs with cpu pinning host can interfere affect to each other.
06:42:45 ileixe I assume that VM with CPU pinning can achieve their own workload though, but the fact is If nosiy VM (which is also CPU pinned) is there they affect to each other.
07:09:38 openstackgerrit Merged openstack/nova stable/stein: Log when port resource is leaked during port delete https://review.opendev.org/657581
07:30:42 openstackgerrit Merged openstack/nova stable/stein: Include all network devices in nova diagnostics https://review.opendev.org/657125
07:30:55 openstackgerrit Merged openstack/nova master: Update description of valid whitelist for non-admin user https://review.opendev.org/642403
07:34:07 openstackgerrit Hamdy Khader proposed openstack/os-vif master: OVS DPDK port representors support https://review.opendev.org/658786
07:37:45 openstackgerrit Alexandre arents proposed openstack/nova master: Fix live-migration when glance image deleted https://review.opendev.org/659054
07:43:05 openstackgerrit Merged openstack/nova master: Skip existing VMs when hosts apply force_config_drive https://review.opendev.org/659703
07:44:42 openstackgerrit Lee Yarwood proposed openstack/nova stable/stein: Skip existing VMs when hosts apply force_config_drive https://review.opendev.org/660914
07:45:08 openstackgerrit Lee Yarwood proposed openstack/nova stable/rocky: Skip existing VMs when hosts apply force_config_drive https://review.opendev.org/660915
07:45:26 openstackgerrit Lee Yarwood proposed openstack/nova stable/queens: Skip existing VMs when hosts apply force_config_drive https://review.opendev.org/660917
08:05:00 alex_xu ileixe: what kind of nosiy you pointed to?
08:05:13 ileixe CPU load
08:05:24 ileixe alex_xu: I used stress and ssl encryption
08:05:32 ileixe tool called 'stress'
08:08:07 alex_xu ileixe: the vcpu and pcpu is 1:1 pinning, but in the same cpu socket, they still share cpu cache
08:08:50 ileixe I made them to be separated by using cpu_thread_policy
08:08:59 ileixe to 'require'
08:09:17 ileixe so I think no VMs share physical core for the test environment.
08:11:40 ileixe alex_xu: Have you ever expereince cpu pinning? Is it assure that VMs could not interfere each other?
08:12:03 ileixe (if there are only cpu loads)
08:13:21 lyarwood # git review
08:13:24 lyarwood Could not connect to gerrit.
08:13:34 lyarwood hmm anyone seeing this or have I just borked something locally?
08:15:35 lyarwood `RSA host key for review.openstack.org has changed and you have requested strict checking.` - nope it's remote
08:21:33 alex_xu ileixe: i only can suggestion you to check whether is system process running on the those pinned pcpu. and setting the emulator thread policy ensure the emulator thread not running on those pcpu.
08:23:14 ileixe alex_xu: Hm.. I double checked libvirt configuration and PID/stats at proc interface, and it was expected. Thanks anyway.
08:25:59 alex_xu ileixe: np, good luck
09:15:25 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove 'ComputeManager._reschedule' https://review.opendev.org/660774
09:15:25 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove cells v1 parameter from 'ComputeTaskAPI.resize_instance' https://review.opendev.org/651311
09:15:26 openstackgerrit Stephen Finucane proposed openstack/nova master: Stop passing 'kwargs' to 'rebuild_instance' https://review.opendev.org/651312
09:15:26 openstackgerrit Stephen Finucane proposed openstack/nova master: Stop passing 'delete_type' to 'terminate_instance' https://review.opendev.org/651313
09:15:27 openstackgerrit Stephen Finucane proposed openstack/nova master: filters: Stop handling cells v1 https://review.opendev.org/651314
09:15:27 openstackgerrit Stephen Finucane proposed openstack/nova master: Remove nova.compute.*API() shims https://review.opendev.org/660527
09:15:28 openstackgerrit Stephen Finucane proposed openstack/nova master: Ensure controllers all call super https://review.opendev.org/660950
09:49:36 kashyap stephenfin: Hi
09:50:03 kashyap stephenfin: I want to submit this doc to one of the upstream Nova admin guides -- which guide do you suggest: https://kashyapc.fedorapeople.org/Nova-Mitigation-for-Intel-MDS-Flaws.txt
09:52:20 stephenfin kashyap: probably the reference guide (doc/source/reference)
09:58:45 openstackgerrit Balazs Gibizer proposed openstack/nova master: Remove global state from the FakeDriver https://review.opendev.org/656709
09:59:28 openstackgerrit Balazs Gibizer proposed openstack/nova master: Enhance service restart in functional env https://review.opendev.org/512552
10:02:11 openstackgerrit Balazs Gibizer proposed openstack/nova master: Reproduce bug 1724172 in the functional test env https://review.opendev.org/512553
10:02:12 openstack bug 1724172 in OpenStack Compute (nova) "Allocation of an evacuated instance is not cleaned on the source host if instance is not defined on the hypervisor" [Medium,In progress] https://launchpad.net/bugs/1724172 - Assigned to Balazs Gibizer (balazs-gibizer)
10:02:26 openstackgerrit Balazs Gibizer proposed openstack/nova master: cleanup evacuated instances not on hypervisor https://review.opendev.org/512623
10:02:38 openstackgerrit Arnaud Morin proposed openstack/nova master: Force refresh instance network info on deletion https://review.opendev.org/660761
10:44:51 openstackgerrit Balazs Gibizer proposed openstack/nova master: Remove global state from the FakeDriver https://review.opendev.org/656709
10:44:51 openstackgerrit Balazs Gibizer proposed openstack/nova master: Enhance service restart in functional env https://review.opendev.org/512552
10:44:52 openstackgerrit Balazs Gibizer proposed openstack/nova master: Reproduce bug 1724172 in the functional test env https://review.opendev.org/512553
10:44:53 openstack bug 1724172 in OpenStack Compute (nova) "Allocation of an evacuated instance is not cleaned on the source host if instance is not defined on the hypervisor" [Medium,In progress] https://launchpad.net/bugs/1724172 - Assigned to Balazs Gibizer (balazs-gibizer)
10:44:53 openstackgerrit Balazs Gibizer proposed openstack/nova master: cleanup evacuated instances not on hypervisor https://review.opendev.org/512623
11:06:29 openstackgerrit Balazs Gibizer proposed openstack/nova master: Do not start nova-network in the notification func test https://review.opendev.org/660975
11:22:16 hrw morning
11:38:46 openstackgerrit Lee Yarwood proposed openstack/nova master: Use os-brick locking for volume attach and detach https://review.opendev.org/614190
11:43:33 openstackgerrit Hamdy Khader proposed openstack/os-vif master: OVS DPDK port representors support https://review.opendev.org/658786
11:59:40 openstackgerrit Arnaud Morin proposed openstack/nova master: Use force_refresh=True on network info update https://review.opendev.org/660982
12:50:44 openstackgerrit Adam Spiers proposed openstack/nova master: Move patch_exists() to nova.test.TestCase for reuse https://review.opendev.org/660500
13:22:16 kashyap stephenfin: Thx
13:27:24 efried kashyap: Did you see my patch answering the "how do we support renamed arch-specific cpu flag traits" question?
13:28:07 kashyap efried: Afraid, not yet. I'm at a one-day conference in the Netherlands, so my connectivity is a bit spotty today.
13:28:33 kashyap efried: Got the URL handy?
13:28:50 efried kashyap: https://review.opendev.org/#/c/660515/
13:29:21 efried kashyap: it's all red in the CI because it depends on your os-traits patch, but the dep chain doesn't work easily
13:29:51 efried kashyap: But once your patch is in an os-traits release, we put that in requirements.txt/lower-constraints.txt in this patch and it should go green.
13:30:16 efried anyway, point is it's not super hard to make it work, and we're probably going to have to do it eventually anyway, so might as well.
13:30:37 efried what this means to you is: no reason to hold off "moving" those traits in your patch.
13:31:51 openstackgerrit Adam Spiers proposed openstack/nova master: Move patch_exists() to nova.test.TestCase for reuse https://review.opendev.org/660500
13:32:08 aspiers efried: ^^^ this feels like the cleanest approach to me
13:33:01 kashyap efried: Thank you ... "For backward compatibility reasons, we have to continue exposing those versions of those traits" :-)
13:33:05 aspiers efried: I'm working on something similar for open()
13:33:17 aspiers but for now I need to go vote and get lunch. biab
13:33:35 kashyap efried: So my patch, shall I go back to making it a new PS of the older PS to preserve history? (Sorry for not doing that to begin with)
13:33:53 kashyap Or shall I just mention it in this new one, saying: "please read here for history"?
13:34:06 efried I would like someone libvirt-y, like stephenfin and aspiers, to have a look at that solution and make sure it's not insane.
13:34:25 efried kashyap: I would like it all kept in the original patch please.
13:34:42 efried aspiers: Well, you were the last one to touch the CPU_TRAITS_MAPPING anyway :)
13:34:47 aspiers true
13:35:08 aspiers reminder to self: voting and lunching means not being here :-/
13:35:10 kashyap efried: Will keep it in the original (then you need to update your PS, too)
13:35:37 efried kashyap: Thanks. Yeah, at this point I've proven the concept; I'll just wait until your patch is merged and released and then update mine properely.
13:35:39 efried properly
13:35:44 efried (sean-k-mooney moment)
13:36:18 kashyap efried: Yeah, noted. Tomm I'm off the afternoon. Can get to it only next week. Since there's no urgency here, hope that's OK
13:36:39 efried kashyap: Sure, that's fine, no hurry from my end
13:36:45 kashyap efried: Also you might want to have a gander at this (which I'll submit as a doc to Nova): https://kashyapc.fedorapeople.org/Nova-Mitigation-for-Intel-MDS-Flaws.txt
13:37:58 openstackgerrit Dan Smith proposed openstack/nova master: Make nova-next archive using --before https://review.opendev.org/661002
13:39:35 efried kashyap: lgtm
13:39:49 efried I'll have a few comments when it's up in gerrit
13:39:58 efried but overall clear and concise
13:40:23 kashyap efried: Thx; sure. Posting a variant of it now...before I forget.
13:40:43 kashyap Anything that improves clarity.
13:42:02 mfeder Hi all,
13:42:02 mfeder Is there any option how to execute command as another user (linux way : "sudo -u user `command`") ?
13:42:02 mfeder I would like to access regular-user's key pair using admin user.
13:42:02 mfeder E.g.
13:42:02 mfeder # as regular user :
13:42:03 mfeder `openstack keypair create kp1`
13:42:03 mfeder # as admin user:
13:42:04 mfeder `openstack keypair show kp1 `
13:42:38 mfeder Thanks for any hint !

Earlier   Later