Posted	Nick	Remark
#openstack-nova - 2017-06-07
10:47:03	bauzas	that's where I was confused
10:47:10	mdbooth	kashyap: However, this would be a nasty hack. We'd need to know how to identify that the hack was required. We'd also need to identify exactly at which point the hack was required.
10:47:57	openstackgerrit	Merged openstack/nova master: Use plain routes list for instance actions endpoint https://review.openstack.org/468338
10:48:12	mdbooth	kashyap: Nova certainly doesn't explicitly track or expose the balloon device in any way. However, if we needed to manipulate it during the migration workflow the libvirt driver could absolutely do that.
10:48:19	kashyap	mdbooth: What would be a nasty nack?
10:48:20	mdbooth	But it would be nasty.
10:48:29	mriedem	bauzas: ok, that's simpler than i thought it would be
10:48:32	kashyap	mdbooth: You mean, _if_ there weren't a clean libvirt API?
10:48:44	openstackgerrit	Merged openstack/nova master: Use plain routes list for os-fping endpoint https://review.openstack.org/468349
10:48:48	mdbooth	kashyap: If there's no clean libvirt api then we can't do it.
10:49:09	mdbooth	kashyap: If there is a clean libvirt api we can do it, but it would be a hack.
10:49:30	openstackgerrit	Pooja Jadhav proposed openstack/nova master: Nova flavor-list swap should be 0 when no value is specified https://review.openstack.org/471715
10:49:39	openstackgerrit	Matt Riedemann proposed openstack/nova master: Remove unused CONF import from placement/auth.py https://review.openstack.org/471716
10:49:43	kashyap	mdbooth: Yep, noted. Apparently, this bug (https://bugzilla.redhat.com/show_bug.cgi?id=1450524) Dave is trying to address is already fixed in a certain newer QEMU.
10:49:43	openstack	bugzilla.redhat.com bug 1450524 in qemu-kvm-rhev "qemu-kvm: VQ 0 size 0x100 Guest index 0x2010 inconsistent with Host index 0x2171: delta 0xfe9f" [High,Assigned] - Assigned to jasowang
10:50:15	bauzas	mriedem: FWIW, once I feel we get consensus on the patch, will backport it to stable/newton and ocata
10:50:30	mdbooth	kashyap: TBH, that reduces my motivation to introduce dirty hacks. If it's already fixed, the user can upgrade. If they don't want to upgrade... well that's a price they have to pay.
10:50:49	kashyap	mdbooth: Yes, I agree. That's why I thought it is worthwhile to mention that
10:51:08	kashyap	No need to add "hacks" when they aren't warranted
10:51:46	mdbooth	RHEL 7.2 to RHEL 7.3 is not an onerous update (stupid 3rd party kernel modules not withstanding)
10:52:08	kashyap	mdbooth: I _think_ the bug is during migration -- so even if the user upgrades
10:52:34	kashyap	IIUC, for the fix to take effect, they need a cold boot (stop / start) for the Nova guest to pick up the new machine type
10:53:01	mdbooth	Why is a new machine type required?
10:54:49	sfinucan	mriedem: Could you take another look at the PCI weigher today? See if I've cleared things up enough?
10:58:13	openstackgerrit	Stephen Finucane proposed openstack/nova master: libvirt: Make 'get_domain' private https://review.openstack.org/417378
11:00:02	openstackgerrit	Matt Riedemann proposed openstack/nova master: Fix cell0 naming when QS params on the connection https://review.openstack.org/471487
11:00:20	mriedem	bauzas: fixed that typo, +2
11:00:24	mriedem	sfinucan: oh maybe
11:00:31	mriedem	it hurts my brain though, looking at weighers
11:00:38	mriedem	*especially* pci weighers
11:00:45	openstackgerrit	falseuser proposed openstack/nova master: Delete the instance files after revert a resize task https://review.openstack.org/469814
11:01:18	mriedem	sdague: bauzas: sfinucan: this is a simple file move patch https://review.openstack.org/#/c/471425/ which would unblock a bp from getting completed
11:01:41	sfinucan	mriedem: I'd recommend a strong drink and two aspirin
11:01:56	mriedem	sfinucan: ok let me look after the joy of the quotas changes
11:03:03	sfinucan	ta
11:08:36	sfinucan	mriedem: Is it just me, or is that moved file totally unused?
11:08:46	sfinucan	wait - nvm
11:09:10	sfinucan	(it's an actual test - not a base class 🙈)
11:09:44	kashyap	mdbooth: Sorry, got distracted :-(
11:10:18	kashyap	To answer your question...
11:11:21	kashyap	The new machine type comes as part of the upgrade of QEMU
11:11:56	kashyap	And, for the fix to take effect, the guest should pick it up.
11:12:37	kashyap	(And the fix seem to take effect only via an explicit start / stop; not via migrating -- that's the issue).
11:18:17	openstackgerrit	Matt Riedemann proposed openstack/nova master: Add functional test for two-cell scheduler behaviors https://review.openstack.org/452006
11:22:21	openstackgerrit	Sean Dague proposed openstack/python-novaclient master: Prevent 2.32 release note from showing up in 9.0.0 release notes https://review.openstack.org/471383
11:31:44	openstackgerrit	Merged openstack/python-novaclient master: Skip rebuild functional test due to persistent vif plugging timeout https://review.openstack.org/471533
11:35:02	openstackgerrit	Stephen Finucane proposed openstack/nova-specs master: Clarify 'cells-count-resources-to-check-quota-in-api' https://review.openstack.org/471729
11:36:40	openstackgerrit	Merged openstack/nova master: Completely remove mox from unit/network/test_linux_net.py https://review.openstack.org/445504
11:40:31	mriedem	sfinucan: done https://review.openstack.org/#/c/379524/
11:42:41	sfinucan	mriedem: Am I correct in saying that scheduler filters run before the weighers?
11:43:32	mriedem	yes
11:43:42	mriedem	FilterScheduler._schedule method
11:51:32	openstackgerrit	Merged openstack/nova master: Use plain routes list for hosts endpoint instead of stevedore https://review.openstack.org/459231
11:53:12	mriedem	sean-k-mooney: would you agree with this? https://review.openstack.org/#/c/441183/15/nova/network/model.py@87
11:53:16	mriedem	s/would/do/
11:59:58	openstackgerrit	OpenStack Proposal Bot proposed openstack/python-novaclient master: Updated from global requirements https://review.openstack.org/469581
12:11:37	openstackgerrit	Michael Henkel proposed openstack/nova master: contrail: add vrouter VIF plugin type support https://review.openstack.org/441183
12:19:07	openstackgerrit	Matt Riedemann proposed openstack/nova master: contrail: add vrouter VIF plugin type support https://review.openstack.org/441183
12:21:45	bauzas	mriedem: I tend to think we should be supporting unicode
12:22:02	bauzas	mriedem: I'm trying to verify that
12:22:16	mriedem	bauzas: for the db connection url? otherwise that's a very random statement
12:22:21	bauzas	yup
12:22:28	bauzas	like "
12:22:30	bauzas	oops
12:22:31	bauzas	like é
12:22:45	mriedem	do people use unicode username/password/db names?
12:22:45	bauzas	I'm just verifying this
12:22:51	bauzas	mriedem: in France, yes
12:23:02	mriedem	oh france
12:23:41	openstackgerrit	Merged openstack/nova master: Add InstanceGroupList.get_counts() https://review.openstack.org/457857
12:23:42	bauzas	like logins with "hervé"
12:24:19	openstackgerrit	Merged openstack/nova master: Fix default_availability_zone docs https://review.openstack.org/438887
12:25:07	openstackgerrit	Merged openstack/nova master: Fix live migration devstack hook for multicell environment https://review.openstack.org/458537
12:25:50	openstackgerrit	Merged openstack/nova master: Move notifications/objects/test_base.py https://review.openstack.org/471425
12:27:09	mriedem	johnthetubaguy: sfinucan: bauzas: melwitt: am i crazy or did this move an option to another group and it's not tested? https://review.openstack.org/438887
12:27:13	openstackgerrit	Stephen Finucane proposed openstack/nova master: Add PCIWeigher https://review.openstack.org/379524
12:27:14	openstackgerrit	Stephen Finucane proposed openstack/nova master: Prefer non-PCI host nodes for non-PCI instances https://review.openstack.org/379625
12:27:51	bauzas	mriedem: nope, it's the global
12:27:58	bauzas	I verified it
12:28:02	bauzas	see my comment
12:28:30	mriedem	oh i see, nvm
12:28:33	mriedem	confusing
12:28:37	mriedem	compute_group vs compute_group_opts
12:30:02	sfinucan	bauzas, mriedem: were ye so inclined, I did modify that again here https://review.openstack.org/#/c/462469/
12:36:46	mriedem	sfinucan: what happens with the PCIWeigher if you don't have NUMATopologyFilter enabled?
12:36:52	mriedem	which isn't enabled by default,
12:36:59	mriedem	but all weighers are loaded and run automatically i believe
12:37:34	bauzas	holy fuck, how can I pdb nova-manage commands ?
12:37:46	bauzas	looks like my raw_input is crapped
12:38:17	bauzas	I just wanna know which kind of string we're getting from the conf object
12:38:32	bauzas	in a py2 world
12:38:42	bauzas	unless someone can answer this ^
12:39:58	mriedem	raise Exception
12:40:06	mriedem	debug like a pro
12:41:31	sfinucan	https://sd.keepcalm-o-matic.co.uk/i/keep-calm-and-printf-debugging.png
12:42:34	bauzas	please no
12:43:29	artom	If that fails because stdout gets lost somwhere, open a temp file and write to it
12:44:12	bauzas	mriedem: do you have a pattern ? I can raise some exception, but that's not helping me
12:44:41	artom	Well damn, looks like my idea is actually useful :D
12:44:55	sfinucan	mriedem: RE: what happens with the PCIWeigher if you don't have NUMATopologyFilter enabled?

Earlier   Later