Posted	Nick	Remark
#openstack-nova - 2019-04-12
01:48:13	mriedem	tetsuro: eh
01:48:29	mriedem	i don't love trying to future-proof that
01:48:56	mriedem	tetsuro: for what use case? user specifies a host and there are >1 nodes and you want to filter allocation candidates by all nodes?
01:49:02	mriedem	that would be 1000 for CERN
01:49:04	mriedem	*could
01:51:25	tetsuro	Not sure what happens on the performance
01:52:38	mriedem	i tend to think that's not a case we need to care about too much right now, until someone says they really need it
01:52:49	tetsuro	But I remember Surya said in the spec that he's not curious to have that for ironic cases
01:52:56	mriedem	i think in most cases that someone is forcing a server to a specific baremetal node, they are specifying the node
01:53:06	mriedem	right
01:53:24	openstackgerrit	Boxiang Zhu proposed openstack/nova-specs master: Add host and hypervisor_hostname flag to create server https://review.openstack.org/645458
01:53:29	mriedem	the only times i've heard of someone use the JsonFilter query hint for example is to force to a specific baremetal node
01:53:37	mriedem	via the hypervisor_hostname parameter
01:54:06	tetsuro	in which case the found node ==1
01:54:45	yaawang	mriedem: Hi, I have posted a new spec https://review.openstack.org/#/c/651681/. This spec will use the flavor extra spec and image metadata to datermine whether to use auto converge and post copy.
01:55:38	tetsuro	mriedem: okay, I'll leave this field as is i.e. UUIDField instead of ListUUIDField.
01:56:31	mriedem	yaawang: yeah i saw but haven't dug into it yet. i agree that might be a better option, but there are cases to consider like what happens if the image or extra spec says to use auto_converge or post-copy but the host is unable to provide those? do we fail the server create? do we ignore?
01:56:42	mriedem	for SLA reasons i think we'd have to fail
01:56:58	mriedem	tetsuro: ack
02:00:56	mriedem	yaawang: ok i left some thoughts on the spec,
02:01:00	mriedem	but traits might be the way to go here
02:04:57	yaawang	Yes, we need to check libvirt supports auto converge or post copy before live migration. I think end user prefer to use auto converge or post copy during live migration, but the end user hopes to be able to successfully carry out live migartion.
02:06:27	yaawang	Trait is a good idea, or schedule instances to the host which support auto converge or post copy.
02:07:13	mriedem	right, so i linked to the tpm spec because that case is very similar
02:07:30	mriedem	so read up on that and see if it could fit the same issue you're trying to solve
02:51:21	openstackgerrit	Merged openstack/nova stable/rocky: Move create of ComputeAPI object in websocketproxy https://review.openstack.org/649375
02:52:21	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Add in_tree field to RequestGroup object https://review.openstack.org/649534
02:52:22	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Pass target host to RequestGroup.in_tree https://review.openstack.org/650878
02:52:22	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Add get_compute_nodes_by_host_or_node() https://review.openstack.org/650877
02:52:23	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Query `in_tree` to placement https://review.openstack.org/649535
03:26:40	yaawang	mriemem: Okay, i'll update the spec.
05:24:09	openstackgerrit	Boxiang Zhu proposed openstack/nova master: Make evacuation respects anti-affinity rule https://review.openstack.org/649963
05:24:10	openstackgerrit	Boxiang Zhu proposed openstack/nova master: [WIP] Fix live migration break group policy simultaneously https://review.openstack.org/651969
05:48:05	openstackgerrit	Merged openstack/nova stable/rocky: Do not persist RequestSpec.ignore_hosts https://review.openstack.org/649326
06:13:42	boxiang	hi, does anyone know when/whether the requested_destination of request_spec will be save into db(persistent)? After I cold migrate the instance, I find it's null. Need help : )
06:58:01	openstackgerrit	Michael Still proposed openstack/nova master: Remove unused FP device creation and deletion methods. https://review.openstack.org/635433
06:58:01	openstackgerrit	Michael Still proposed openstack/nova master: Privsepify ipv4 forwarding enablement. https://review.openstack.org/635431
06:58:02	openstackgerrit	Michael Still proposed openstack/nova master: Move adding vlans to interfaces to privsep. https://review.openstack.org/635436
06:58:02	openstackgerrit	Michael Still proposed openstack/nova master: Privsep the ebtables modification code. https://review.openstack.org/635435
06:58:03	openstackgerrit	Michael Still proposed openstack/nova master: Move dnsmasq restarts to privsep. https://review.openstack.org/639280
06:58:03	openstackgerrit	Michael Still proposed openstack/nova master: Move iptables rule fetching and setting to privsep. https://review.openstack.org/636508
06:58:04	openstackgerrit	Michael Still proposed openstack/nova master: Move calls to ovs-vsctl to privsep. https://review.openstack.org/639282
06:58:04	openstackgerrit	Michael Still proposed openstack/nova master: Move router advertisement daemon restarts to privsep. https://review.openstack.org/639281
06:58:05	openstackgerrit	Michael Still proposed openstack/nova master: Move final bridge commands to privsep. https://review.openstack.org/639580
06:58:05	openstackgerrit	Michael Still proposed openstack/nova master: Move setting of device trust to privsep. https://review.openstack.org/639283
06:58:06	openstackgerrit	Michael Still proposed openstack/nova master: We no longer need rootwrap. https://review.openstack.org/554438
06:58:06	openstackgerrit	Michael Still proposed openstack/nova master: Cleanup the _execute shim in nova/network. https://review.openstack.org/639581
06:58:07	openstackgerrit	Michael Still proposed openstack/nova master: Cleanup no longer required filters and add a release note. https://review.openstack.org/639826
07:04:38	openstackgerrit	Michael Still proposed openstack/nova master: Remove fake_libvirt_utils from connection tests. https://review.openstack.org/642557
07:04:39	openstackgerrit	Michael Still proposed openstack/nova master: Remove fake_libvirt_utils from virt driver tests. https://review.openstack.org/643894
07:04:39	openstackgerrit	Michael Still proposed openstack/nova master: Remove fake_libvirt_utils from snapshot tests. https://review.openstack.org/642558
07:04:40	openstackgerrit	Michael Still proposed openstack/nova master: Remove remaining vestiges of fake_libvirt_utils from unit tests. https://review.openstack.org/643896
07:04:40	openstackgerrit	Michael Still proposed openstack/nova master: Remove fake_libvirt_utils from libvirt imagebackend tests. https://review.openstack.org/643895
07:04:41	openstackgerrit	Michael Still proposed openstack/nova master: Remove fake_libvirt_utils users in functional testing. https://review.openstack.org/644793
07:06:36	openstackgerrit	Boxiang Zhu proposed openstack/nova master: Make evacuation respects anti-affinity rule https://review.openstack.org/649963
07:09:00	alex_xu	kashyap: I replied https://review.openstack.org/#/c/642030/, let me know whether i understand correctly
07:13:24	alex_xu	boxiang: I guess it is here https://github.com/openstack/nova/blob/master/nova/conductor/manager.py#L361
07:17:01	kashyap	alex_xu: Will do
07:17:17	alex_xu	kashyap: thanks
07:20:25	kashyap	alex_xu: On your question ... you're right: we're mentioning it as an "extra" flag because a certain named CPU model may not contain some important CPU flags, e.g IvyBridge does not contain 'pcid'.
07:21:13	kashyap	(But the *host* itself is capable of it, and QEMU can recognize it. So we supply it explicitly)
07:22:05	kashyap	I will answer on the review, so you can respond at your own pace :-)
07:27:51	alex_xu	kashyap: for the use-case of config extra_flag, we want the IvyBridge + pcid, not just ensure pcid in the model of IvyBridge, right?
07:28:36	kashyap	Exactly. We're specifying '+ pcid' because it is not included in the libvirt's "idea of IvyBridge", but host is capable of PCID
07:29:46	alex_xu	kashyap: also, there won't have any model incompatible with pcid?
07:29:47	kashyap	alex_xu: To refresh your memory:
07:30:16	kashyap	(1) Intel Haswell, Broadwell, Skylake libvirt/QEMU models include 'pcid' by default.
07:30:48	kashyap	(2) The libvirt / QEMU CPU models "Nehalem", "Westmere", "SandyBridge", and "IvyBridge" will _not_ expose the 'PCID' capability by default, even if the host CPUs by the same name include it. I.e. 'PCID' needs to be explicitly specified when using the said virtual CPU models.
07:31:25	alex_xu	kashyap: nice refresh, that is interesting :)
07:31:43	kashyap	alex_xu: Okay, was wondering if what I wrote above was useful :-)
07:32:07	alex_xu	kashyap: so I guess the problem is that paragraph makes me confusing on what you want to check.
07:32:10	kashyap	alex_xu: I wrote that documentation in Nova's documentation.
07:32:12	kashyap	(And here too: https://kashyapc.fedorapeople.org/Reducing-OpenStack-Guest-Perf-Impact-from-Meltdown.txt)
07:32:34	kashyap	alex_xu: Actually, I didn't write that original paragraph in the spec, Jack Ding wrote it
07:32:52	kashyap	alex_xu: Let me think a bit more on to rephrase it to be more clear.
07:33:09	alex_xu	kashyap: ah, sorry, I thought it was removed from the original spec
07:33:17	kashyap	It was not
07:33:27	alex_xu	kashyap: yea, thanks, good to be clear what is checked we want
07:36:22	kashyap	alex_xu: As a general point: for most operators, using 'host-model' should suffice, because that will auto-add necessary CPU flags for Meltdown/Spectre
07:47:52	alex_xu	kashyap: thanks, good to know
08:20:49	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Pass target host to RequestGroup.in_tree https://review.openstack.org/650878
08:20:49	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Add get_compute_nodes_by_host_or_node() https://review.openstack.org/650877
08:20:50	openstackgerrit	Tetsuro Nakamura proposed openstack/nova master: Query `in_tree` to placement https://review.openstack.org/649535
08:23:07	openstackgerrit	Boxiang Zhu proposed openstack/nova-specs master: Add host and hypervisor_hostname flag to create server https://review.openstack.org/645458
08:48:50	openstackgerrit	Boxiang Zhu proposed openstack/nova master: Make evacuation respects anti-affinity rule https://review.openstack.org/649963
09:22:05	openstackgerrit	Boxiang Zhu proposed openstack/nova-specs master: Add host and hypervisor_hostname flag to create server https://review.openstack.org/645458
09:25:48	alex_xu	boxiang: I think we can persistent the destination_host
09:26:05	alex_xu	boxiang: since we reset it when do other migration, this is the rebuild https://github.com/openstack/nova/blob/master/nova/conductor/manager.py#L1034
09:27:31	kashyap	gibi: lyarwood: I hope I answered your questions on the Secure Boot spec: https://review.openstack.org/#/c/506720/
09:27:35	kashyap	Thanks for the review and questions!
09:27:50	boxiang	alex_xu: let me take a look
09:29:25	openstackgerrit	ya.wang proposed openstack/nova-specs master: Expose auto converge and post copy https://review.openstack.org/651681
09:33:18	boxiang	alex_xu:Yeah, I see other migration will call reset_forced_destinations. I will make some changes later.
09:36:13	alex_xu	boxiang: cool
09:36:31	alex_xu	boxiang: I think I can +2 after you update
09:36:59	openstackgerrit	Stephen Finucane proposed openstack/nova master: Bump to hacking 1.1.0 https://review.openstack.org/651553
09:37:00	openstackgerrit	Stephen Finucane proposed openstack/nova master: hacking: Resolve W503 issues https://review.openstack.org/651555
09:37:00	openstackgerrit	Stephen Finucane proposed openstack/nova master: hacking: Resolve E731 issues https://review.openstack.org/651554
10:28:09	openstackgerrit	Stephen Finucane proposed openstack/python-novaclient stable/rocky: Add test for console-log and docs for bug 1746534 https://review.openstack.org/652033
10:28:10	openstack	bug 1746534 in python-novaclient "encoding error when doing console-log" [High,Fix released] https://launchpad.net/bugs/1746534 - Assigned to Thomas Goirand (thomas-goirand)

Earlier   Later