Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-12-13
18:45:35 mpiwowarczy hello
18:46:36 mpiwowarczy can I have a question to nova deprecated personality parameter?
18:47:14 mpiwowarczy seems it will be removed in microversion 2.57
18:47:42 mpiwowarczy but can not found when 2.57 will be used?
18:47:55 mpiwowarczy for Rocky release it was microversion 2.27
18:48:06 mpiwowarczy which version will be used in Stein?
18:49:01 mpiwowarczy or is there any fixed date when 2.57 version will be dropped?
18:54:22 artom mpiwowarczy, microversions are never dropped, we still support 2.1 if the client asks for it
18:54:34 artom mpiwowarczy, in terms of releases, have a look at https://docs.openstack.org/nova/latest/reference/api-microversion-history.html
18:56:49 mpiwowarczy artom: thanks for the answer
18:56:56 mpiwowarczy so when personality files (files injection) feature will be dropped?
18:58:11 mpiwowarczy will be this supported in Stein release?
19:01:35 artom mpiwowarczy, it'll never be dropped per se
19:02:05 artom mpiwowarczy, you can keep using it in Stein if the client asks for a microversion < 2.57
19:03:49 mpiwowarczy ok, thanks for your help
19:07:37 melwitt mpiwowarczy: it's dropped in a sense that if/when you find yourself needing to request microversion >= 2.57 when you call the REST API to get a different feature/addition, you will not be able to use personality files at the same time/in the same call
19:08:40 mpiwowarczy melwitt: I see
19:10:15 mpiwowarczy do you intend to maintain all versions, starting from 2.1, in all future OpenStack releases?
19:11:22 mpiwowarczy day by day it is more difficult to maintain :)
19:18:33 mriedem what is more difficult to maintain? writing apps against the API? or maintaining the various microversions in the API?
19:18:54 mriedem the client opts into the behavior, so if you are ok with v2.1 then you don't need to ever change that
19:20:01 mpiwowarczy oh yeah, you're right
19:20:49 mpiwowarczy just one more question - latest novaclient 11.1.0, which microversion it uses?
19:20:59 mriedem the alternative is we don't use microversions, change the api from release to release, and all client software everywhere in the world has to deal with that based on the cloud they are talking to
19:21:00 mpiwowarczy where I can check that?
19:21:27 mriedem the client-side support is defined here https://github.com/openstack/python-novaclient/blob/master/novaclient/__init__.py#L28
19:21:35 mriedem novaclient python APIs by default use 2.1
19:21:51 mriedem the CLI by default negotiates the highest supported microversion between the client and the server
19:22:24 mriedem so if you're using novaclient as a python sdk, then it's opt-in,
19:22:37 mriedem if you're using the CLI, it's negotiated unless you tell it otherwise
19:23:09 mpiwowarczy it is about SDK
19:23:14 mriedem then it's opt-in
19:23:27 mpiwowarczy we have defined version python-novaclient==11.1.0 and we use file injection
19:23:32 mriedem api methods in novaclient are wrapped / bound based on what versions they support
19:23:49 mriedem as i said, unless you opt into a newer microversion, it will use 2.1
19:23:51 mriedem for the sdk
19:23:55 openstackgerrit Merged openstack/nova stable/rocky: Harden placement init under wsgi https://review.openstack.org/617297
19:24:11 mpiwowarczy *python-novaclient>=9.1.0 # Apache-2.0
19:25:05 mpiwowarczy thanks, you helped me a lot
19:25:32 mriedem yw
19:33:26 melwitt mriedem: do happen to know if we have the opposite of --availability-zone ZONE:HOST:NODE for admins where it would mean launch the server anywhere but this node? curious because I know in the request spec we have both force_hosts and ignore_hosts
19:33:50 melwitt I didn't think we did but not 100% sure
19:34:13 mriedem not from the api i don't think so
19:34:24 mriedem ignore_hosts is for when you're moving a server and you ignore the source host from schedluing
19:34:27 cdent i've had people ask about that too, and also couldn't find it
19:34:30 mriedem or ignore hosts you've rescheduled from
19:35:01 melwitt thanks. someone asked recently and I told them I think forbidden traits is the closest thing to something like that
19:35:21 mriedem well,
19:35:45 mriedem for that to work, you'd have to temporarily put a custom trait on the blacklisted host, schedule to !host via forbidden trait, and then remove the trait
19:36:03 mriedem if you're going to do that, you might as well just disable the compute service and then re-enable it after the scheduling is done
19:36:11 mriedem b/c the ComputeFilter essentially does the same thing
19:36:19 mriedem unless you're cern and you drop the max placement results down to 10
19:36:57 melwitt yeah, well I was thinking more general if there was an underlying reason to avoid that host, like there's special hardware there that you want to avoid taking up
19:37:26 melwitt but as a one shot thing if there's no other meaning to it, yeah, I agree disabling is easier
19:38:18 mriedem i believe tetsuro's forbidden aggregates spec goes into detail about that
19:38:20 mriedem for blazar
19:38:45 mriedem it is essentially the same problem, you have a special pool of resources that non-special requests can't have access to
19:39:18 melwitt yeah
20:09:16 openstackgerrit Matt Riedemann proposed openstack/nova master: Remove GROUP BY clause from CellMapping.get_by_project_id https://review.openstack.org/619061
20:09:17 mriedem zzzeek: i did that subquery and it works ^ thanks for the tip there
20:09:45 zzzeek mriedem: that's great. keep pinging me for thse so we can make the DB code better :)
21:15:45 openstackgerrit Jack Ding proposed openstack/nova master: Preserve UEFI NVRAM variable store https://review.openstack.org/621646
22:11:05 openstackgerrit Merged openstack/python-novaclient master: Fix flavor keyerror when nova boot vm https://review.openstack.org/582147
22:11:05 openstackgerrit Merged openstack/python-novaclient stable/rocky: Improve the description of optional arguments https://review.openstack.org/602056
22:15:43 mriedem melwitt: you want to just fast approve this backport https://review.openstack.org/#/c/623546/ for rocky and then we can queue up a rocky release with a couple of depends on
22:15:53 mriedem the only other one i'm watching is https://review.openstack.org/#/c/623507/
22:17:57 melwitt ok. depends-on being this one I'd think? https://review.openstack.org/624982
22:18:31 openstackgerrit Merged openstack/nova master: Use external placement in functional tests https://review.openstack.org/617941
22:37:03 mriedem melwitt: i would like to let that bake a bit
22:37:12 melwitt k
22:37:15 mriedem dan and i were talking about something in there as an unexpected side effect this morning
22:37:17 mriedem hence the depends-on
22:37:47 mriedem so just the docs change and lee's backport
22:37:54 melwitt ok
22:38:42 mriedem if you're curious, the side effect is if the cell is configured to hit the api db, which lots of people still do for up-calls to work, and upgrade levels is 'auto', it means all cells will use the minimum compute version even if a particular cell is fully upgraded
22:38:48 mriedem which is different from before
22:38:51 mriedem not bad, but different
22:40:13 melwitt I was, thanks for saving me from having to read back in the irc log
22:43:10 mriedem Kevin_Zheng: i've got some questions/confusion on yikun's live migration completion timeout action patch if you want to kick him when you're in the office https://review.openstack.org/#/c/619143/
22:57:53 openstackgerrit Merged openstack/os-resource-classes master: Tune up documentation to be more useful https://review.openstack.org/624384
23:08:34 mriedem dansmith: you might have an opinion on the os-vif version side-step in this spec https://review.openstack.org/#/c/607610/ but i'm not sure it's really worth holding up on that
23:10:38 dansmith mriedem: yeah I think I saw talk of that recently
23:11:26 dansmith I'm ambivalent, although amused/bemused that it's been however many years and we still aren't sharing objects with neutron
23:11:56 Kevin_Zheng mriedem: ack
23:15:03 mriedem we didn't do the port binding thing, talked about since ocata, until rocky, so yeah
23:19:54 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Make compute rpcapi version calculation check all cells https://review.openstack.org/625060
#openstack-nova - 2018-12-14
01:06:01 openstackgerrit Merged openstack/nova master: Only construct SchedulerReportClient on first access from API https://review.openstack.org/623246
01:06:07 openstackgerrit Merged openstack/nova master: DRY up SchedulerReportClient init https://review.openstack.org/623247
01:06:14 openstackgerrit Merged openstack/nova stable/rocky: Note the aggregate allocation ratio restriction in scheduler docs https://review.openstack.org/623546
02:54:32 openstackgerrit Dave McCowan proposed openstack/nova stable/rocky: Add secret=true to fixed_key configuration parameter https://review.openstack.org/623507
02:55:09 openstackgerrit Dave McCowan proposed openstack/nova stable/queens: Add secret=true to fixed_key configuration parameter https://review.openstack.org/623509
02:57:07 openstackgerrit Matt Riedemann proposed openstack/nova stable/queens: Don't persist RequestSpec.requested_destination https://review.openstack.org/611945
05:01:37 openstackgerrit Merged openstack/nova stable/queens: Refix disk size during live migration with disk over-commit https://review.openstack.org/602478
05:16:52 openstackgerrit Merged openstack/nova stable/queens: Fix stacktraces with redis caching backend https://review.openstack.org/606896
05:17:05 openstackgerrit Merged openstack/nova stable/queens: Add regression test for bug 1797580 https://review.openstack.org/611944
05:17:05 openstack bug 1797580 in OpenStack Compute (nova) queens "NoValidHost during live migration after cold migrating to a specified host" [High,In progress] https://launchpad.net/bugs/1797580 - Assigned to Matt Riedemann (mriedem)
05:17:10 openstackgerrit Merged openstack/nova stable/queens: Not set instance to ERROR if set_admin_password failed https://review.openstack.org/608179
05:17:16 openstackgerrit Merged openstack/nova stable/queens: Ignore VirtDriverNotReady in _sync_power_states periodic task https://review.openstack.org/607730
06:07:08 izza_ hi...i created instance with VGPU flavor in Openstack TripleO rocky version with Windows Server 2012 R2 guest OS,but as checked, graphics dirver is not visible in the guest OS
06:07:21 izza_ we're using the tesla P100

Earlier   Later