Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-12-09
17:41:25 openstackgerrit Matt Riedemann proposed openstack/nova master: Deal with cross-cell resize in _remove_deleted_instances_allocations https://review.opendev.org/639453
17:41:26 openstackgerrit Matt Riedemann proposed openstack/nova master: Flesh out RevertResizeTask.rollback https://review.opendev.org/695334
17:41:26 openstackgerrit Matt Riedemann proposed openstack/nova master: Add RevertResizeTask https://review.opendev.org/638046
17:41:27 openstackgerrit Matt Riedemann proposed openstack/nova master: Revert cross-cell resize from the API https://review.opendev.org/638048
17:41:27 openstackgerrit Matt Riedemann proposed openstack/nova master: Add revert_snapshot_based_resize conductor RPC method https://review.opendev.org/638047
17:41:28 openstackgerrit Matt Riedemann proposed openstack/nova master: Confirm cross-cell resize while deleting a server https://review.opendev.org/638268
17:41:28 openstackgerrit Matt Riedemann proposed openstack/nova master: Add functional cross-cell revert test with detached volume https://review.opendev.org/695335
17:41:29 openstackgerrit Matt Riedemann proposed openstack/nova master: Add CrossCellWeigher https://review.opendev.org/614353
17:41:29 openstackgerrit Matt Riedemann proposed openstack/nova master: Add archive_deleted_rows wrinkle to cross-cell functional test https://review.opendev.org/651650
17:41:30 openstackgerrit Matt Riedemann proposed openstack/nova master: Add functional test for anti-affinity cross-cell migration https://review.opendev.org/661859
17:41:30 openstackgerrit Matt Riedemann proposed openstack/nova master: Add test_resize_cross_cell_weigher_filtered_to_target_cell_by_spec https://review.opendev.org/695336
17:41:31 openstackgerrit Matt Riedemann proposed openstack/nova master: Support cross-cell moves in external_instance_event https://review.opendev.org/658478
17:46:53 artom stephenfin, thanks!
17:47:00 openstackgerrit Vladyslav Drok proposed openstack/nova master: Minor improvements to cell commands https://review.opendev.org/698053
17:52:53 openstackgerrit Eric Fried proposed openstack/nova master: Nix os-server-external-events 404 condition https://review.opendev.org/698037
17:53:19 efried gmann, mriedem: done ^
18:05:19 efried stephenfin: are you working your way up to https://review.opendev.org/#/c/696992/ eventually? (aka: "would you please, and thank you?")
18:09:17 openstackgerrit sean mooney proposed openstack/nova master: Disable NUMATopologyFilter on rebuild https://review.opendev.org/689861
18:09:34 openstackgerrit sean mooney proposed openstack/nova master: support pci numa affinity policies in flavor and image https://review.opendev.org/674072
18:10:37 sean-k-mooney efried: do you have time this week to review https://review.opendev.org/#/c/687957/11 and the two patches that follow. i would like to get those merged before i go on PTO
18:10:53 efried is that the numa affinity thing?
18:11:30 sean-k-mooney the frist two are rebuild of numa instance and the last is the numa affinity policy thing
18:12:20 sean-k-mooney so those 3 patch fix 2 bugs and close two blueprints/specs
18:12:47 openstackgerrit Merged openstack/nova master: Fup for I63c1109dcdb9132cdbc41010654c5fdb31a4fe31 https://review.opendev.org/697678
18:13:20 sean-k-mooney efried: also i think you just missed stephenfin. he was just heading home for the day
18:13:32 openstackgerrit Eric Fried proposed openstack/nova master: FUP: Remove noqa and tone down an exception https://review.opendev.org/698054
18:14:17 efried ack.
18:14:23 efried I'll look at those now sean-k-mooney...
18:24:56 openstackgerrit Mykola Yakovliev proposed openstack/nova master: Fix boot_roles in InstanceSystemMetadata https://review.opendev.org/698040
18:25:59 mnaser has anyone seen any race conditions in starting up instances in stein? :X
18:26:33 openstackgerrit Mykola Yakovliev proposed openstack/nova master: Fix boot_roles in InstanceSystemMetadata https://review.opendev.org/698040
18:27:19 mnaser https://www.irccloud.com/pastebin/8hYnM4Lx/
18:27:40 mnaser i mean this is a grep on the instance uuid but the instance destroyed successfully twice 15 seconds apart?
18:36:58 sean-k-mooney it could be an interaction with the periodic task and the actul instnace action that tried to showdown the instance
18:37:23 sean-k-mooney mnaser: but no i have not seen that specifically
18:37:53 mnaser sean-k-mooney: well instance destroyed successfully is called by `_wait_for_destroy`
18:38:02 mnaser https://github.com/openstack/nova/blob/stable/stein/nova/virt/libvirt/driver.py#L1023-L1025
18:38:23 mnaser which has the timer here https://github.com/openstack/nova/blob/stable/stein/nova/virt/libvirt/driver.py#L1038-L1039
18:38:55 mnaser which is `_destroy()`
18:39:45 mnaser hmm, we dont have a 'powering-off' state in nova do we
18:40:19 mnaser we do
18:41:27 mnaser task state gets updated inside nova.compute.api -- does that mean that technically there could be 2 concurrent api requests?
18:42:06 sean-k-mooney you can have two concurrent request to destroy yes at the api level
18:42:27 sean-k-mooney but the window would be quite short
18:42:27 mnaser so i guess in that case if two api requests came at the same time, we can end up with two requests in the compute level
18:42:52 sean-k-mooney yes
18:43:32 sean-k-mooney although we normaly do a check of the taskstate wehn ever we updat it so one would fail
18:49:57 mordred mriedem, efried: sdk release was cut including your ironic fix
18:51:05 efried ack, thx
18:51:08 mnaser sean-k-mooney: _record_action_start is called, ill check logs
18:54:09 efried sean-k-mooney: theoretically...
18:54:17 efried if a rebuild is done with the same image...
18:54:36 efried couldn't I have changed the metadata on that image since my instance was originally booted?
18:57:21 mnaser sean-k-mooney: sigh, i wonder if we call _destroy before starting up an instance which is why it says "Instance destroyed successfully."
18:58:02 mnaser sean-k-mooney: yes thats exactly what happens, we call destroy for power on because we call hard reboot
19:05:12 dustinc [1] https://blueprints.launchpad.net/nova/+spec/provider-config-file
19:05:12 dustinc disable the entire feature by default instead
19:05:12 dustinc dansmith: I was working on docs/etc for [1] and was thinking that it might be a good idea to have a master on/off switch for the feature with a default of off...the reason I was thinking that is because as it is specced right now, the directory is the only config option and a cautious operator would probably want to make sure the default directory is present with permissions set..it would maybe be easier to just
19:05:16 dustinc what do you think?
19:05:56 dansmith dustinc: how is a conf knob turned off different from the directory/file(s) being missing?
19:06:37 dustinc in theory if not present someone else could create the directory and place files
19:07:18 efried I was also thinking maybe instead of having the directory conf option default to something sane, have it default to None, and that's the "off" switch. You have to turn it on explicitly by setting it to something.
19:07:23 dansmith dustinc: in /etc/nova? not anyone without sufficient privileges
19:09:47 dustinc dansmith: if that's sufficient then maybe leave it as it is then..
19:11:47 dustinc thanks
19:14:32 mnaser sean-k-mooney: off the top of your head? i could swear there was a nova option which skipped waiting for port plugs when starting up an instance
19:14:37 mnaser am i imagining things
19:15:49 dansmith mnaser: vif_plugging_timeout=0 and vif_plugging_fatal=False
19:15:58 dansmith mnaser: but don't do it except for debugging
19:16:01 mnaser dansmith: ah yes, ok that makes sense
19:16:27 mnaser dansmith: yep, i don't want to either, we're just trying to figure out why rdocloud can bring instances up so quickly (yet this new cloud is taking so long the tripleo-ci b its are timing out)
19:16:54 mnaser so given its taking ~1-2 minutes to turn up an instance, because a lot of it is unplugging (in cleanup) and plugging (in startup)
19:17:02 dustinc dansmith: dont_be_slow=true
19:36:16 openstackgerrit Alexandre arents proposed openstack/nova stable/rocky: Do not update root_device_name during guest config https://review.opendev.org/696353
20:21:55 sean-k-mooney mnaser: i see dansmith alreay answered your question. are the new ci and rdocloud both using hte same netwrok backend.
20:22:13 mnaser sean-k-mooney: afaik that runs ovs and so do we, dont have too much details except ocata vs stein
20:22:37 sean-k-mooney i think rdocloud is running ml2/ovs too
20:23:44 sean-k-mooney efried: if a rebuild is done with the same iamge we do not go to the schduler and use the cached copy of the image mentadata
20:24:08 sean-k-mooney efried: we only go to the schduler if the image changes
20:34:43 efried interesting
20:38:45 openstackgerrit Matt Riedemann proposed openstack/nova master: Add sequence diagrams for cross-cell-resize https://review.opendev.org/698051
20:38:45 openstackgerrit Matt Riedemann proposed openstack/nova master: Flesh out docs for cross-cell resize/cold migrate https://review.opendev.org/696212
20:38:46 openstackgerrit Matt Riedemann proposed openstack/nova master: Implement cleanup_instance_network_on_host for neutron API https://review.opendev.org/697162
20:38:46 openstackgerrit Matt Riedemann proposed openstack/nova master: Simplify FinishResizeAtDestTask event handling https://review.opendev.org/695337
20:38:47 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Implement reschedule logic for cross-cell resize/migrate https://review.opendev.org/696213
20:38:47 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Add negative test to delete server during cross-cell resize claim https://review.opendev.org/688832
20:38:50 sean-k-mooney resource usage is not ment to change on a rebuild so since the flavor does not change and the image does not change you dont need to check.
20:40:19 mriedem except for silly ol numa
20:40:27 sean-k-mooney yep
20:40:30 mriedem which is why eric finds himself in this situation
20:41:20 sean-k-mooney well even in the numa case if you dont change the image then your fine but ya people want to do that for some reason :)
20:41:49 mordred changing the image doesn't seem like a rebuild to me - it seems like creating a new server
20:41:55 mordred but - you know - what do I know?
20:42:54 sean-k-mooney :) well we are stuck with the terms we have
20:43:11 sean-k-mooney atleast its less confusing then evacuate
21:36:09 openstackgerrit Mykola Yakovliev proposed openstack/nova master: Validate aggregate IDs before querying database https://review.opendev.org/698094
21:38:01 openstackgerrit Merged openstack/nova master: Handle ServiceNotFound in DbDriver._report_state https://review.opendev.org/697301
22:08:46 openstackgerrit Merged openstack/nova master: VMware: disk_io_limits settings are not reflected when resize https://review.opendev.org/680296
22:21:10 openstack bug 1829479 in OpenStack Compute (nova) "The allocation table has residual records when instance is evacuated and the source physical node is removed" [Medium,In progress] https://launchpad.net/bugs/1829479 - Assigned to Matt Riedemann (mriedem)
22:21:10 openstackgerrit Merged openstack/nova stable/stein: Add functional recreate test for bug 1829479 and bug 1817833 https://review.opendev.org/695932

Earlier   Later