| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-08-31 | |||
| 21:12:49 | mriedem | note that the server details don't show a host or node when it's shelved | |
| 21:13:00 | cfriesen | mriedem: sure, but the AZ was selected at boot time as part of the boot criteria | |
| 21:13:00 | mriedem | if it's not on a host, it's clearly not in an AZ | |
| 21:13:06 | mriedem | nope | |
| 21:13:13 | mriedem | i didn't create the server with an az | |
| 21:13:23 | mriedem | so when i unshelve it, it can go into any othre az | |
| 21:14:02 | mriedem | related https://review.openstack.org/#/c/567701/ | |
| 21:15:10 | cfriesen | I had thought that once we pick an AZ at boot it might have implications on which cinder backends are available, so you couldn't leave it afterwards | |
| 21:15:23 | mriedem | oh fuck no | |
| 21:15:36 | mriedem | related https://review.openstack.org/#/c/469675/ | |
| 21:15:37 | cfriesen | cool, learned something new | |
| 21:15:59 | mriedem | we only give a shit about volume az if cinder.cross_az_attach=False, which it's not by default | |
| 21:16:09 | mriedem | if it is, well, good luck friend | |
| 21:16:12 | mriedem | b/c nova sucks at handling that | |
| 21:16:39 | mriedem | https://review.openstack.org/#/c/467674/ will show you exactly why | |
| 21:17:02 | mriedem | if you remove that nova fix as a dependency, you can't create a server with a pre-existing volume at all | |
| 21:17:08 | mriedem | because 'nova' != None | |
| 21:17:11 | mriedem | KABLAMMO | |
| 21:18:40 | mriedem | granted, my patch for that cross_az_attach fix is fugly | |
| 22:20:11 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Null out instance.availability_zone on shelve offload https://review.openstack.org/599087 | |
| #openstack-nova - 2018-09-01 | |||
| 00:12:07 | openstackgerrit | Merged openstack/nova master: Fix image-defined numa claims during evacuate https://review.openstack.org/588657 | |
| 00:37:43 | openstackgerrit | Merged openstack/nova master: Move conductor wait_until_ready() delay before manager init https://review.openstack.org/598353 | |
| 21:25:25 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Move conductor wait_until_ready() delay before manager init https://review.openstack.org/599200 | |
| #openstack-nova - 2018-09-02 | |||
| 00:35:10 | openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: implement reshaper for vgpu https://review.openstack.org/599208 | |
| 00:35:15 | mriedem | fried_rice: bauzas: welcome to reshape city, population: libvirt ^ | |
| 14:35:43 | openstackgerrit | Matt Riedemann proposed openstack/nova master: libvirt: implement reshaper for vgpu https://review.openstack.org/599208 | |
| #openstack-nova - 2018-09-03 | |||
| 00:16:08 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Fix a failure to format config sample https://review.openstack.org/597986 | |
| 02:57:35 | openstackgerrit | Brin Zhang proposed openstack/nova master: Need further updates, no need to review https://review.openstack.org/599276 | |
| 03:33:16 | openstackgerrit | Chen proposed openstack/nova master: Fix SOFT_DELETE vm_state string https://review.openstack.org/598482 | |
| 06:29:48 | prometheanfire | https://github.com/openstack/nova/commit/50658eee4fe26a55854642119beeff46c0d0108a seems to be causing problems | |
| 06:30:00 | prometheanfire | 2018-09-03 01:29:14.620 11567 ERROR nova.objects.cell_mapping [req-0b912c05-d2bf-4e68-bcdf-7953c87dcc88 - - - - -] Failed to parse [database]/connection to format cell mapping: ValueError: expected ':' after conversion specifier | |
| 06:30:10 | prometheanfire | tonyb: think I found another one ^ | |
| 06:30:44 | prometheanfire | not sure what's wrong with my db | |
| 06:30:52 | prometheanfire | I'm guessing my password isn't being escaped | |
| 06:32:04 | tonyb | prometheanfire: can you run nova in a venv and import cell_mapping.py and check the parse results with your netloc? | |
| 06:32:23 | prometheanfire | maybe the root? | |
| 06:32:23 | prometheanfire | 2018-09-03 01:29:14.919 11567 ERROR oslo_db.sqlalchemy.exc_filters [req-0b912c05-d2bf-4e68-bcdf-7953c87dcc88 - - - - -] DBAPIError exception wrapped from (psycopg2.ProgrammingError) relation "projects" does not exist | |
| 06:32:36 | prometheanfire | don't have a venv atm | |
| 06:33:02 | tonyb | prometheanfire: venvs are cheap, but if you hav nova installed as a site package you can just use that | |
| 06:33:02 | prometheanfire | another thing to blame not testing on postgres? | |
| 06:33:04 | prometheanfire | ya | |
| 06:33:25 | tonyb | prometheanfire: No I doubt that a postgres thing | |
| 06:34:58 | prometheanfire | my password actually is fine for the one in question | |
| 06:35:04 | prometheanfire | so probably not that | |
| 06:37:37 | tonyb | prometheanfire: :( Well I guess double check you're pointing at the right DB | |
| 06:39:57 | prometheanfire | it worked before the upgrade | |
| 06:43:34 | prometheanfire | not seeing anything in upgrade nodes or deprication notes or https://docs.openstack.org/nova/rocky/user/upgrade.html denoting any change needed | |
| 06:43:36 | tonyb | prometheanfire: ... Okay but that doesn't mean it wasn't using a deprecated feature that has gone between queens and rocky | |
| 06:43:45 | tonyb | prometheanfire: Okay | |
| 06:57:59 | prometheanfire | problem seems to be with this | |
| 06:58:00 | prometheanfire | https://github.com/openstack/nova/commit/50658eee4fe26a55854642119beeff46c0d0108a#diff-61f32a08cd0554ad482db124dcdc546aR128 | |
| 06:58:04 | prometheanfire | the url.format is failing | |
| 06:59:28 | prometheanfire | https://gist.githubusercontent.com/prometheanfire/c98816d477331bcc0a0272adb93cc5e2/raw/c746f1b1780f1202286b9902178ff4e613329f44/gistfile1.txt | |
| 07:25:16 | prometheanfire | tonyb: it IS my password | |
| 07:51:59 | openstackgerrit | Zhenyu Zheng proposed openstack/nova-specs master: Make scheduling weight more granular https://review.openstack.org/599308 | |
| 08:02:27 | openstackgerrit | Zhenyu Zheng proposed openstack/nova-specs master: Make scheduling weight more granular https://review.openstack.org/599308 | |
| 08:14:09 | prometheanfire | ok, the nova_api database has a non-escaped database_connection in the cell_mappings table | |
| 08:15:24 | prometheanfire | which fixed THAT error at least | |
| 08:38:47 | bauzas | good morning stackers | |
| 08:57:42 | openstackgerrit | Matthew Thode proposed openstack/nova stable/rocky: Don't use '_TransactionContextManager._async' https://review.openstack.org/597421 | |
| 09:11:58 | openstackgerrit | Chen proposed openstack/nova master: Fix SOFT_DELETE vm_state string https://review.openstack.org/598482 | |
| 09:23:57 | prometheanfire | this isn't helping, for some reason re-running the service isn't using the file I just changed, wtf | |
| 09:24:36 | prometheanfire | https://gist.githubusercontent.com/prometheanfire/76b31b592b1780a72b32c289ab1bfb06/raw/b58c80c3d22860397781c53915d7350d6dbe0bab/gistfile1.txt | |
| 09:25:00 | prometheanfire | but async_ isn't defined at all on that line anymore, I editited it to be async | |
| 09:25:17 | prometheanfire | anyone have any idea where that's cached? | |
| 09:28:01 | prometheanfire | I've removed any pycache files I could find | |
| 09:30:38 | openstackgerrit | Merged openstack/nova master: libvirt: guest: introduce blockStats instead of domain.blockStats https://review.openstack.org/526833 | |
| 09:55:11 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Making instance listing skipping down cells configurable https://review.openstack.org/592428 | |
| 10:00:15 | openstackgerrit | Sujeet Banerjee proposed openstack/nova master: WIP: Added a periodic task for network management. https://review.openstack.org/599338 | |
| 10:15:14 | prometheanfire | well, it looks like https://review.openstack.org/597421 may not be the right fix | |
| 11:23:28 | gibi | o/ | |
| 11:27:09 | sean-k-mooney | gibi: o/ | |
| 11:30:10 | openstackgerrit | Merged openstack/nova master: hardware: fix memory check usage for small/large pages https://review.openstack.org/532168 | |
| 12:20:40 | amarao | How to access object attributes in policy json? I'd like to forbid some operations based on flavor. Is someone tried that before? | |
| 12:26:29 | sean-k-mooney | amarao: as in based on the flavor of an instance you want to change what api actions are allowable and or by whom? | |
| 12:27:07 | sean-k-mooney | amarao: that sound like a nightmare from an interoperablity stand point so i doubt that has been tried/tested before | |
| 12:30:13 | amarao | Yep, I want to use flavor of the instance (or another instance attribute). | |
| 12:30:42 | amarao | Basically, I want to lock instance. I can use nova lock for this, but user still have access to console, and I want to remove it. | |
| 12:34:57 | moshele | sean-k-mooney: hi | |
| 12:35:47 | moshele | sean-k-mooney: in intel driver if spoofchk is enabled does it allow to set the mac address of the vf to 00:00:00:00:00:00 | |
| 12:58:51 | sean-k-mooney | moshele: i am not sure. if it did i would expect that to mean if the dest/source mac is not 00:00:00:00:00:00 specically drop the packet e.g. only broadcast traffic would be allowed | |
| 13:02:03 | sean-k-mooney | moshele: also at least in niantic (x520/x510) cards i know the mac filter is before the vlan filter so if you had 2 vf with that adress that would cause issues | |
| 13:03:27 | sean-k-mooney | amarao: locking in nova is intended to prevent api acess to an instance not to prevent users form using it | |
| 13:04:27 | sean-k-mooney | amarao: so a user can still poweroff a locked instance by sshing in and executing power off if they choose too. | |
| 13:05:23 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Making instance/migration listing skipping down cells configurable https://review.openstack.org/592428 | |
| 13:05:41 | sean-k-mooney | amarao: is there a partical usecase you have in mind. perhaps a dedicated feature would be better then a flavor hack. | |
| 13:06:04 | amarao | sean-k-mooney, in our scenario we want to run it with our own key (user have no private key from it). 'Hidden' key plus 'lock' almost lock instance from user. What is left is console: Ctrl-Alt-Del, Sysrq-b, and (most importantly) grub access at boot time. | |
| 13:06:18 | amarao | I'd like to take away console from user as well (for specific instances only). | |
| 13:10:26 | moshele | sean-k-mooney: the reason I am asking is that mellanox driver block this and it can cause issues. So the mac address 00:00:00:00:00:00 is just to allow libvirt to clean vf when the VM with VF is deleted see https://www.redhat.com/archives/libvir-list/2017-March/msg00455.html | |
| 13:11:24 | moshele | sean-k-mooney: but now if spoofchk enabled first nova(libvirt delete the vm) restoring the mac to 00:00:00:00:00:00 and then the sriov agent will disable the spofchk | |
| 13:12:01 | moshele | sean-k-mooney: so the vf mac will not set to 00:00:00:00:00:00 and we can't reuse the mac on the same compute | |
| 13:12:44 | moshele | sean-k-mooney: I wonder how intel driver behavior to undestand were to fix it, in the driver or in openstack ... | |
| 13:14:26 | sean-k-mooney | moshele: i really need to get a redhat account at somepoint.. i cant read that linke but anyway. i dont think setting 00:00:00:00:00:00 is a good idea on intel cards either | |
| 13:14:52 | sean-k-mooney | moshele: i belive that the intel driver generates unique macs automaticaly for the pfs | |
| 13:14:57 | sean-k-mooney | * vfs | |
| 13:16:17 | sean-k-mooney | personally the fact that libvirt is seting 00:00:00:00:00:00 on deleteing the vm sound like a libvirt bug to me | |
| 13:16:40 | sean-k-mooney | or a nova bug if we are doing that in the libvirt vir driver | |
| 13:18:30 | moshele | sean-k-mooney: basically it set the restore the previously used mac so if you auto generate it. it will restore the auto generate one. in mellanox case the vf is start with 00:00:00:00:00:00 so we allowed to restore to 00:00:00:00:00:00 | |
| 13:18:37 | sean-k-mooney | amarao: well would a flavor extra-spec for disableing the console on an instace work? | |
| 13:19:53 | amarao | sean-k-mooney are there such property for extra-spec? | |