Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-26
15:50:38 mriedem bauzas: by default the tests won't have anything set in the CONF.placement values
15:50:42 mriedem unless we have defaults
15:50:45 openstackgerrit Chris Dent proposed openstack/nova: Add more debug logging on RP inventory update failures https://review.openstack.org/414230
15:50:53 mriedem mfeoktistov: he moved some small things around it looks like
15:51:02 mriedem mdbooth: ^
15:51:09 mdbooth IIRC I was +1 on this before, so if they only thing you've change is privsep I'm still +1.
15:51:11 cdent mriedem: that's ^ the long promised test additions to that logging thing you did a while back. with some adjustments
15:51:17 andreykurilin mriedem: only one(+one not critical in the chain) left - https://review.openstack.org/#/c/424256/ . It has one +2 and answered all comments
15:51:58 bauzas mriedem: dansmith: the only problem I see is that we expect keystone config options
15:52:05 mriedem andreykurilin: ok, looks like johnthetubaguy has been reviewing https://review.openstack.org/#/c/424256/ so i'll leave the +W to him
15:52:10 bauzas so I need to verify them
15:52:18 melwitt mriedem: I'm pretty confident in the counting stuff. I think it's no worse than what's already there. we would trade known issues and out-of-sync problems for something with a lot smaller footprint, but obviously has some element of the unknown compared to something we've had for years
15:53:19 mfeoktistov mdbooth: no, just rootwrap instead of privsep, move two lines of code from libvirt_driver to libvirt_utils and fix tests for it
15:53:23 mriedem bauzas: i think you just need to check for CONF.placement.password
15:53:26 mriedem that *has* to be set
15:53:38 bauzas mriedem: okay, I can just do that
15:53:42 melwitt mriedem: as for the bottom patches, if we assume a single cell, I think it still works, it's just reservations and usages would be in the main db whereas the limits and classes would be in the api db.
15:53:54 mriedem http://logs.openstack.org/72/404472/20/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/8aafdf2/logs/etc/nova/nova.conf.txt.gz
15:54:02 mriedem bauzas: ^ for what we set in devstack
15:54:21 mriedem but there are some deprecated things in ksa so i'd just stick with checking for the password option
15:54:22 bauzas mriedem: yup, I know
15:54:29 mriedem like user_id vs username, etc
15:54:38 bauzas mriedem: I'm just hesitating about checking things related to keystone diretly
15:54:43 mriedem and the 10 different ways you can specify a project in ksa
15:54:50 mfeoktistov mdbooth: code which gets default filesystem type from nova conf or ext4 as default value
15:54:59 bauzas mriedem: given we don't verify that by us but instead straight passing it to keystoneauth
15:55:08 mdbooth mfeoktistov: Yep, spotted
15:55:17 mriedem bauzas: i don't know what that means, or why we care
15:55:22 bauzas anyway, let's not overcomplicate things
15:55:23 cdent sfinucan: on https://review.openstack.org/#/c/404990/ there's this to fix that regex (next cycle): https://review.openstack.org/#/c/418393/
15:55:32 mriedem bauzas: init_host just checks, if CONF.placement.password is None: raise Exception
15:55:33 bauzas my patch is almost ready and I have to leave in 5
15:55:36 dansmith mriedem: on the quotas thing, I think it's risky, but I also think it's more likely that anyone else will be able to fix bugs in the counting stuff, so if you want to push, I'll dig into that last one
15:56:02 sfinucan cdent: Aha - thank you. That did seem like a peculiar oversight
15:56:06 dansmith mriedem: and I think there's probably truth to the argument that it's no worse than what we have now, but it is more obvious than what we have now
15:56:11 mdbooth pkoniszewski pkoniszewski_ (irc is the worst): Are you able to determine what I'm talking about through gerrit's formatting?
15:56:16 dansmith mriedem: but also, we're merging other stuff with high risk, so...
15:56:26 cdent sfinucan: it's because 409 is overloaded for at least two different types of conflict
15:56:31 mriedem dansmith: heh, so while we're playing with house money...
15:56:42 mriedem dansmith: feel free to push, i just won't be able to get to it
15:56:59 pkoniszewski_ mdbooth: but the point is that there is no such race
15:57:08 dansmith mriedem: I meant "maybe we shouldn't take out another loan" but okay
15:57:25 mdbooth pkoniszewski_: Explain
15:57:29 cdent it's called leverage
15:57:41 mriedem dansmith: oh i thought you wanted to go balls out
15:57:56 mdbooth pkoniszewski_: I described the race. Can you point out the step I got wrong?
15:58:13 dansmith mriedem: I meant the last statement was a hedge, not a "we've already got a second mortgage, why not go for a third"
15:58:37 dansmith mriedem: let me look through it and see how I feel after that
15:58:43 mriedem hey, the dow was over 20K so i think that's a safe bet right now
15:58:43 mdbooth pkoniszewski_: I think cfriesen(?) said they'd hit it in practise, too.
15:58:46 mriedem what could possibly go wrong
15:58:48 dansmith heh
15:58:58 mfeoktistov mriedem: thanks for discussion in mailing list
15:59:13 pkoniszewski_ mdbooth: live migration will always be set to 'completed', there is no chance we will end up with 'failed' or 'finished' state
15:59:26 dansmith johnthetubaguy: are you still around?
15:59:28 pkoniszewski_ mdbooth: the RPC has been changed to call
15:59:31 pkoniszewski_ mdbooth: https://review.openstack.org/#/c/389687/
15:59:35 johnthetubaguy dansmith: I am
15:59:48 bauzas there it is
16:00:02 dansmith johnthetubaguy: how are you feeling about the counting quotas stuff?
16:00:06 openstackgerrit Sylvain Bauza proposed openstack/nova: Scheduler calling the Placement API https://review.openstack.org/417961
16:00:40 mdbooth pkoniszewski_: So it has, I must have been looking at an old version when looking at the old commit.
16:00:54 mdbooth Well... for the reasons stated I also think that's a bad idea :)
16:01:02 mdbooth But yeah, that means there's no race.
16:01:15 pkoniszewski_ mdbooth: but there is still an issue with rollback, thats a very good point
16:02:30 melwitt mriedem, dansmith, johnthetubaguy: one thing I realized about the counting quotas approach is that once we go multicell, if we lose contact with a cell db for whatever reason, we lose the ability to count resources in it, which would basically free up a person's resource usage for the time period the cell db is inaccessible
16:02:35 sfinucan mfeoktistov: +W'd
16:02:43 sfinucan also, worst IRC nick ever :)
16:02:54 sfinucan ...and mine is bad
16:03:47 dansmith melwitt: yeah, that would be bad..
16:05:05 dansmith melwitt: that would be for what kind of resources?
16:05:18 dansmith melwitt: we can count instances and server groups and keypairs all in the api db
16:05:30 mdbooth pkoniszewski_: Incidentally, you could add a comment to the 'completed' code that it can be safely deleted in Queens, because we can guarantee the destination will have already updated migration status.
16:05:41 melwitt dansmith: it seems like we'd need to have some sort of caching table for the usage that gets updated anytime we actually count
16:06:10 melwitt dansmith: currently, I think instances, cores, ram. but that's because I rolled the instance tally into the cores and ram count
16:06:32 dansmith melwitt: that kinda defeats the purpose,in that it becomes a thing that can get out of sync and that we can't refresh without being able to contact the other cell
16:06:45 dansmith melwitt: okay well we could count instances in the api db to mitigate that
16:07:06 johnthetubaguy melwitt: maybe we hard fail on that
16:07:29 dansmith melwitt: so maybe we should get a read from operator(s) about how bad it would be be to not be if ram/cores leaks while a whole cell is offline
16:07:37 johnthetubaguy melwitt: if we know something is there
16:07:44 dansmith johnthetubaguy: hard fail to create instances anywhere if one cell is offline? that'd be bad :)
16:07:52 pkoniszewski_ mdbooth: AFAIR nova does not assume that there will only be live migration from N-1 to N (N-2 to N must also be possible and safe)
16:07:58 melwitt dansmith: yeah. I haven't thought about it much but it seemed like it makes sense for it to be stale if the cell is gone. but if it went away permanently, then yeah that would be the refresh/sync case. so nevermind that idea
16:08:29 mdbooth pkoniszewski_: I understood the contract was that N and N-2 never exist in the same deployment, right?
16:08:34 melwitt dansmith, johnthetubaguy: yeah, we could fall back to InstanceMappings for the instance count but cores/ram would be unknown
16:08:37 dansmith melwitt: well, no I meant even if it's offline temporarily, you wouldn't be able to refresh that value from N-1 cells because you don't know how much of it is in the offline cell, unless you keep more data
16:08:41 mdbooth So no migration between them is implicit.
16:08:45 mdbooth dansmith: ^^^ ?
16:08:55 dansmith mdbooth: correct
16:09:20 bauzas mriedem: dansmith: cdent: rev'd https://review.openstack.org/417961
16:09:38 melwitt dansmith: oh, right
16:09:40 dansmith mdbooth: pkoniszewski you shouldn't break your RPC api between N-2 and N, but we also don't support that kind of deployment, so higher level assumptions might not be possible between those two
16:09:48 cdent bauzas: cool, will look asap, currently double booked in two meetings ...
16:10:14 mriedem bauzas: ok. i think we should move the compute/service version checking stuff into a separate patch for sanity
16:10:21 mriedem bauzas: but sounds like you're going to be afk for awhile
16:10:22 dansmith bauzas: that service version and fail should be a separate patch
16:10:25 dansmith what mriedem said
16:10:50 bauzas mriedem: yup, I could potentially lightweight the change
16:10:56 pkoniszewski_ dansmith: got it
16:11:20 melwitt dansmith: actually I guess the long term answer was we're going to ask for cores/ram count from placement when that's doable

Earlier   Later