| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-04-19 | |||
| 22:04:24 | dansmith | melwitt: what we need to avoid is returning unversioned data to the compute, which I don't think is solved purely by putting the straddle in quotas.py | |
| 22:04:37 | melwitt | dansmith: I was thinking of the Quotas object as an abstraction but wasn't think of the remotable methods part | |
| 22:04:45 | dansmith | melwitt: what we really need is a composite operation that just says "yo dawg, check this quota limit for me and raise if there's a problem" right? | |
| 22:05:00 | dansmith | melwitt: then all the data is on the conductor side and compute is just asking for a check instead of getting the data itself | |
| 22:05:03 | melwitt | dansmith: it's not but it would be doing the same thing it is now (without having a Quotas object version attached to it) | |
| 22:05:17 | melwitt | dansmith: yeah, that makes sense | |
| 22:05:39 | dansmith | melwitt: right now we remote the call to the quotas db code through the object action without returning data, right? | |
| 22:05:47 | dansmith | and your patch ends up with us returning that data to compute I think | |
| 22:05:54 | melwitt | dansmith: check_deltas is supposed to do that, but atm I'm confused as to where that occurs or if it solves the problem | |
| 22:06:29 | melwitt | check_deltas is the composite operation that calls count followed by limit_check | |
| 22:08:08 | dansmith | melwitt: where is check_deltas? | |
| 22:08:58 | dansmith | that must be something later in your patch series? | |
| 22:09:49 | melwitt | dansmith: third patch I think | |
| 22:10:29 | melwitt | dansmith: https://review.openstack.org/#/c/446239 | |
| 22:11:10 | cfriesen | config option question: when specifying the "live_migration_inbound_addr" config option, is it expected that an ipv6 address will be enclosed in square brackets? Or should nova handle that when parsing the config option? | |
| 22:11:39 | dansmith | melwitt: ah yep.. that just takes resources and counts (from compute in this case), remotes to conductor where it looks at all the stuff and either explodes or doesn't, right? | |
| 22:13:17 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP/PoC: nova.utils.get_service_url(group) https://review.openstack.org/458257 | |
| 22:13:23 | melwitt | dansmith: it takes the deltas but it does the count and pulls the limits itself | |
| 22:13:54 | dansmith | melwitt: right, it just takes a resource and count, and explodes if the quota would be violated | |
| 22:14:09 | dansmith | melwitt: that's what we need yeah | |
| 22:14:34 | clarkb | cfriesen: I thought I could answer based on how we configure things in multinode testing but that appears to be an entirely ipv4 control plane :/ | |
| 22:14:48 | melwitt | dansmith: if by count you mean delta, then yes. if you want to check if you have room for 5 more instances, you pass it 5 and then it will count how many instances your project and user have allocated, then it pulls the limits and compares | |
| 22:15:07 | cfriesen | clarkb: I'm wondering if maybe nova should do: | |
| 22:15:35 | cfriesen | migrate_data.target_connect_addr = utils.safe_ip_format(CONF.libvirt.live_migration_inbound_addr) | |
| 22:15:53 | dansmith | melwitt: yeah | |
| 22:17:21 | melwitt | dansmith: okay, so that solves that problem. so I need to figure out what to do about the db access methods I have in the Quotas object | |
| 22:18:11 | melwitt | I put them there as a universal place to call api/main db to do the right thing, and the caller should only be the quota driver in nova/quota.py | |
| 22:18:29 | melwitt | I think ... | |
| 22:18:56 | dansmith | melwitt: right but does compute call into nova.quota for anythign? I thought it did | |
| 22:19:09 | dansmith | if so, then your check_deltas conversion needs to come first | |
| 22:19:19 | melwitt | dansmith: it does, for the reserve/commit/rollback | |
| 22:19:24 | openstackgerrit | Peter Hamilton proposed openstack/nova master: Add configuration options for certificate validation https://review.openstack.org/457678 | |
| 22:19:48 | dansmith | melwitt: yeah, then you need to do something in a different order I guess | |
| 22:19:50 | melwitt | dansmith: check_deltas only works if the quota has been converted to Countable | |
| 22:19:56 | dansmith | yeah | |
| 22:24:34 | melwitt | I shall think on it | |
| 22:25:03 | melwitt | my brain's too fried at the moment | |
| 22:28:23 | eandersson | Is there a way to extend the data in metadata? (e.g. adding tenant name) | |
| 22:53:45 | mriedem | eandersson: you can update the metadata associated with a server yes | |
| 22:54:09 | mriedem | eandersson: https://developer.openstack.org/api-ref/compute/#server-metadata-servers-metadata | |
| 22:54:42 | mriedem | https://docs.openstack.org/cli-reference/nova.html#nova-meta | |
| 23:04:53 | mriedem | melwitt: did you get a patch up for that ironic unit test failure? | |
| 23:05:20 | melwitt | mriedem: yes, https://review.openstack.org/458161 | |
| 23:05:28 | mriedem | sweet | |
| 23:06:00 | melwitt | I think the only thing it misses out on is running the _wait_state function but I think one of the other unit tests runs it | |
| 23:06:10 | melwitt | double check it though | |
| 23:06:39 | mriedem | the commit message says it's covered in other tests | |
| 23:06:43 | mriedem | i guess you're lying | |
| 23:06:45 | mriedem | potentially | |
| 23:07:42 | mriedem | but yeah there are | |
| 23:07:45 | mriedem | off the hook this time witt | |
| 23:10:48 | openstackgerrit | Mathieu Gagné proposed openstack/nova master: Add ability to signal and perform online volume size change https://review.openstack.org/454322 | |
| 23:10:56 | melwitt | haha | |
| 23:11:57 | melwitt | I thought I saw it covered in other tests but maybe I misunderstood something | |
| 23:12:38 | mriedem | +2 | |
| 23:14:07 | melwitt | mriedem: thanks. at first I removed those overrides but then kept them to use in the assert_called_with to verify they're being passed along to the timer as expected | |
| 23:14:36 | mriedem | alright - that's fine for the configurable one, the module scope variable was just there for tricking the timer | |
| 23:14:46 | melwitt | oh | |
| 23:22:27 | mriedem | mdbooth: do you plan on rebasing this https://review.openstack.org/#/c/430213/ ? - something came into triage today that was a duplicate | |
| 23:23:18 | mriedem | sean-k-mooney: who is running the intel 3rd party ci now that wznoinsk is gone? | |
| 23:39:25 | imacdonn | mriedem, still around ? | |
| 23:41:29 | mriedem | imacdonn: yeah | |
| 23:42:11 | openstack | Launchpad bug 1662483 in OpenStack Compute (nova) "duplicate for #1683972 detach_volume races with delete" [Undecided,In progress] - Assigned to Matthew Booth (mbooth-9) | |
| 23:42:11 | imacdonn | hey .. re. https://bugs.launchpad.net/nova/+bug/1683972 | |
| 23:42:20 | imacdonn | thanks, openstack | |
| 23:42:57 | imacdonn | I'm not sure it's really the same problem as the one you marked it a dup of .. it's slightly related, I guess | |
| 23:43:30 | mriedem | feel free to unduplicate it if you want, but i thought the solution was the same | |
| 23:43:36 | mriedem | lock on detach_volume in the compute manager | |
| 23:44:01 | imacdonn | my locking was way more extensive (and possibly way more wrong, but it works for me) | |
| 23:44:12 | mriedem | care to post your change? | |
| 23:44:18 | mriedem | i'm sure mdbooth would like to look at it | |
| 23:44:24 | imacdonn | but my case was overlap between attach and detach .. not two detach's | |
| 23:44:39 | imacdonn | if mdbooth is the right guy, I can get with him ... what TZ is he in ? | |
| 23:45:08 | mriedem | UK | |
| 23:45:27 | mriedem | so he'd be around in the morning | |
| 23:46:05 | imacdonn | OK, I'll look for him them ... I can post what I did somewhere, but I'm fairly sure it's not "right" | |
| 23:46:27 | mriedem | hard to know if you don't post it | |
| 23:46:40 | mriedem | posting to gerrit would be preferrable, else you could attach a patch file in the bug report | |
| 23:47:44 | imacdonn | I think it's too much of a hack to put in a review .. should I attach to my bug (the one marked as a dup) ? | |
| 23:49:41 | mriedem | nothing is too much of a hack to put it in a review - just put WIP in the prefix of the commit title | |
| 23:49:47 | mriedem | or -Workflow it in gerrit | |
| 23:49:53 | mriedem | no one is going to think less of you :) | |
| 23:50:15 | imacdonn | heh | |
| 23:50:45 | imacdonn | I'll take another look at it first ... don't want to embarrass myself any more than necessary :) | |
| 23:52:49 | imacdonn | I have another thing I wanted to get your opinion on (completely unrelated) | |
| 23:55:36 | mriedem | ok | |
| 23:55:52 | imacdonn | so when nova plugs an instance's vif into a neutron network that has MTU != 1500, something needs to set the MTU on the tap interface. Whose job is that? nova-compute? os_vif? the neutron l2 agent? ... ? | |
| 23:56:02 | imacdonn | asking because it's currently not getting done, and I'm not sure who to blame. | |
| 23:56:18 | imacdonn | For more detail, see https://ask.openstack.org/en/question/105521/whos-supposed-to-set-the-mtu-on-a-nova-instances-vifs-tap-interface/ | |
| 23:56:52 | mriedem | sean-k-mooney: ^ | |
| 23:57:02 | mriedem | imacdonn: in part it depends on the vif type | |
| 23:57:15 | mriedem | the libvirt driver isn't using os-vif yet for all vif types | |
| 23:57:35 | imacdonn | it is in my case .. vif_type is bridge | |
| 23:58:08 | mriedem | then i believe os-vif does it | |
| 23:58:28 | imacdonn | I think I have the terminology right there ... in any case, vif_plug_linux_bridge under os_vif creates the bridge | |
| 23:58:52 | imacdonn | then libvirt creates the domain, which creates the tap interface, and puts it on the bridge | |
| 23:58:57 | mriedem | does the network in neutron have the mtu attribute set? | |
| 23:59:06 | imacdonn | yes | |
| 23:59:40 | imacdonn | I've studied the code, and I can't find anything that's supposed to do the job | |
| 23:59:56 | imacdonn | I don't think os_vif can do it, because the tap interface doesn't exist yet | |