| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-01-28 | |||
| 14:08:00 | sean-k-mooney | gibi: given python is garbage collected any memory leask is likely form interaction with libvirt right | |
| 14:08:17 | sean-k-mooney | e.g. the proxy objects | |
| 14:08:33 | sean-k-mooney | or well interaction with external things via real threads | |
| 14:08:50 | gibi | sean-k-mooney: or thing the code actually stores and accumlates for no reason | |
| 14:09:05 | sean-k-mooney | i know we can have meemory issue if we return excptions instead of raising them too on python 3 | |
| 14:09:06 | sean-k-mooney | *2 | |
| 14:09:22 | gibi | like this https://github.com/openstack/oslo.messaging/blob/00d15eaeaba0ded0330cdcec7b19eee3adbfb1e1/oslo_messaging/_drivers/amqpdriver.py#L426 | |
| 14:09:42 | sean-k-mooney | gibi: ah well that is less a leak but rather using more memory then it need to due to a poor algorithim | |
| 14:10:13 | gibi | yepp | |
| 14:10:30 | gibi | actually I can hit that log in pike devstack by poking rabbitmq | |
| 14:10:37 | sean-k-mooney | oh in this case it also qould be an rabbitissue | |
| 14:11:21 | sean-k-mooney | ya that could grow if the reply are lost right | |
| 14:12:27 | gibi | I see two types of increase one that eventually recovers. after a long timeout the disconnect is logged in nova side and the connection is removed | |
| 14:12:28 | sean-k-mooney | back to debuging why an osp env that is freshly deployed passes tempest but then after 9+ hours apparenlty stops working | |
| 14:12:54 | gibi | but there are connections that does not freed even after an hour | |
| 14:13:29 | sean-k-mooney | that sound like an amqp bug | |
| 14:13:52 | sean-k-mooney | like the clinet has gone away but for some reason it keeps the file desctprot for the connection open | |
| 14:14:00 | sean-k-mooney | and never recognises the disconnect | |
| 14:14:05 | gibi | could be | |
| 14:14:12 | sean-k-mooney | althoug | |
| 14:14:16 | gibi | now I have to proove that this is what happens in the customer env | |
| 14:14:27 | gibi | as it is still in my local devstack | |
| 14:14:44 | sean-k-mooney | ya not sure how to determin that | |
| 14:15:20 | gibi | I bet on the warning log from oslo | |
| 14:15:28 | gibi | If the customer sees that then I have a lead | |
| 14:27:59 | lyarwood | ~. | |
| 15:02:03 | stephenfin | lyarwood: comments left on one, +2 on the other | |
| 15:06:09 | lyarwood | kashyap: libvirt is always starts with a lower case l right? | |
| 15:06:20 | lyarwood | kashyap: I'm sure this came up in the past and has confused me for ages | |
| 15:07:24 | kashyap | lyarwood: On a call; bbiab | |
| 15:07:29 | sean-k-mooney | i have seen both | |
| 15:07:30 | kashyap | But yes, lower case | |
| 15:07:31 | lyarwood | np | |
| 15:07:36 | sean-k-mooney | but its normally lowercase | |
| 15:07:57 | lyarwood | yeah this came up before and I was told always lowercase for $reasons | |
| 15:08:45 | sean-k-mooney | if its in a commit mesage or specs then i always use libvirt instead of Libvirt | |
| 15:08:51 | sean-k-mooney | same for release notes | |
| 15:11:03 | sean-k-mooney | although funally enough it's uppercase here https://github.com/libvirt/libvirt/blob/30703564c2ac8d95279801a821cf5510fa4b8149/docs/ci.rst#libvirt-continuous-integration but i think thats wrong | |
| 15:11:56 | sean-k-mooney | also a few places in the readme too https://github.com/libvirt/libvirt/blob/30703564c2ac8d95279801a821cf5510fa4b8149/README.rst | |
| 15:12:36 | sean-k-mooney | but again that feel more like there editor auto capitalising then intentional as the use lowercase when its not the start of a sentence | |
| 15:16:59 | openstackgerrit | Ghanshyam proposed openstack/nova master: DNM:try l-c with direct deps https://review.opendev.org/c/openstack/nova/+/772780 | |
| 15:24:52 | kashyap | lyarwood: Back; unless it's at the start of a sentence, it's always lowercase. | |
| 15:25:28 | kashyap | sean-k-mooney: Both those occurrences are correct -- because the word came at the start of a sentence | |
| 15:25:39 | kashyap | So it's only natural to capitalize it :-) | |
| 15:26:05 | kashyap | lyarwood: What made you ask? Are you writing some docs / code comments or something? | |
| 15:28:21 | sean-k-mooney | kashyap: ya althoguh some porject for trademark reason say you should not capatilise it or use a spefic capatialistion when you do | |
| 15:28:48 | sean-k-mooney | kashyap: so the english rule dont alway apply if you are follow the project style/trademark guidelines | |
| 15:29:29 | kashyap | sean-k-mooney: Right; but pretty sure I saw what I described "lowercase all the time, except at the start of a sentence" somewhere in libvirt upstream discussions | |
| 15:29:37 | kashyap | But I can't find a link for you at the moment | |
| 15:31:38 | openstackgerrit | Ghanshyam proposed openstack/nova master: DNM:try l-c with direct deps https://review.opendev.org/c/openstack/nova/+/772780 | |
| 15:31:42 | sean-k-mooney | ya makes sense | |
| 15:32:01 | sean-k-mooney | i assume lyarwood was looking at documentaiton or simiarl and just wanted to make sure it was correct | |
| 15:44:36 | lyarwood | sean-k-mooney / kashyap ; https://review.opendev.org/c/openstack/nova/+/771981 was the context, stephenfin had a few nits and I couldn't remember the reasons for libvirt being lowercase all the time etc. | |
| 15:44:47 | lyarwood | thanks kashyap, I'll sort the change out now | |
| 15:45:24 | kashyap | lyarwood: No prob | |
| 15:50:26 | gibi | nova meeting starts in 10 minutes on #openstack-meeting-3 | |
| 15:55:52 | openstackgerrit | Lee Yarwood proposed openstack/nova master: docs: Move the LibvirtDistroSupportMatrix wiki page into our docs https://review.opendev.org/c/openstack/nova/+/771981 | |
| 16:08:44 | openstackgerrit | Merged openstack/nova master: tox: Enable parallel docs build https://review.opendev.org/c/openstack/nova/+/751034 | |
| 16:30:18 | sean-k-mooney | lyarwood: ah ok that was in the wiki for easy editing but we very very rarely update it so moving it too docs makes sense i guess | |
| 16:32:15 | lyarwood | yeah it wasn't easy and it's just nice to have it in tree and updated in the same change as the actual bump etc | |
| 16:33:10 | artom | Lol, I just realized running functional tests locally that need an un-mergde os-traits patch is... dumb. | |
| 16:35:24 | sean-k-mooney | artom: you can do that | |
| 16:35:28 | lyarwood | artom: install it into the env | |
| 16:35:34 | sean-k-mooney | you just pip install it in the env | |
| 16:35:50 | artom | Right, I'm sure there are ways | |
| 16:36:25 | sean-k-mooney | basicaly .tox/functional/bin/python3 -m pip install -e ../os-traits | |
| 16:36:44 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: [WIP] libvirt: Remove compareCPU() check on the destination https://review.opendev.org/c/openstack/nova/+/772917 | |
| 16:36:52 | sean-k-mooney | assuming you have os-traits cloneed and checked out beside your nova dir | |
| 17:27:58 | openstackgerrit | Stephen Finucane proposed openstack/nova master: mypy: Add nova.cmd, nova.conf, nova.console https://review.opendev.org/c/openstack/nova/+/705657 | |
| 17:27:58 | openstackgerrit | Stephen Finucane proposed openstack/nova master: mypy: Add type annotations to top-level modules https://review.opendev.org/c/openstack/nova/+/705658 | |
| 17:27:59 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Clean manager.Manager, service.Service signatures https://review.opendev.org/c/openstack/nova/+/764806 | |
| 17:27:59 | openstackgerrit | Stephen Finucane proposed openstack/nova master: WIP: Expand type hints for nova.block_device https://review.opendev.org/c/openstack/nova/+/742170 | |
| 17:28:00 | openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: Remove dead code https://review.opendev.org/c/openstack/nova/+/772928 | |
| 18:34:15 | openstackgerrit | Lee Yarwood proposed openstack/nova master: Disallow CONF.compute.max_disk_devices_to_attach = 0 https://review.opendev.org/c/openstack/nova/+/755360 | |
| 18:35:24 | artom | Didn't melwitt already do that? | |
| 18:36:34 | lyarwood | artom: that was just a rebase | |
| 18:36:48 | artom | Ag | |
| 18:36:49 | artom | Ah | |
| 18:38:36 | artom | *facepalm* | |
| 18:39:00 | artom | When 0 is an "allowed" value, but None is not, don't just if <value> :( | |
| 18:39:08 | artom | That's a couple of hours wasted | |
| 18:54:13 | sean-k-mooney | its the same as the rpc workers issue we fixed a whiel back | |
| 18:54:19 | sean-k-mooney | jsut missign the min=1 | |
| 18:54:50 | sean-k-mooney | ah we have -1 for unlimited | |
| 18:55:02 | sean-k-mooney | so so we cant do that | |
| 18:55:08 | sean-k-mooney | really we should have used 0 for that | |
| 18:55:21 | sean-k-mooney | in this case since 0 is not valid | |
| 20:12:20 | ganso | Hi Nova folks! I just logged this bug https://bugs.launchpad.net/nova/+bug/1913641 and while thinking about the possible ways to solve it, I started wondering whether it was intentional (and therefore not a bug) to consider shelved_offloaded instances as running, metric-wise. | |
| 20:12:21 | openstack | Launchpad bug 1913641 in OpenStack Compute (nova) "Incorrect Shelved_offloaded instance metrics on openstack usage show output" [Undecided,New] | |
| 21:03:34 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: WIP: libvirt: start tracking NUMACell.socket for hosts https://review.opendev.org/c/openstack/nova/+/766816 | |
| 21:03:35 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: WIP: extra specs/image pros: add `socket PCI NUMA affinity https://review.opendev.org/c/openstack/nova/+/772748 | |
| 21:03:35 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: WIP: Add `socket` PCI NUMA affinity policy request prefilter https://review.opendev.org/c/openstack/nova/+/772749 | |
| 21:03:36 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: WIP: pci: implement the SOCKET NUMA affinity policy https://review.opendev.org/c/openstack/nova/+/772779 | |
| 21:04:21 | artom | ganso, I *think* `openstack show usage` is about quotas, and shelved offloaded still count against the quota | |
| 21:04:53 | artom | But you're correct in assuming their resources (disk, VM) should not be counted... | |
| 21:10:18 | ganso | artom: Thanks for the response. hmm given that the metrics shown are cpu-hours, ram-hours and disk-hours, it is a constantly updating, and therefore increasing the value to be billed, if this info is used for billing, whereas quotas remain constant if the number of instances do not change. And indeed, since "count_usage_from_placement" was introduced, shelved_offloaded instances do not even have allocations anymore (nor quotas), but they are listed | |
| 21:10:18 | ganso | in usage show because they show up in "InstanceList". So, considering all that, sounds like a valid bug. | |
| 21:52:28 | openstackgerrit | Merged openstack/nova master: Allow extending an existing instance allocation https://review.opendev.org/c/openstack/nova/+/757109 | |
| 22:06:13 | openstackgerrit | Merged openstack/nova master: Make remove allocation symmetric with add allocation https://review.opendev.org/c/openstack/nova/+/757110 | |
| 22:19:40 | openstackgerrit | Merged openstack/nova master: Factor out port resource deallocation https://review.opendev.org/c/openstack/nova/+/757111 | |