| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-06-11 | |||
| 20:24:44 | gmann | its different problem. i can see lof of LaTeX warning but not sure that leads to error | |
| 20:24:53 | melwitt | yeah same | |
| 20:28:57 | elod | gmann: yes, I think it would be good :) | |
| 20:30:15 | dansmith | elod: I guess we could, on stable, fail if we find no cherry-pick lines, unless the "StableOnly" tag is in the commit message | |
| 20:31:32 | elod | dansmith: sounds good! | |
| 20:36:00 | elod | is see e.g. train-only, stein-only, etc tags in commit messages, too, however. maybe those can be accepted next to stable-only as well | |
| 20:36:09 | gmann | +1, with 'StableOnly' tag it can work fine | |
| 20:36:25 | gmann | just stable-only is easy | |
| 20:36:54 | dansmith | maybe StableOnly: train ? | |
| 20:36:55 | elod | I'm OK with that | |
| 20:37:16 | dansmith | I'll just make it grep for StableOnly and we can decide | |
| 20:37:22 | gmann | but that may be backported to stein and so. | |
| 20:37:42 | gmann | like i start fixing from train till rocky | |
| 20:38:04 | gmann | devstack has such cases very often | |
| 20:38:13 | elod | StableOnly: train is a bit superfluous | |
| 20:38:48 | elod | let just be it simply stable-only or [stable-only] :) | |
| 20:39:01 | gmann | we can say 'StableOnly' means not coming from master but it can be backported to older stable | |
| 20:40:06 | openstackgerrit | Dan Smith proposed openstack/nova master: Check cherry-pick hashes in pep8 tox target https://review.opendev.org/735205 | |
| 20:40:14 | dansmith | gmann: yeah I think that's the way to do it | |
| 20:40:25 | dansmith | "StableOnly means it originated on some stable branch" | |
| 20:40:44 | gmann | yeah | |
| 20:41:13 | openstackgerrit | Dan Smith proposed openstack/nova stable/ussuri: Check cherry-pick hashes in pep8 tox target https://review.opendev.org/735208 | |
| 20:41:54 | openstackgerrit | Dan Smith proposed openstack/nova stable/ussuri: DNM: Testing a thing https://review.opendev.org/735241 | |
| 21:48:56 | openstack | bug 1550919 in OpenStack Compute (nova) train "[Libvirt]Evacuate fail may cause disk image be deleted" [Medium,In progress] https://launchpad.net/bugs/1550919 - Assigned to Lee Yarwood (lyarwood) | |
| 21:48:56 | openstackgerrit | Merged openstack/nova stable/train: Add functional test for bug 1550919 https://review.opendev.org/731272 | |
| 22:19:29 | openstackgerrit | Dan Smith proposed openstack/nova master: Check cherry-pick hashes in pep8 tox target https://review.opendev.org/735205 | |
| 22:19:49 | openstackgerrit | Dan Smith proposed openstack/nova stable/ussuri: Check cherry-pick hashes in pep8 tox target https://review.opendev.org/735208 | |
| 22:19:55 | dansmith | elod: gmann: ^ I'm out tomorrow FYI | |
| 23:33:58 | openstackgerrit | Merged openstack/nova stable/train: libvirt: Don't delete disks on shared storage during evacuate https://review.opendev.org/731273 | |
| #openstack-nova - 2020-06-12 | |||
| 01:48:36 | openstackgerrit | sunhao proposed openstack/nova master: Add checks for volume status when rebuilding https://review.opendev.org/734734 | |
| 03:22:35 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Workaround for pdf-docs build failing with sphinx 3.1.0 https://review.opendev.org/735279 | |
| 03:31:04 | gmann | melwitt: stephenfin gibi dansmith ^^ this is workaround on gate fix for pdf-docs build failing. We are hitting the TeX memory issue here due to our large sample files. | |
| 03:32:03 | gmann | i am not sure how to fix that as sample files is too large to include. It started failing with new Sphinx 3.1.0 which is also mystery as i cannot find any related change in that. | |
| 03:35:57 | gmann | workaround passing on gate. I am going away as my memory also exhausted :) on this pdf debugging where no relevant log on failure. | |
| 03:36:38 | gmann | can we make pdf build as non-failing ? | |
| 03:48:09 | ikla | when using sriov neutron ports can I name them a specific interface name when booting it to linux? like 10eth01 and 20eth01 | |
| 07:34:16 | openstackgerrit | Alexandre Arents proposed openstack/nova master: Snapshot: offload glance upload in a native thread https://review.opendev.org/734776 | |
| 11:21:58 | zigo | env=OS_NOVA_DISABLE_EVENTLET_PATCHING=yes | |
| 11:21:58 | zigo | Hi there ! I am wondering, would it be safe to run nova-api with: | |
| 11:22:12 | zigo | We're running with uwsgi, so not under eventlet... | |
| 11:27:54 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Workaround for pdf-docs build failing with sphinx 3.1.0 https://review.opendev.org/735279 | |
| 11:45:14 | gibi | gmann: updated ^^ | |
| 12:37:00 | gmann | gibi: thanks. checking. | |
| 12:43:36 | gmann | gibi: found the real issue on this. we are expanding the sample policy file in two places and that is why two times giant policy file inclusion end up excaussting memory - 1. https://github.com/openstack/nova/blob/a9e0b0573da65074db686ceb0c5cc000ec7110f1/doc/source/configuration/sample-policy.rst 2. https://github.com/openstack/nova/blob/master/doc/source/admin/configuration/samples/policy.yaml.rst | |
| 12:43:57 | gmann | we should ref the second one to first location and then it solve the memory issue. | |
| 12:44:16 | gmann | i thought it was just one place and still eating up all memory. | |
| 12:47:12 | gmann | and pdf doc also should be fine as it can refer the another section in same pdf doc. | |
| 12:55:34 | gibi | gmann: cool. feel free to propose the real solution | |
| 12:56:07 | gmann | i am testing on pdf doc also so that we do not endup web link there and then i will update | |
| 12:59:16 | gibi | yeah, I saw that many of our pdf links are weblinks. But I think this is a generic problem | |
| 13:13:55 | lvdombrkr | Hello guys. changing root password with nova set-password and nova get-password is deprecated now? | |
| 13:14:03 | gmann | gibi: you mean external links right not nova doc link right? | |
| 13:19:34 | gmann | lvdombrkr: no, its not deprecated. | |
| 13:20:01 | gmann | any warning you are getting for deprecation ? | |
| 13:23:33 | lvdombrkr | gmann: no no i heard somewhere that its depricated. can i pass by this metod password to instance during booting? | |
| 13:25:28 | gmann | lvdombrkr: yeah you can pass with adminPass field | |
| 13:26:00 | gmann | --admin-pass for CLI | |
| 13:26:57 | lvdombrkr | gmann: and then this pass will be stored on metadata as plaintext? | |
| 13:32:09 | gibi | gmann: looked deeper, most of the nova doc links are local in the pdf but some (e.g. on page 125) are web links even if it could be a pdf local link | |
| 13:33:02 | zigo | gibi: gmann: Do you know the answer about OS_NOVA_DISABLE_EVENTLET_PATCHING=yes? Is it fine to run nova-api with it? If so, I'll hard-wire it in the nova package in Debian, as it looked working better for us. I really would love to get upstream opinion here... | |
| 13:33:38 | zigo | I don't know the code well enough to tell what this implies. | |
| 13:35:27 | gibi | zigo: based on the code comment in https://github.com/openstack/nova/blob/a9e0b0573da65074db686ceb0c5cc000ec7110f1/nova/monkey_patch.py#L95 I don't think it is intended the services | |
| 13:36:08 | zigo | Right. | |
| 13:36:55 | zigo | That's kind of what I thought, though my colleague insisted... | |
| 13:37:45 | gibi | zigo: also I think the nova-api service uses eventlet for parallel querying multiple cells | |
| 13:40:59 | lvdombrkr | gmann : openstack server create: error: unrecognized arguments: --admin-pass | |
| 13:43:42 | gmann | lvdombrkr: you are using osc or novaclient? for nova client --admin-pass works but i cannot see that in osc which might be missing and not up to date | |
| 13:43:57 | gmann | and password in get is always encrypted not plaintext. | |
| 13:45:51 | gmann | and clear password does not actually clear the server password but only remove it from metadata | |
| 13:47:48 | gmann | gibi: i cannot see page 125 with link? or you mean some other page? | |
| 13:48:38 | lvdombrkr | gmann: i need to remove ir from metadata by myself? | |
| 13:49:02 | gmann | I mean if you want then there is API for that. | |
| 13:49:17 | gmann | otherwise in metadata it is in encrypted form. | |
| 13:50:24 | gibi | gmann: https://ibb.co/ftWttM2 | |
| 13:51:34 | gmann | gibi: ohk, that is 114 in this - https://docs.openstack.org//nova/latest/doc-nova.pdf | |
| 13:51:57 | gmann | yeah those should be fixed. I will do it but later | |
| 13:52:10 | gibi | gmann: yeah this is from the pdf workaround patch so I guess the difference in numbering is due to the remove policy sample | |
| 13:52:18 | gibi | gmann: sure, no rush fixing these | |
| 13:52:19 | gmann | i see | |
| 13:54:36 | lvdombrkr | should be set? | |
| 13:54:36 | lvdombrkr | gmann: to pass password by --admin-pass in nova.conf inject_password=true | |
| 13:54:44 | lvdombrkr | or its not related? | |
| 13:54:54 | lvdombrkr | and are two saparete ways | |
| 13:56:15 | gibi | lvdombrkr: did you tried the nova CLI I'm not sure openstack CLI has the --admin-pass flag implemented | |
| 13:56:38 | gmann | yeah osc does not have | |
| 13:58:34 | gibi | lvdombrkr: and I also think you need inject_password=true in the nova.conf on the compute side | |
| 13:58:44 | gibi | maybe even inject_partition = -1 | |
| 13:59:21 | gibi | and also python-guestfs package installed on the compute node | |
| 13:59:46 | lvdombrkr | gibi: inject_password=true inject_partition = -1 not working with python3 thats why im looking other way to inject password | |
| 13:59:49 | frickler | openstack server set --root-password exists, but no get-password seemingly | |
| 14:00:09 | lvdombrkr | gibi : https://review.opendev.org/#/c/733982/ | |
| 14:01:23 | lvdombrkr | frickler: yes yes but still need this patch https://review.opendev.org/#/c/733982/ correct? | |
| 14:01:41 | gmann | humm, and we do not test that? | |
| 14:02:27 | frickler | gmann: doesn't look like it, unit testing is all mocked, no real functional test I've found | |
| 14:02:35 | gibi | lvdombrkr: thanks for the context, I think I was able to reproduce your bug | |
| 14:03:38 | frickler | lvdombrkr: the set-pw seems to try to use QEMU guest agent instead, when I try it I get: QEMU guest agent is not enabled (HTTP 409) | |
| 14:05:15 | lvdombrkr | frickler : good than i need to test it | |
| 14:10:58 | gmann | frickler: lvdombrkr yeah, for kvm or qemu you need to enable the qemu agent | |
| 14:29:28 | lvdombrkr | gmann: i can enable it permanently in nova.conf? | |
| 14:32:41 | gmann | lvdombrkr: not in nova conf but you need to enable it on image. you can set hw_qemu_guest_agent in image meta. | |