| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-10-08 | |||
| 19:36:24 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Helper to start computes with different HostInfos https://review.opendev.org/686832 | |
| 19:36:25 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: Functional tests for NUMA live migration https://review.opendev.org/672595 | |
| 19:36:27 | openstackgerrit | Artom Lifshitz proposed openstack/nova master: [WIP] NUMA LM: Add func test for bug 1845146 https://review.opendev.org/687404 | |
| 19:36:27 | openstack | bug 1845146 in OpenStack Compute (nova) train "NUMA aware live migration failed when vCPU pin set" [High,Fix committed] https://launchpad.net/bugs/1845146 - Assigned to Dan Smith (danms) | |
| 20:18:33 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Bulk query floating IPs in _nw_info_get_ips https://review.opendev.org/465792 | |
| 20:18:34 | openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: Time how long _nw_info_get_ips takes https://review.opendev.org/687410 | |
| 20:31:07 | openstackgerrit | Merged openstack/nova master: doc: fix formatting in mitigation-for-Intel-MDS-security-flaws https://review.opendev.org/687350 | |
| 20:33:03 | openstackgerrit | Eric Fried proposed openstack/nova master: Deprecate [api]auth_strategy and noauth2 https://review.opendev.org/687373 | |
| 20:33:04 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP: Murder NoAuthMiddleware https://review.opendev.org/687416 | |
| 20:33:10 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: link to user/availability-zones from user home page https://review.opendev.org/687165 | |
| 20:33:10 | openstackgerrit | Matt Riedemann proposed openstack/nova master: doc: link to user/index from main home page https://review.opendev.org/687292 | |
| 20:33:41 | efried | mriedem: It turned out to be actually quite easy to bypass noauth and just stub out keystone --^ | |
| 20:34:04 | efried | there's a lot more to be done there, but I was able to get through enough pieces to convince myself that that approach is going to work. | |
| 20:34:23 | efried | the remainder is just cleaning up the leftovers. | |
| 20:34:55 | mriedem | does ksa/keystonemiddleware provide any fixtures for some of this? | |
| 20:35:15 | mriedem | that would be ideal rather than stubbing keystonemiddleware.auth_token.filter_factory | |
| 20:35:35 | mriedem | oh hello https://github.com/openstack/keystonemiddleware/blob/master/keystonemiddleware/fixture.py | |
| 20:35:43 | efried | mriedem: I agree, but that would be a subsequent step in any case. | |
| 20:36:18 | efried | as in, doing this is not taking away anything we were doing before, because we weren't hitting those factories at all | |
| 20:36:34 | efried | I mentioned same in my commit message | |
| 20:36:59 | efried | lbragstad: any idea if there's a fixture for keystonemiddleware.auth_token.filter_factory ? | |
| 20:38:01 | efried | ah, neat | |
| 20:38:13 | efried | lbragstad: cancel, I did not read scrollback thoroughly. | |
| 20:38:32 | lbragstad | yup - looks like you're on the right track | |
| 20:40:23 | mriedem | efried: ultra mega -1 on the bottom change | |
| 20:40:36 | mriedem | not the logging thing, the reno | |
| 20:41:04 | efried | mriedem: o ya, I thought of that while I was updating and then forgot again, will do. | |
| 20:42:42 | mriedem | so with https://review.opendev.org/#/c/686414/ i guess we can push rc2 now yeah? | |
| 20:42:44 | mriedem | are you doing that? | |
| 20:43:19 | efried | mriedem: I can, or you can, up to you. | |
| 20:43:24 | mriedem | i'll push it real good | |
| 20:43:29 | efried | ++ | |
| 20:46:47 | mriedem | https://review.opendev.org/687418 | |
| 21:02:01 | eandersson | This makes me sad | |
| 21:02:02 | eandersson | https://github.com/openstack/nova/blob/stable/rocky/nova/virt/libvirt/vif.py#L232 | |
| 21:02:03 | eandersson | :'( | |
| 21:02:50 | eandersson | I am pretty sure CentOS 7.x supports more than 8. | |
| 21:02:50 | mriedem | why for | |
| 21:03:00 | mriedem | prove it! | |
| 21:03:02 | eandersson | In fact we have it set to a lot more than 8, but didn't realize that after upgrading it downgraded them | |
| 21:03:31 | mriedem | does the kernel in centros 7.x have some patch that makes that more than 8? | |
| 21:03:32 | eandersson | Probably worth checking with someone from RedHat to confirm | |
| 21:03:36 | mriedem | yar | |
| 21:05:32 | eandersson | My colleague told me he pulled down the source code and confirmed | |
| 21:30:16 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove dead reserve_volume compat code in _validate_bdm https://review.opendev.org/687427 | |
| 21:30:22 | mriedem | eandersson: but was it patched in or something? | |
| 21:31:24 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove dead reserve_volume compat code in _validate_bdm https://review.opendev.org/687427 | |
| 21:42:08 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove Stein compute compat checks for volume type support https://review.opendev.org/687428 | |
| 21:48:45 | mriedem | this is a good easy fix https://review.opendev.org/#/c/678991/ | |
| 21:54:11 | ceryx | mriedem: I'm the colleague eandersson mentioned. https://elixir.bootlin.com/linux/v3.10/source/drivers/net/tun.c#L115 has the cap set to 8, but that same file from the centos source RPM for that kernel version is https://pastebin.com/2415QjTW. Unsure what version they changed that in though. | |
| 21:55:25 | ceryx | but they set '#define MAX_TAP_QUEUES 256 | |
| 21:55:26 | ceryx | ' rather than DEFAULT_MAX_NUM_RSS_QUEUES (which is still set to 8 in the rhel/centos version) | |
| 21:56:37 | mriedem | ok, i guess that's the question | |
| 21:57:09 | mriedem | and if it's a centos only patch then it's pretty hard to have a condition in nova for that | |
| 22:01:29 | kosa777777 | I am also the colleague of eandersson and ceryx, how about having option to set it to certain number on compute itself if operator chooses to not rely on default? | |
| 22:03:02 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Pre-validate volumes for cross_az_attach=False during create https://review.opendev.org/687430 | |
| 22:03:54 | mriedem | kosa777777: i'd prefer not having options like that but not sure what others think - maybe start by reporting a bug and taking it from there | |
| 22:05:29 | ceryx | I checked through the centos kernel versions up through 7.7 and they all have that driver hard-coded to #define MAX_TAP_QUEUES 256 :/ | |
| 22:05:57 | mriedem | can you do a git blame and find out the change that patched it? | |
| 22:11:22 | kosa777777 | also "platform.linux_distribution()" would determine if it is CentOS, if that is needed to distinguish Ubuntu from Centos | |
| 22:14:55 | mriedem | let's start with a reported bug with options and discuss there - there could be a [workarounds] group option if people are ok with that | |
| 22:17:56 | eandersson | Yea sounds good | |
| 22:26:43 | kosa777777 | great | |
| 22:29:35 | eandersson | mriedem, do you happen to have an estimate on when we could get an initial version of the backport for making restating computes safe :P | |
| 22:29:57 | eandersson | I could probably take a stab at it, but last time the number of differences between the master and rocky code base was pretty overwhelming | |
| 22:30:57 | mriedem | it starts in train which GAs on oct 17 https://wiki.openstack.org/wiki/Nova/Train_Release_Schedule - after that we can release in train 20.0.1 and move to stein + release and then rocky + release | |
| 22:31:12 | mriedem | so my guess would be early november at the earliest | |
| 22:31:48 | mriedem | meaning, it's not a regression in train so we're not putting it into the train GA right now, it will be in the first patch release after the train GA | |
| 22:32:02 | mriedem | then stein and rocky as normal | |
| 22:36:26 | eandersson | Yea - I mean for me it's more about the patch being pushed, but not necessarily merged. | |
| 22:36:54 | mriedem | you can propose the stein and rocky backports, there are at least 2 patches for that fix b/c there is a functional recreate test patch and then the fix patch | |
| 22:37:07 | mriedem | or you can bribe gibi to move on that | |
| 22:37:16 | eandersson | How do I bribe gibi? beer? :D | |
| 22:37:20 | mriedem | goulash | |
| 22:43:35 | openstackgerrit | Eric Fried proposed openstack/nova master: Deprecate [api]auth_strategy and noauth2 https://review.opendev.org/687373 | |
| 22:43:35 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP: Murder NoAuthMiddleware https://review.opendev.org/687416 | |
| 23:28:42 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Pre-validate volumes for cross_az_attach=False during create https://review.opendev.org/687430 | |
| 23:29:27 | mriedem | dansmith: for your morrow, ^ is an alternative and much less fugly solution for the cross_az_attach=False server create issue that i was trying to fix in https://review.opendev.org/#/c/469675/ | |
| 23:29:41 | mriedem | basically all contained in the api so less passing of garbage down and up through the layers | |
| #openstack-nova - 2019-10-09 | |||
| 00:03:54 | openstackgerrit | Merged openstack/nova master: api-ref: Fix security groups parameters https://review.opendev.org/686517 | |
| 02:32:58 | openstackgerrit | Jing Zhang proposed openstack/nova master: [WIP] Default mem_page_size to small for pinned instances https://review.opendev.org/686079 | |
| 03:58:31 | openstackgerrit | Dongcan Ye proposed openstack/nova master: libvirt: Check the guest support UEFI https://review.opendev.org/685678 | |
| 07:16:07 | SonPham | hi . how to boot to instance with this conf | |
| 07:16:11 | SonPham | <disk type="network" device="disk"> | |
| 07:16:45 | SonPham | i changed xml file type="raw" to type="qcow2" | |
| 07:17:02 | SonPham | Nova error: Error: Failed to perform requested operation on instance "4444", the instance has an error status: Please try again later [Error: CantStartEngineError]. | |
| 07:20:06 | openstackgerrit | Brin Zhang proposed openstack/nova master: Follow-up: Add delete_on_termination to volume-attach API https://review.opendev.org/680655 | |
| 07:36:57 | gibi | eandersson: hi! I read about bribery above but not 100% sure which bugfix you are talking about. is it https://bugs.launchpad.net/nova/+bug/1833581 ? | |
| 07:36:57 | openstack | Launchpad bug 1833581 in OpenStack Compute (nova) train "instance stuck in BUILD state if nova-compute is restarted" [Low,In progress] - Assigned to Balazs Gibizer (balazs-gibizer) | |
| 07:39:32 | SonPham | hi . how to boot to instance with this conf | |
| 08:40:05 | kashyap | melwitt: stephenfin: Added my review notes here: https://review.opendev.org/#/c/589769/ (skip CPU comparison on AAr ch64) | |
| 09:55:20 | openstackgerrit | Merged openstack/nova master: Skip functional test jobs for doc redirect changes https://review.opendev.org/687357 | |
| 09:55:25 | openstackgerrit | Merged openstack/nova master: docs: Add redirects for '/user/aggregates' https://review.opendev.org/687353 | |
| 09:55:31 | openstackgerrit | Merged openstack/nova master: doc: link to user/availability-zones from user home page https://review.opendev.org/687165 | |
| 09:55:37 | openstackgerrit | Merged openstack/nova master: doc: link to user/index from main home page https://review.opendev.org/687292 | |
| 10:08:39 | efried | gibi: Yes, eandersson was talking about https://bugs.launchpad.net/nova/+bug/1833581 | |
| 10:08:39 | openstack | Launchpad bug 1833581 in OpenStack Compute (nova) train "instance stuck in BUILD state if nova-compute is restarted" [Low,In progress] - Assigned to Balazs Gibizer (balazs-gibizer) | |
| 10:11:15 | efried | gibi: Was there not a func repro patch for that? | |
| 10:13:28 | gibi | efried: thanks. I can propose stable patches to stable branches even before the stein backport merges. | |
| 10:13:38 | gibi | efried: stable/stein has the func reproduce as it merged earlier | |