| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-03-28 | |||
| 14:05:37 | efried | sounds good, thanks. | |
| 14:05:45 | gibi | efried: a similar question regrading nova onboarding in Denver. Do you need help? | |
| 14:05:53 | efried | gibi: absolutely! | |
| 14:06:12 | efried | Here's where I will eagerly take all the help I can get. | |
| 14:06:34 | efried | I'm never sure what form onboarding should take | |
| 14:06:56 | efried | the audience always seems so diverse in terms of background/experience | |
| 14:07:21 | efried | I was toying with the idea of doing it as a live demo of developing a bug fix. | |
| 14:07:52 | efried | So like, look at the bug, write and upload a test case patch, write and upload the fix patch | |
| 14:08:05 | efried | which of course is your wheelhouse completely. | |
| 14:08:32 | gibi | efried: we migth be able to do that, but it needs to be a canned bug / bugfix to make sure we can really fix it | |
| 14:09:00 | gibi | in the given time | |
| 14:09:11 | efried | gibi: Yeah, I thought maybe prepare by uploading a patch that reverts a known bug fix (and its test case) and then basically redo the work on top of it. | |
| 14:09:27 | efried | So it would be a three-patch series, but we only talk about two of the patches. | |
| 14:09:34 | efried | and abandon all three at the end. | |
| 14:09:45 | gibi | efried: good idea | |
| 14:10:35 | efried | need to pick a bug that's pretty nova-y; otherwise it really just turns into a demo of "best practices for fixing a bug in the gerrit workflow" | |
| 14:10:58 | efried | focus attention on the flow of the test case and the nature of the fix. | |
| 14:11:07 | gibi | pci? bdm? live migration? :) | |
| 14:11:33 | gibi | the heavy nova stuff | |
| 14:11:56 | efried | something with a test case that's not like hundreds of lines long. | |
| 14:14:05 | gibi | efried: this might be too heavy on test side https://review.openstack.org/#/c/624444/5/nova/tests/functional/libvirt/test_pci_sriov_servers.py | |
| 14:15:20 | gibi | efried: maybe this https://review.openstack.org/#/c/640302/ | |
| 14:15:53 | gibi | small fix, on the API with API sample test | |
| 14:16:13 | efried | gibi: Well, in that pci/sriov one the delta is bigger than I would like, but if we used it for the demo, it would be just L243-88 | |
| 14:17:13 | efried | that second one doesn't really have a clear [test case] and [fix] | |
| 14:17:27 | gibi | efried: it is api sample so it is well automataed | |
| 14:18:04 | gibi | but I hear you | |
| 14:18:57 | cdent | do we still have people who actively understand and like the api sample tests? | |
| 14:19:18 | efried | I'm looking for commits whose title includes the word 'recreate' | |
| 14:19:21 | openstackgerrit | Silvan Kaiser proposed openstack/nova master: WIP: Adds systemd detection result caching in Quobyte driver https://review.openstack.org/648093 | |
| 14:20:47 | gibi | cdent: at least I would not be affraid to dig into it if needed | |
| 14:20:52 | gibi | efried: what about https://review.openstack.org/#/c/632904/3/nova/tests/functional/test_servers.py | |
| 14:21:06 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Add functional recreate test for bug 1819963 https://review.openstack.org/648409 | |
| 14:21:07 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Update instance.availability_zone on revertResize https://review.openstack.org/648410 | |
| 14:21:07 | openstack | bug 1819963 in OpenStack Compute (nova) rocky "Reverting a resize does not update the instance.availability_zone value to the source az" [Medium,Confirmed] https://launchpad.net/bugs/1819963 | |
| 14:21:32 | openstackgerrit | Silvan Kaiser proposed openstack/nova master: Adds systemd detection result caching in Quobyte driver https://review.openstack.org/648093 | |
| 14:21:36 | efried | gibi: heh, what about --^ | |
| 14:22:17 | gibi | efried: could be. I feel https://review.openstack.org/#/c/632904/3/nova/tests/functional/test_servers.py is more self contained | |
| 14:23:06 | efried | gibi: I was being lazy and looking for one where the recreate was separate from the fix, but it would be easy enough to split out. | |
| 14:23:34 | efried | again with https://review.openstack.org/#/c/648409/1/nova/tests/functional/test_availability_zones.py we wouldn't have to include the refactor part in the demo, just L145-73 | |
| 14:24:39 | gibi | efried: OK. you convinced me | |
| 14:24:48 | kaisers | efried: FYI, added the requested tests/nit-corrections to https://review.openstack.org/648093 , pls check if that fits your view | |
| 14:25:57 | efried | kaisers: ack | |
| 14:26:27 | openstackgerrit | Takashi NATSUME proposed openstack/nova stable/stein: doc: Fix openstack CLI command https://review.openstack.org/648412 | |
| 14:28:03 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Add functional recreate test for bug 1819963 https://review.openstack.org/648414 | |
| 14:28:04 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Update instance.availability_zone on revertResize https://review.openstack.org/648415 | |
| 14:28:04 | openstack | bug 1819963 in OpenStack Compute (nova) rocky "Reverting a resize does not update the instance.availability_zone value to the source az" [Medium,In progress] https://launchpad.net/bugs/1819963 - Assigned to Matt Riedemann (mriedem) | |
| 14:31:04 | Zara | hi! I have a question; aiui when disk space is overcommitted, disk_available_least shows a negative value (eg: with `openstack hypervisor stats show`). Why is a negative value displayed instead of '0'? It seems like the value could be infinitely low, so I'm wondering how the information would then be used? | |
| 14:31:26 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Update instance.availability_zone on revertResize https://review.openstack.org/648415 | |
| 14:32:14 | efried | Zara: Just spitballing here, but I assume it gives you an indication of how overcommitted you are. | |
| 14:32:20 | sean-k-mooney | Zara: it means that the total amount of worse case diskspace for the current instance on the host exceed the availble diskspace | |
| 14:32:36 | sean-k-mooney | efried: yes it does | |
| 14:33:17 | sean-k-mooney | Zara: i found this really good github gist that explains it https://gist.github.com/JCallicoat/43505cab0535057ca4fb | |
| 14:33:20 | efried | arguably this could be "fixed" to floor out at zero, but that would require a new microversion, so it probably hasn't been considered worth the hassle since you could do that on the client side if you don't care about the "extra information". | |
| 14:33:27 | sean-k-mooney | we porably shoudl have somthing similar in our docs | |
| 14:33:44 | sean-k-mooney | efried: i dont think this is a bug | |
| 14:33:49 | sean-k-mooney | it was by design | |
| 14:34:23 | openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/os-vif master: Remove IP proxy methods https://review.openstack.org/643115 | |
| 14:34:27 | efried | Right, right, it's a feature, not a bug. | |
| 14:34:57 | sean-k-mooney | efried: that said this was origianly used by the disk filters i nthe scheduler but we do that in placmenet now | |
| 14:35:11 | sean-k-mooney | orperhaps it was the disk wheigher | |
| 14:35:45 | Zara | thanks for the gist (yeah, that says nova-scheduler). | |
| 14:36:24 | sean-k-mooney | hum the weigher actully used free_disk_mb | |
| 14:37:02 | Zara | An official docs note somewhere would be good; I encountered it in the context of someone thinking it was a bug. | |
| 14:39:18 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/pike: Add functional recreate test for bug 1819963 https://review.openstack.org/648421 | |
| 14:39:19 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/pike: Update instance.availability_zone on revertResize https://review.openstack.org/648422 | |
| 14:39:19 | openstack | bug 1819963 in OpenStack Compute (nova) rocky "Reverting a resize does not update the instance.availability_zone value to the source az" [Medium,In progress] https://launchpad.net/bugs/1819963 - Assigned to Matt Riedemann (mriedem) | |
| 14:39:36 | sean-k-mooney | lookin futher we do https://github.com/openstack/nova/blob/9db1c6061d10617b40dbc1d0d64faed79503cb81/nova/scheduler/host_manager.py#L213 so we set free_gb in the context of the host state object to the smalles to free_gb or lesast_gb | |
| 14:40:41 | sean-k-mooney | so yes the disk weighter will prefer host that are less over commited | |
| 14:40:43 | sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/scheduler/weights/disk.py | |
| 14:40:58 | sean-k-mooney | which is why we have the negitive value | |
| 14:41:49 | openstackgerrit | Takashi NATSUME proposed openstack/nova stable/rocky: doc: Fix openstack CLI command https://review.openstack.org/648425 | |
| 14:43:31 | Zara | sean-k-mooney: ah, I see, thanks | |
| 14:44:57 | efried | Zara: It would be great if you wanted to find the appropriate spot in the docs and propose a patch to enhance it :) | |
| 14:46:06 | Zara | efried: I'll see what I can do. :) | |
| 14:46:14 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add retry_on_deadlock to migration_update DB API https://review.openstack.org/648428 | |
| 14:56:32 | efried | Zara: Thanks! | |
| 15:23:35 | Zara | (aw, I found a nice place to add it... but the section is in the ocata docs; I haven't yet found the correlate of those docs for more recent releases (https://docs.openstack.org/ocata/config-reference/compute/schedulers.html ) | |
| 15:24:50 | sean-k-mooney | Zara: we generate our config docmuemnt from code now | |
| 15:25:21 | amodi | what was the spam about? :? | |
| 15:25:27 | Zara | ah okay, guess I'll need to add a docstring somewhere | |
| 15:25:28 | sean-k-mooney | Zara: https://github.com/openstack/nova/blob/master/nova/conf/scheduler.py | |
| 15:25:54 | Zara | sean-k-mooney: thanks | |
| 15:26:54 | sean-k-mooney | we do have ohter schduler docs somewhere actully | |
| 15:27:50 | sean-k-mooney | Zara: i think this is what you want https://github.com/openstack/nova/blob/master/doc/source/admin/configuration/schedulers.rst | |
| 15:34:14 | Zara | seank-k-mooney: ahaaa, yes that looks like the same doc | |
| 15:34:16 | Zara | thanks | |
| 15:54:41 | openstackgerrit | Dan Smith proposed openstack/os-traits master: Add COMPUTE_IMAGE_TYPE_* traits https://review.openstack.org/648147 | |
| 16:02:03 | ricolin | cdent, jaypipes, zhipeng do you think you guys can help with http://lists.openstack.org/pipermail/openstack-discuss/2019-March/004246.html for Resource Management SIG? | |
| 16:02:25 | cdent | ricolin: there hasn't been any resource mgt sig activity, well, ever | |
| 16:02:29 | cdent | just hope | |
| 16:02:59 | cdent | zhipeng may have some thoughts, but though we wanted to we never really got it off the ground | |
| 16:03:40 | ricolin | cdent, is there any plan for mgt sig in Summit or PTG this time? | |
| 16:03:57 | cdent | not that I'm aware of | |
| 16:04:10 | cdent | but again, zhipeng may have more info | |
| 16:19:34 | openstackgerrit | Zara proposed openstack/nova master: Explain why disk_available_least can be negative https://review.openstack.org/648454 | |
| 16:21:26 | ricolin | cdent, thanks, will try to reach out with zhipeng | |
| 16:22:27 | Zara | (my patch may be a bit brief, but I guess it's a start) | |
| 16:22:44 | openstackgerrit | Merged openstack/os-traits master: Add COMPUTE_IMAGE_TYPE_* traits https://review.openstack.org/648147 | |
| 16:46:47 | openstackgerrit | Eric Fried proposed openstack/os-traits master: Add contextual comment about image type traits https://review.openstack.org/648466 | |