| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-02-03 | |||
| 14:09:20 | bauzas | sean-k-mooney: you can surely review them if you want | |
| 14:09:37 | bauzas | I'm not really a nova-net expert so having more eyes is better | |
| 14:09:42 | stephenfin | I will never turn down reviewers | |
| 14:10:01 | bauzas | any review is helpful FWIW | |
| 14:10:25 | sean-k-mooney | well basically im just going to double check the commit message and the code make sense and you are not breaking the neutorn path | |
| 14:10:37 | bauzas | even if I (sometimes/always/somehow/pick the one you prefer) disagree with the review and associated comments | |
| 14:11:51 | sean-k-mooney | stephenfin: by the way once this is complete i want to do a pass over os-vif to delete the nova-net only bits | |
| 14:13:17 | sean-k-mooney | we have a few object fields that i think are only there for nova-net so i want to deprecate them and then likely remove them early in V although we could maybe do it in u. time will tell. | |
| 14:14:11 | sean-k-mooney | things like this https://github.com/openstack/os-vif/blob/master/os_vif/objects/network.py#L33 | |
| 14:14:36 | sean-k-mooney | the multi_host flag is for nova networks and i dont think we ever use it in os-vif | |
| 14:15:30 | sean-k-mooney | ok we technically do but only in the linux bridge driver and even so only with nova-networks https://github.com/openstack/os-vif/blob/1e4cf6160efc7fb4fd8279bd2a666e4d72f4fb6c/vif_plug_linux_bridge/linux_bridge.py#L101 | |
| 14:17:58 | sean-k-mooney | dansmith: by the way is your ephemeral object class merged in ovo yet? | |
| 14:18:30 | sean-k-mooney | i would like to use it to clean up this crap https://github.com/openstack/os-vif/blob/master/os_vif/objects/network.py#L41-L47 but i might just copy the base class for now like we did in nova if it is not merged | |
| 14:21:08 | francoisp | hello lyarwood, stephenfin, doing code rework for https://review.opendev.org/#/c/669674/ and I have a question for you: | |
| 14:21:09 | francoisp | I understand we want do do away with the config option, and stick with a hard-coded value for max number of re-tries. | |
| 14:21:09 | francoisp | "If something is failing to complete after 5 loops, surely that's an issue that bumping the value will only mask?" | |
| 14:21:09 | francoisp | stephenfin you put in a comment at nova/conf/compute.py, line 871: | |
| 14:21:10 | francoisp | are you suggesting we use a lower value (ex.: max 3 retries) ? | |
| 14:22:37 | stephenfin | no, I'm only questioning the need to make this configurable rather than hard code it | |
| 14:22:55 | stephenfin | 5 is fine. I'm don't think we'd want to go bigger than that though | |
| 14:22:57 | stephenfin | francoisp: ^ | |
| 14:23:00 | efried | sean-k-mooney: https://review.opendev.org/#/c/702049/ | |
| 14:23:25 | sean-k-mooney | efried: cool thanks that is the one i was looking for | |
| 14:23:39 | sean-k-mooney | efried: so ya ill just bump the min version and use that | |
| 14:23:48 | sean-k-mooney | ill wait till after m2 however to do that | |
| 14:23:49 | efried | That's in nova, no bump | |
| 14:24:01 | sean-k-mooney | oh sorry i taught that was the ovo one | |
| 14:24:21 | sean-k-mooney | os-vif does not import form nova so i either need to copy that or wait for it to merge in ovo | |
| 14:24:54 | sean-k-mooney | this is the ovo review https://review.opendev.org/#/c/701795/ | |
| 14:24:57 | efried | oh, yeah, the ovo one hasn't merged yet https://review.opendev.org/#/c/701795/ | |
| 14:25:00 | efried | yeah | |
| 14:26:23 | sean-k-mooney | looks like dansmith's -w was to wait for concensus on the nova review but since we have merged it i think we are good provided this version matches the nova one | |
| 14:27:45 | sean-k-mooney | efried: ill proably submit a relase patch for os-vif on friday ahead of m2 next week and ill add you once its up | |
| 14:27:45 | dansmith | sean-k-mooney: I was waiting for us to land it in nova and be happy with that definition | |
| 14:28:10 | dansmith | I haven't circled back since I got back from being out, and I thought you were going to think of better names... :) | |
| 14:28:58 | sean-k-mooney | we could. i know i suggested that but i havent really found one that is better | |
| 14:30:36 | sean-k-mooney | ill likely just submit the patches to os-vif to use it and leave the bottom one with -w until your happy with it then | |
| 14:33:44 | gibi | dansmith: hi! updated and responded in https://review.opendev.org/#/c/702261/ | |
| 14:33:47 | ignaziocassano1 | Hello All, anyone could help me on openstack queens cinder with unity iscsi driver ? Often when a vm is migrated or a volume is detached , nova compute log resports errors with the command multipath -f comman (map in use) | |
| 14:35:24 | sean-k-mooney | that ^ sound vaguly familar but maybe lyarwood would know if that is a know issue that is maybe fixed | |
| 14:36:00 | dansmith | gibi: ack, will look post-caffeination | |
| 14:36:05 | gibi | dansmith: thanks | |
| 14:37:15 | efried | sean-k-mooney: Am I to understand that blockdiag accepted your fix? | |
| 14:37:15 | lyarwood | sean-k-mooney: I think that came up on the ML recently and I asked for logs | |
| 14:37:23 | lyarwood | ignaziocassano1: can you write that up in a bug with logs? | |
| 14:37:44 | sean-k-mooney | efried: they merged a similar one that does the same thing but also updates the change log | |
| 14:37:48 | sean-k-mooney | so basically yes | |
| 14:38:12 | efried | sean-k-mooney: okay, cool, so we should blacklist 2.0.0 in u-c and accept, what, 2.0.1 once they cut it? | |
| 14:38:28 | sean-k-mooney | ya basically | |
| 14:38:49 | sean-k-mooney | we can blacklist it now if we want too but i dont know what the eta is on the 2.0.1 release is | |
| 14:39:15 | efried | Well, we know 2.0.0 is busted, but it also doesn't seem to be affecting anyone but us, so... | |
| 14:39:22 | sean-k-mooney | oh its already done | |
| 14:40:56 | sean-k-mooney | https://github.com/blockdiag/blockdiag/compare/2.0.0...2.0.1 | |
| 14:40:58 | ignaziocassano1 | lyqrwood: I am using centos 7.....how can open a bug ? Which site must I use ? | |
| 14:41:20 | sean-k-mooney | efried: want me to submit the patch to the requiremetns repo | |
| 14:42:00 | lyarwood | ignaziocassano1: https://bugs.launchpad.net/os-brick/+filebug - if you could include DEBUG logs showing the volume attachment and attempt to detach that would be super useful | |
| 14:42:17 | ignaziocassano1 | thanks | |
| 14:44:26 | openstackgerrit | Stephen Finucane proposed openstack/nova master: nova-net: Remove unused nova-network objects https://review.opendev.org/697156 | |
| 14:44:27 | openstackgerrit | Stephen Finucane proposed openstack/nova master: nova-net: Remove unused parameters https://review.opendev.org/703974 | |
| 14:45:20 | efried | sean-k-mooney: sure | |
| 14:45:51 | sean-k-mooney | ok ill do it shortly once i finish my current review | |
| 14:47:09 | sean-k-mooney | efried: did we already merge stephens fix | |
| 14:47:19 | efried | yes | |
| 14:47:36 | efried | sean-k-mooney: https://review.opendev.org/#/c/705263/ | |
| 14:48:06 | sean-k-mooney | cool im just going to blacklist it in gr we could raise the min version in nova if we want to removed the monkey patching seperatly | |
| 14:48:36 | sean-k-mooney | im not going to raise the lower constratin in the requirements repo. | |
| 14:50:00 | ignaziocassano1 | lyarwood: I opened a bug. Did you know if I will receive email when someone will work on it ? | |
| 14:50:33 | lyarwood | ignaziocassano1: can you share the bug URL here? | |
| 14:50:46 | openstack | Launchpad bug 1861691 in os-brick "openstack queens emc iscsi errors migrating volume and or detaching volumes" [Undecided,New] | |
| 14:50:46 | ignaziocassano1 | https://bugs.launchpad.net/os-brick/+bug/1861691 | |
| 14:51:01 | ignaziocassano1 | I attached the nova-compute log | |
| 14:51:44 | ignaziocassano1 | mmm I must change the description | |
| 14:52:41 | ignaziocassano1 | It happens when I migrate a vm or detach a volume | |
| 14:53:25 | lyarwood | ignaziocassano1: do you have any logs showing the volume being attached? | |
| 14:54:36 | ignaziocassano1 | I included the nova-compute log ...the problem occurs non kvm node. I also have the volume log. Dow you want I attach it ? | |
| 14:57:17 | ignaziocassano1 | When I migrate a vm , os-brick try to use the command "multipath -f xxxxxxxx" to flusj the device. As reported in nova compute log, it finds "map in use" | |
| 14:58:21 | sean-k-mooney | efried: oh its already blacklisted | |
| 14:58:29 | efried | cool | |
| 14:58:56 | sean-k-mooney | im not sure we want the upper constratis file to cap it at 2.0.1 however | |
| 14:59:08 | ignaziocassano1 | So, often the vm migrates but remains in error state and horizon shows the vm on old node | |
| 15:00:52 | lyarwood | ignaziocassano1: this channel isn't really for support but yeah `map in use` errors suggest the multipath device is being used by multiple instances, normally due to c-vol or os-brick bugs | |
| 15:01:08 | lyarwood | ignaziocassano1: make sure both nova, os-brick and cinder are fully updated | |
| 15:01:43 | ignaziocassano1 | lyarwood: they are fully updated with last queens released on centos | |
| 15:02:00 | dansmith | efried: I know you have other stuff going on which is probably delaying your ability to run through the cyborg patches. If you could at least hit the profile groups thing, that would be helpful: https://review.opendev.org/#/c/631243 | |
| 15:02:02 | efried | sean-k-mooney: certainly not until that's releaced | |
| 15:02:09 | efried | dansmith: ack | |
| 15:02:30 | dansmith | efried: knowing that latest version of that matches your expectation and what should be done is important | |
| 15:02:33 | dansmith | thanks | |
| 15:03:21 | dansmith | sean-k-mooney: last I heard you had deployed the cyborg set in devstack and had identified some real issue, but I don't know what that was, and it's been a while.. have you done it recently? | |
| 15:03:39 | sean-k-mooney | efried: it is alredy on pypi also its weired i have blockdiag in my local requiremetns but i dont see it on master. ill talk to the release folks and make sure we are good | |
| 15:03:59 | sean-k-mooney | dansmith: no but i can do it again tomorrow | |
| 15:04:06 | ignaziocassano1 | lyarwood: I am sure the device is not used by multiple instances. I tried substituing the command "multipath -f" with "mutipath -r" in /usr/lib/python2.7/site-packages/os_brick/initiator/linuxscsi.py and it works . But flush and reload are different operations :-( | |
| 15:04:22 | sean-k-mooney | dansmith: the issue was the patches needed to be rebased for root required and cyboge had a python 3 issue | |
| 15:04:34 | dansmith | sean-k-mooney: ah okay, good to know, thanks | |
| 15:04:38 | sean-k-mooney | dansmith: sundar has rebased the cybog patch chain so it should work | |
| 15:04:48 | dansmith | gotcha, I remember now | |
| 15:04:57 | openstackgerrit | Stephen Finucane proposed openstack/nova master: objects: Remove custom comparison methods https://review.opendev.org/472285 | |
| 15:05:49 | sean-k-mooney | ill kick off devstack stacking in the backgound and if it stack ill let you know what i find. | |
| 15:05:58 | dansmith | thanks | |
| 15:06:08 | ignaziocassano1 | lyarwood: keep in minf all volumes I attach have lvm inside them. Is it possible lvm on kvm node lock them ? | |