Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-08-30
15:38:28 efried mriedem: No, cyborg has nothing in nova atm
15:38:46 efried oh, yeah, that's poorly worded.
15:41:16 Kevin_Zheng mriedem: we were talking today, about the batch when listing patch
15:42:24 Kevin_Zheng The batch size could vary depending on sort key and dir
15:42:44 Kevin_Zheng Like for sort with uuid
15:43:01 Kevin_Zheng It could be evenly distributed
15:43:24 dansmith Kevin_Zheng: you mean the *optimal* batch size?
15:43:40 Kevin_Zheng dansmith: yeah
15:44:40 dansmith Kevin_Zheng: obviously if you're sorting by uuid it should be fairly evenly distributes and sorting by other things could be massively less well-distributed, but I'm not sure how we could (efficiently) optimize that at runtime
15:44:45 dansmith Kevin_Zheng: do you have ideas?
15:45:00 Kevin_Zheng I was thinking a tool that analyzes dB data
15:45:16 Kevin_Zheng And feed to nova periodically
15:45:30 Kevin_Zheng But that seems to much :)
15:45:36 dansmith yeah :)
15:46:08 dansmith it likely varies by tenant, sort_key, cloud layout, scheduler weights, etc
15:46:17 Kevin_Zheng Just came up when we introduced the new approach to our product team
15:46:18 dansmith it'd be hard to pin that down except for a single-tenant cloud I think
15:46:55 sean-k-mooney mriedem: today i belive you can use cyborg to program a pci device and then you can use nova to pass it through via a pci passthrough flavor alisa but there is no way to force landing on the host with the device you just programed
15:46:57 Kevin_Zheng I’d say it is a powerful tool:)
15:47:33 dansmith Kevin_Zheng: if the goal is to get larger batches from cells likely to have many results, we could do things like scale up the batch size each time you hit a cell again
15:48:03 dansmith Kevin_Zheng: so that if your query is likely to get most results from one cell, we get $batch_size, then $batch_size*2, then $batch_size*4, etc
15:48:25 dansmith but I think I'd want to see a benchmark showing that as worthwhile before I approved it,
15:48:34 Kevin_Zheng Hmm
15:48:51 Kevin_Zheng That could be a good way
15:48:51 dansmith because I expect that since the db query time is so small compared to the processing time, I'm not sure it matters that much (even your hyper-optimized batch sizing tool :)
15:49:26 Kevin_Zheng Yeah, they are just guessing as always
15:49:49 dansmith Kevin_Zheng: yeah :D
15:50:27 dansmith Kevin_Zheng: it's common trap: One big optimization on batch size gives 60% improvement, so assume there are more 60% improvements to be gained through hyper-optimization :)
15:50:34 openstackgerrit Merged openstack/nova-specs master: VMware: add support for live migration https://review.openstack.org/598163
15:59:27 mriedem leave some optimizations for the enterprise fellas
16:01:51 mriedem gibi: you were +2 on this before i robustified the test per mel's prodding https://review.openstack.org/#/c/588943/
16:03:51 Kevin_Zheng maybe left some place for them to be able to do that, like a call to my powerful tool backend :P
16:04:53 mriedem is the toronto lab already working on that?
16:05:15 mriedem research people gotta get grant money somehow
16:05:37 sean-k-mooney mriedem: i will likely be fixing a few things in cyborg in the near future. do you want me to fix the releases note regarding nova inetgration
16:05:43 mriedem need i remind everybody https://www.openstack.org/videos/vancouver-2018/revisiting-scalability-and-applicability-of-openstack-placement-1
16:05:58 mriedem sean-k-mooney: i guess?
16:06:05 mriedem revising release notes is sometimes a tricky business
16:06:47 sean-k-mooney well we update specs retroactivly i done really see release notes as any different
16:07:01 mriedem because release notes are built from git history
16:07:04 mriedem specs are not
16:07:17 stephenfin sean-k-mooney: Any reason we don't squash these? https://review.openstack.org/#/q/topic:bug/1759420+(status:open+OR+status:merged)
16:07:50 sean-k-mooney stephenfin: i wanted to specifcally demonstrate that the behavior was wrong
16:07:59 sean-k-mooney other then that no
16:08:05 Kevin_Zheng No, Xian lab can work on that:)
16:08:37 stephenfin sean-k-mooney: I'm guessing if we reverted the functional part then we'd see the test fail, right? Any chance you could squash them?
16:10:27 sean-k-mooney stephenfin: sure but i need to go get my car NCT tested so ill do it later this evening/tomorow
16:10:34 stephenfin sean-k-mooney: all good
16:10:41 sean-k-mooney anything else you want me to change while im doing it?
16:11:23 sean-k-mooney stephenfin: i might add mel's notes as comments too
16:11:47 mriedem mdbooth: are you ok with the wording here? https://review.openstack.org/#/c/596492/
16:12:01 sean-k-mooney anyway got to run.
16:14:07 cdent sean-k-mooney: my MOT (which I guess is the same thing) is tomorrow and it's almost certainly going to fail
16:18:14 stephenfin sahid: I've still got open comments on https://review.openstack.org/#/c/532168/
16:27:51 stephenfin lyarwood: Can I move this to MODIFIED too? I'm not sure what the process is for non-hotfixes as I didn't have to kick off any builds myself https://bugzilla.redhat.com/show_bug.cgi?id=1187945
16:27:52 openstack bugzilla.redhat.com bug 1187945 in openstack-nova "[RFE] Take into account NUMA locality of physical NICs when plugging instance VIFS from Neutron networks" [Urgent,Post] - Assigned to sfinucan
16:29:06 mnaser so i never ended up doing the full clean up from the stale cell stuff
16:29:37 mnaser but if i have instances with an instance_mapping entry, no build_request, they don't exist in any cells (cell0 or anything else), i can just drop the instance_mapping entry to get rid of it from the listing?
16:29:58 lyarwood stephenfin: ^_^
16:30:16 dansmith mnaser: yeah
16:30:17 dansmith mnaser: that should be the case for any instances you've deleted and then purged from the db
16:30:18 dansmith if you've done that
16:30:39 mnaser dansmith: yeah they're not even purged, cell_id = NULL too
16:30:39 dansmith recently archive started nuking the BR at least
16:30:41 dansmith not sure about the mapping
16:30:55 dansmith oh okay well, if they're really gone there's no need for the mapping
16:30:59 stephenfin lyarwood: 🙈
16:31:11 mnaser this was a whole thing related to the adding entries into nova_api in a single transaction
16:31:27 mnaser which i think i put a patch that i *think* works but i dont know how to test that it works in a single transaction
16:32:23 mnaser https://review.openstack.org/#/c/586824/1 was supposed to be backportable interim solution to avoid listing stuff that shouldnt be there and https://review.openstack.org/#/c/586742/2 was the more fundamental fix but i havent had time to look over them more
16:42:17 melwitt sahid: your review would be appreciated on this bug fix for handling disk_bus for root disk https://review.openstack.org/584999
16:42:45 sahid stephenfin: surprising that it I did not noticed them :)
16:43:26 sahid melwitt: sure i will do that
16:43:32 melwitt thanks
17:19:41 cfriesen in nova/compute/flavors.py we call "from nova.api.validation import parameter_types". This appears to be really expensive (~6 seconds in a recent test) due to the regex stuff. One possibility would be to do the import right before the flavor creation so that it doesn't impact all nova processes. Thoughts?
17:21:14 openstackgerrit Merged openstack/nova master: reshaper gabbit: Nix comments re doubled max_unit https://review.openstack.org/597220
17:35:10 openstackgerrit Merged openstack/nova master: Fix race condition in reshaper handler https://review.openstack.org/596497
17:37:31 sean-k-mooney melwitt: im just back, am would you like me to squash those two disk bus patches together?
17:38:23 sean-k-mooney melwitt: i used the functional regression style partly to prove to my self that the test case was corret since you pointed out my orginial test case worked without the patch applied
17:38:32 melwitt sean-k-mooney: not right now, maybe only if you need to respin. I don't have a strong opinion about it, just pointing it out
17:38:53 melwitt yeah, understood
17:42:52 sean-k-mooney cdent: just got back and ya MOT and NCT are basically the same.
17:43:23 sean-k-mooney cdent: happily in my case it passed the second time.
17:44:13 sean-k-mooney cdent: that said i dont drive my car enough i have only done 8000KM/5000 miles in the last two years...
17:44:17 cdent sean-k-mooney: my car is 21 years old. the emissions check is going to be an issue, I fear. Apparently the trick is to take it in to the test good and hot after racing around like a crazy person
17:45:02 sean-k-mooney cdent: if its 21 years old it shoudl qualify as a vintage car now right?
17:45:26 cdent hmm, that's a good point.
17:46:06 sean-k-mooney i cant remeber what the cut off is in ireland but there is an emaitions cut off at some point in ireland where provided you have converted from lead based fule to unleeded the co2 emmsions are basically ignored
17:46:14 openstackgerrit Merged openstack/nova master: Report client: get_allocations_for_provider_tree https://review.openstack.org/584648
17:47:46 cdent sean-k-mooney: it looks like it may be 40 years here :(
17:50:33 sean-k-mooney cdent: well hopfully it will keep running that long :)
17:53:00 cdent i can only try
17:53:30 cdent sean-k-mooney: in other vaguely related to sean-k-mooney news: I'm sending in my applicaiton for an irish passport today
17:54:11 sean-k-mooney oh. cutting it a little close with brexit no?
17:54:35 cdent i had to get a hold of my mother's birth certificate
17:56:09 sean-k-mooney ya i love that to get a pass port which is ment to be the most secure id you can get in the contry you need a copy of your birth cert which is the only id i have that cant even be used to by alcohol
17:56:44 cdent \o/
17:56:54 sean-k-mooney i know they use it in thery to prove that you our your parent are entiled to citezenship in this case but still
18:00:17 sean-k-mooney ok time for food. laters o/
18:19:29 melwitt dansmith or jaypipes: could one of you hit this to move rocky implemented specs? https://review.openstack.org/592622

Earlier   Later