Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-01-29
18:55:46 openstackgerrit Balazs Gibizer proposed openstack/nova master: Reduce gen conflict in COMPUTE_STATUS_DISABLED handling https://review.opendev.org/704866
18:55:46 openstackgerrit Balazs Gibizer proposed openstack/nova master: Repro gen conflict in COMPUTE_STATUS_DISABLED handling https://review.opendev.org/704865
18:56:37 sean-k-mooney efried: i know we have customer request for numa aware gpus/vgpus but we agreed even internally that that should wait until after the memory/cpu case is done
18:56:52 sean-k-mooney so im fine with that
18:56:54 efried yeah, sure, we're not going to not do it, we're just not going to do it now.
18:57:03 sean-k-mooney yep
18:57:45 efried sean-k-mooney: I'm going to put some examples in the etherpad, but will need you to go through and edit my spelling on the extra specs.
18:58:09 sean-k-mooney hehe i will try
18:58:46 efried I figure if I ask you to edit the spelling of real tokens I have a better chance.
18:58:49 efried vs. English.
18:59:01 sean-k-mooney yes you do
18:59:06 efried ye gods, it just occurred to me: do you speak gaelic?
18:59:35 efried your writing in gaelic must be IMPOSSIBLE to understand, given that gaelic is impossible to read in the first freakin place.
19:00:06 sean-k-mooney yes and no. i went to an all irish play school and got an irish langage exemption becaue i coudl not write in irish
19:00:15 sean-k-mooney i could speak it and could kind of read it
19:00:16 efried "oh, let me just throw an extra consonant and an extra vowel in each syllable. Half of them will be silent. Which ones? Who knows? It's a surprise!"
19:00:50 sean-k-mooney hehe irish is interesting that way
19:01:18 sean-k-mooney espcially since spelling of a verb chagne with gender pluarlity and something the noun it is used ith
19:05:53 sean-k-mooney actully do you want that to be a numa instance example or not
19:06:02 sean-k-mooney i assume a numa one
19:06:38 sean-k-mooney a non numa one would not have hw:mem_page_size set at all and not triat
19:10:21 efried eandersson: can you write me a flavor example with two numa nodes where one has small pages and one has large?
19:10:34 efried whoops, eandersson disregard. ^ was for sean-k-mooney
19:10:38 sean-k-mooney no
19:10:44 sean-k-mooney pagesize is vm wide
19:10:47 efried (my IRC client has been dropping the first char of my messages lately)
19:11:00 efried okay, fine, can you write me a flavor with two numa nodes and an explicit page size?
19:11:24 sean-k-mooney yes ill add 2
19:11:31 efried (turns out 'sean<tab>' minus the first character is 'ean<tab>')
19:12:35 sean-k-mooney hehe ill remove our disscution on the first example to clean up the description
19:16:43 sean-k-mooney you can also split the cpu asemetircly but that is overkill for the exampels i think
19:17:27 sean-k-mooney for the implict splitign its an error if the cpus and ram and not evenly devisable by the number of numa nodes
19:19:16 efried sean-k-mooney: I prefer leaving that third example as it was before, since it shows that we could land on either of the LARGE RPs.
19:19:25 sean-k-mooney ya
19:19:25 efried ++
19:19:30 sean-k-mooney that is why i reverted it
19:19:47 sean-k-mooney any why i used large in the first place
19:19:59 efried sean-k-mooney: any other examples that we should throw in?
19:20:11 sean-k-mooney am maybe a non numa guests
19:20:58 efried yeah
19:21:17 efried to show that it would not land on the sample host at all.
19:22:28 sean-k-mooney yes
19:22:37 sean-k-mooney i was trying to figure out the not num host
19:22:40 sean-k-mooney *numa
19:23:44 sean-k-mooney if we did not add the forbiden trait
19:23:49 sean-k-mooney then it could actully land on that host
19:24:00 sean-k-mooney but we would have to pin it to a numa node
19:24:13 sean-k-mooney which we could do but we proably dont want too
19:24:28 sean-k-mooney so you can continue to create vms that span numa nodes if you dont care
19:24:30 efried right, that's what I was implying earlier when I said we could probably overlap, but should prevent that for now
19:24:40 sean-k-mooney ya
19:24:40 efried oh, that wouldn't allow you to span numa nodes.
19:24:52 sean-k-mooney ya it would not you are right
19:24:56 efried oh, I see, you mean your proc from one and your mem from another
19:24:57 sean-k-mooney not without can split
19:25:20 sean-k-mooney you could have procs and memory span ya
19:25:33 sean-k-mooney but can_split would be need to have any one resouce span numa nodes
19:25:42 efried yeah, actually, the non-NUMA example can be simplified if we just use a granular group for the proc & mem....
19:26:01 sean-k-mooney ya
19:26:04 sean-k-mooney that would work
19:26:32 efried simpler, I like.
19:26:36 sean-k-mooney oh we are missign something from all of them
19:26:44 sean-k-mooney we need to add group_policy=none
19:26:58 sean-k-mooney because we are using granular groups
19:27:58 sean-k-mooney we still get the correct affintiy because fo same tree in the multi numa case
19:28:09 sean-k-mooney even with group_policy=none
19:28:56 efried yeah, I was just going to check whether we in fact did remove group_policy in the latest microversions.
19:29:16 efried it's irrelevant in this case. We get the same result with isolate or none
19:29:21 sean-k-mooney looking at the api doces its still there
19:29:35 efried okay, boo. But as noted above, it's irrelevant, so set it to whatever. Do you agree?
19:29:56 sean-k-mooney it can be in this case
19:30:03 sean-k-mooney but isolate breaks eaisly
19:30:15 sean-k-mooney for example 2 port with bandwith requests
19:30:23 sean-k-mooney so i dont like publicising its use
19:30:43 efried Right, so the point is, we can leave it to the user and/or the bandwidth code to decide on group_policy.
19:30:52 sean-k-mooney yes
19:30:57 efried IIRC the bandwidth code is defaulting it rn.
19:31:07 efried We can ask gibi to confirm
19:31:07 efried or at least we talked about doing that.
19:31:16 sean-k-mooney rn as in none?
19:31:31 efried 'right now'
19:31:47 sean-k-mooney oh i think its defaultin to none if it is
19:31:56 sean-k-mooney that should be our default unless you say otherwise
19:32:25 sean-k-mooney ok so it required becasuee we are using groups but the value is not relevent to the spec
19:32:30 sean-k-mooney im fine with that
19:32:39 sean-k-mooney its an implemenation detail
19:33:18 efried yeah. That was a gripe I had during the nested magic design, that group_policy should default (can't remember which way I said) at this point because we can control everything we need using other, more granular (heh) mechanisms.
19:33:24 efried but we didn't actually do that.
19:33:33 efried And, confirmed, we're defaulting to 'none' these days.
19:33:43 efried I found the reno yaml, lemme find where it is in the docs.
19:34:10 sean-k-mooney well really we should not have it be global either
19:34:21 sean-k-mooney it should be per set of groups or per subtreee or something
19:34:27 efried https://docs.openstack.org/releasenotes/nova/train.html#other-notes
19:34:28 sean-k-mooney anyway out of scope
19:35:04 efried right, that was my gripe. We now have same_subtree to dictate which can be shared and which can be spread. Because same_subtree also devolves to same rp in relevant cases.
19:35:34 sean-k-mooney ya
19:36:12 sean-k-mooney provide we model nics with an RP per PF same_subtree or the negation should hanel that too
19:36:37 sean-k-mooney so we might be able to remvoe the group polices thing at some point
19:37:04 sean-k-mooney anyway if this does not require it to make it work it makes me happy
19:40:15 sean-k-mooney i have saved a copy of that etherpad locally just in case by the way.

Earlier   Later