Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-01-16
15:27:18 sean-k-mooney but if you have the choice between not being able to use an entire socket
15:27:35 sean-k-mooney or slightly reduced performace im guess your ok with the slight perfromace hit
15:27:48 sean-k-mooney and slightly in accurate tracking
15:28:49 sean-k-mooney belmoreira: my conser is only for instace with a numa toplogy that dont use hugepages
15:29:26 belmoreira got it
15:29:47 sean-k-mooney hugepage guess will be tracked properly in either case
15:30:14 belmoreira thanks a lot for this discussion and proposal
15:32:22 Sundar sean-k-mooney: Please make sure to pull in https://review.opendev.org/698846 and https://review.opendev.org/700261
15:32:45 Sundar They are called out as dependencies in the create/bind patch
15:33:14 Sundar i.e. https://review.opendev.org/#/c/631244/
15:33:38 sean-k-mooney i cant have devstack pull in both if they are not in the same patch change
15:33:42 sean-k-mooney *chain
15:33:53 sean-k-mooney so ill have to create a local branch and merge them in
15:34:10 sean-k-mooney i have https://review.opendev.org/#/c/698846/ but not https://review.opendev.org/#/c/700261/
15:34:13 kashyap belmoreira: stephenfin: efried: The kernel issue (https://review.opendev.org/#/c/684375/) has been latent. (Just confirmed w/ a KVM mainatiner.)
15:34:19 Sundar Both are listed as dependencies in the same Nova patch
15:34:42 kashyap sean-k-mooney: (Responded on the change.) Thanks!
15:34:46 sean-k-mooney yes but devstack can only clone one it does not have the zuul logic to create a merge commti from both
15:35:00 sean-k-mooney but i can do it manually and restart the services
15:35:19 stephenfin kashyap: I wonder why there wasn't a sudden spike so? Surely this would have been affecting belmoreira and co. for years now if it's been there forever
15:35:22 sean-k-mooney Sundar: in general depends-on should not be used within the same repo
15:35:46 openstackgerrit Lee Yarwood proposed openstack/nova master: libvirt: Add support for stable device rescue https://review.opendev.org/700812
15:35:46 openstackgerrit Lee Yarwood proposed openstack/nova master: virt: Provide block_device_info during rescue https://review.opendev.org/700811
15:35:47 kashyap stephenfin: Yeah, I don't have a solid answer to that yet :-(
15:35:47 openstackgerrit Lee Yarwood proposed openstack/nova master: docs: Add stable device rescue docs https://review.opendev.org/700837
15:35:50 sean-k-mooney https://review.opendev.org/#/c/698846/ should be rebased on https://review.opendev.org/#/c/700261/
15:36:14 kashyap stephenfin: Let me see the original kernel report I quoted in the commit, maybe there's something in there
15:36:39 sean-k-mooney stephenfin: i have never seen hardware that actully does not have a dma regoin per numa node
15:37:10 openstackgerrit Lee Yarwood proposed openstack/nova master: DNM test stable rescue tempest/devstack changes https://review.opendev.org/700815
15:37:18 sean-k-mooney stephenfin: and i mean ever in the 7+ years i have been working on this stuff
15:37:50 sean-k-mooney so i am not surprised people have not hit this before
15:38:10 stephenfin efried: I think I'm going to have to get you to weigh in on https://review.opendev.org/#/c/696745/, because I _think_ I've addressed the concerns but everyone else is still avoiding it :)
15:38:38 kashyap sean-k-mooney: Ah, interesting
15:38:59 stephenfin It's almost all refactor noise _except_ the rename of 'nova.network.neutronv2.api' to just 'nova.network.neutron', and the subsequent removal of the 'nova.network.api' shim
15:39:35 kashyap sean-k-mooney: I don't have the data about the hardware assmuption here (which hosts might not have the DMA32 region): https://lkml.org/lkml/2018/7/24/152
15:40:26 sean-k-mooney kashyap: i also suspect that htis was not alway the behavior or kvm by the way
15:41:15 sean-k-mooney i proably had been that way for a long time but i susspect it was altered at somepoint
15:41:20 kashyap sean-k-mooney: FWIW, a bit ago I checked w/ Paolo (one of the upstream KVM devs); he confirmed it's "latent"
15:41:23 kashyap Probably
15:42:49 sean-k-mooney maybe i dont knwo why but if feel like if you grabed ubuntu 12.04 or centos 5.4 it likely wont happen
15:42:55 belmoreira stephenfin our batch VMs (running in dedicated compute nodes with NUMA/huge pages aware) are long lived VM
15:43:32 sean-k-mooney kashyap: i just get the feelign like there was a refactor of this at some point but i dont recall why i have that feeling
15:43:36 belmoreira true that we are seeing this issue for awhile. Until now our solution has been to reboot the affected node.
16:00:12 openstackgerrit Lee Yarwood proposed openstack/nova stable/train: Avoid spurious error logging in _get_compute_nodes_in_db https://review.opendev.org/702902
16:06:12 openstackgerrit Balazs Gibizer proposed openstack/nova stable/pike: Error out interrupted builds https://review.opendev.org/687918
16:06:26 openstackgerrit Balazs Gibizer proposed openstack/nova stable/pike: Fix false ERROR message at compute restart https://review.opendev.org/699496
16:10:14 dansmith efried: do you have any opinions on the newer literal set syntax (i.e {1, 2}) ?
16:10:30 dansmith IMHO, it's one of the worst things ever, because it's super confusing and looks like a dict
16:10:59 dansmith just wondering if anyone else would be interested in a developer convention in nova of not adding code with that syntax to our tree
16:11:48 gibi dansmith: I have no problem with the literal set sytnax for me the absense of ':' is enough to notice that it is a set
16:12:06 dansmith it's super easy to not notice that there's no : though
16:12:17 gibi I guess we are different
16:12:20 openstackgerrit Lee Yarwood proposed openstack/nova stable/train: Ensure source service is up before resizing/migrating https://review.opendev.org/701757
16:12:40 dansmith gibi: that's fine, I asked because I was wondering if I'm the only one
16:18:53 artom dansmith, as per your request, I'm not letting you forget https://review.opendev.org/#/c/687404/8
16:20:30 gibi dansmith: I think for me it is mostly about the fact that I used to it. I guess I would be pretty against assignment expression because that would be something I never used before in python
16:21:09 dansmith artom: okay I'm in the middle of one of the cyborg patches, which is making me angry because apparently validating that our new methods are called like we expect them to is far too high a bar
16:21:39 artom dansmith, to adjust nagging frequency, press 1 now ;)
16:22:11 dansmith gibi: I think it's confusing that foo={} is a dict but foo={1} is a set, and easy to confuse foo={1:2} with foo={1,2}, and set comprehensions look a lot like dict comprehensions
16:22:24 dansmith gibi: obviously you *can* tell the difference, it's just very easy to miss, IMHO
16:25:36 gibi I accept your view
16:25:41 dansmith hah
16:26:03 efried dansmith: I personally have no problem with {1, 2} set syntax. A set is a dict with only keys. And I don't have trouble distinguishing between commas and colons.
16:26:17 dansmith aight, well, that settles it then
16:26:26 artom " A set is a dict with only keys" Umm...
16:26:29 efried dansmith: but jaypipes expressed a distaste for it (for the same reasons) so I avoided it... until he was gone.
16:26:46 dansmith oh did he? damn, should have gotten my rule landed then
16:28:29 efried a) I think set((1, 2)) is overly verbose and unnecessary, b) my IDE warns that I should use a set literal, c) I believe set((1, 2)) is also less efficient because it first creates a tuple, then calls a method to turn it into a set.
16:29:34 dansmith yeah, I don't like how verbose set() has to be, and yes there's some overhead but I think it's trivial in this case
16:29:34 efried obv b) I could ignore / turn off, and I'm sure c) is negligible in 99.99% of cases. But given aforementioned not-problem with the literal syntax, it's enough for me.
16:30:05 dansmith ack
16:36:57 efried Is that what you've done?
16:36:57 efried stephenfin: before I dig into that: I'm of the opinion that a) we should make it look like the other service API shims. nova/$service_type/$project_name.py; b) I don't see a need for a generic abstract API layer.
16:41:17 sean-k-mooney dansmith: for what its worth i dislike {} as that allways meen a dict so i dont like useing it to initalise a default. i alway use set() if i want a set in that case
16:41:36 sean-k-mooney well if its empty that is
16:41:57 dansmith well, confirmation that I'm not crazy, at least
16:42:00 efried If set(1, 2) were allowed I would be less grouchy about it.
16:42:09 dansmith or.. "at least as crazy as sean-k-mooney" I guess :P
16:42:38 sean-k-mooney efried you have to do set({...}) right
16:42:41 efried but having multiple meanings for the same operator/punctuation doesn't bother me as long as it's clear from the context, which IMO it is in this case, because colons.
16:42:42 sean-k-mooney that is annoying
16:42:54 efried sean-k-mooney: set($iterable)
16:43:03 sean-k-mooney ya
16:43:16 dansmith stephenfin: same
16:43:16 efried so {...} if that's a set or a dict would work to get you a set of the keys; (...) for a tuple, [...] for a list etc.
16:43:31 sean-k-mooney stephenfin: becasue paython is annoying enough to parse as it is
16:43:51 efried I don't really get why that would be better; but I'm going to guess they chose {} because, as I said, a set is a dict without values.
16:43:59 sean-k-mooney <> i think is technically not equal in python 2
16:44:07 dansmith a set is not a dict with no values :)
16:44:16 stephenfin sean-k-mooney: indeed, it is
16:44:29 stephenfin I've seen that once in my life, I think
16:44:31 stephenfin (docutils)
16:44:41 efried dansmith: totally is. O(1) hashing, just no pointers out of the buckets.
16:44:51 dansmith sean-k-mooney: as hateful as they are, I can imagine them doing that in py3 to break compatibility for fun and profit.. I guess dict literals just got added too early :)
16:44:52 sean-k-mooney although i think its also the same as != and is checking identiy not equality but i might be wrong on that
16:45:07 dansmith efried: totally is not :)
16:45:38 dansmith it may be the same implementation under the covers, but that does not make the statement true :)
16:47:58 efried nova-live-migration seems to be ill. Anyone available to dig in?
16:53:40 dansmith I would think artom would be a good person to maybe start taking on some responsibility for that stuff given his numa LM work
16:54:05 dansmith building some context for the future and what not

Earlier   Later