Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-11-15
23:26:04 mriedem and we'll just always remove ports that nova created on reschedule
23:26:15 mriedem dansmith: do you recall any reasons why we don't just always remove ports we created when we reschedule?
23:26:39 eandersson Yea - I don't like 335788
23:26:47 eandersson but utils.is_neutron would be awesome for this
23:27:16 eandersson that way if there really is some nova-net edge case we would be covered
23:27:25 dansmith um, I don't remember the exact reason, but I think there is one... like setup time or resources for certain backends maybe?
23:27:52 mriedem the bug reports i've been digging through seem to indicate that certain ML2 backends are cool with us not deallocating ports between reschedules
23:28:00 mriedem however, i think from the nova side,
23:28:09 mriedem we're still creating a port each time we call build_and_run_instance
23:28:20 mriedem so even if the underlying network is tying those together somehow, or ignoring them,
23:28:25 mriedem we're at worst eating up port quota for the project
23:28:28 mriedem for ports that we don't use
23:28:32 mriedem s/use/bind/
23:28:53 dansmith tbh, it's been a long time since I looked at that path, probably before the great compute/conductor split for build
23:29:06 mriedem yeah, so,
23:29:20 mriedem in the interest of our "damn the torpedos" pledge from the ptg
23:29:23 mriedem and just break everything
23:29:37 mriedem i say we add an "is_neutron" check in this case and just always deallocate before reschedluing
23:29:46 mriedem if that breaks some unicorn ML2 backend, let them speak up
23:29:47 dansmith we delete ones we create unceremoniously on delete now?
23:30:03 dansmith and what about offload?
23:30:04 mriedem when the instance is deleted yes we cleanup and delete anything nova created
23:30:34 gmann alex_xu: hi, returned back to tokyo today
23:30:36 mriedem well, as you can guess,
23:30:44 mriedem dansmith: shelve offload won't delete or cleanup the ports at all,
23:30:52 mriedem it only unplugs the vifs via driver.destroy()
23:31:06 mriedem the actual port will still show as being bound to the original host,
23:31:15 dansmith and unshelve can continue using those ports?
23:31:15 mriedem until we unshelve the instance and bind it to a new host
23:31:23 mriedem i assume so...
23:31:40 dansmith I guess I would expect reschedule to do the same thing.. unbind/rebind
23:31:48 mriedem i only have a single node devstack atm so can't really test unshelve to another host and ssh into the guest after it's unshelved
23:31:54 dansmith but using shelve as the poster child is probably a weaker argument than delete
23:32:14 mriedem yeah, anything to do with ports and volumes + shelve is a roll of the dice
23:32:25 mriedem given we didn't actually terminate connetions for the last 4 years for volumes on unshelve
23:32:25 dansmith yeah
23:32:41 dansmith well, doing the same thing as delete is defensible so if that's the plan it seems reasonable
23:32:52 mriedem alright
23:33:12 mriedem i'll see if can wordsmith a reasonable commit message
23:33:24 mriedem because there is also prior art in this area which always confuses me
23:35:11 gmann yikun: re: on https://review.openstack.org/#/c/518644/11
23:36:04 gmann yikun: all param in query are as string so we use type as string, if any param is integer then we can use non_integer or positive_integer schema
23:37:13 gmann yikun: and on additionalProperties=True we kept that for backward compatibility but at validation layer all the additional properties will be ignored and stripout
23:38:17 gmann yikun: if you are doing migration pagination with microversion then you can make it False. anyways i will check the patch series today
23:38:39 mriedem gmann: i updated that one today
23:38:47 mriedem the schema thing anyway
23:39:04 openstackgerrit Matt Riedemann proposed openstack/python-novaclient master: Microversion 2.54 - Enable reset keypair while rebuild https://review.openstack.org/519572
23:39:35 gmann mriedem: it is with version bump ?
23:39:51 mriedem no
23:39:55 mriedem it's all backward compatible
23:39:59 mriedem because additionalProperties=True
23:40:24 mriedem i removed cell_name from the list of query parameters in the schema because it's not used anywhere in the actual nova code,
23:40:29 mriedem novaclient sends it, but shouldn't
23:40:38 mriedem so i'm just treating cell_name like additionalProperties
23:40:49 mriedem in a far flung future microversion we can change all of that
23:42:06 gmann mriedem: i see.
23:42:12 gmann got confuse with https://review.openstack.org/#/q/topic:bp/add-pagination-and-change-since-for-migration-list+(status:open+OR+status:merged)
23:42:34 gmann that patch is just putting schema for existing filters.
23:42:38 mriedem yes,
23:42:46 gmann patch topic was new BP
23:42:47 mriedem it was part of https://review.openstack.org/#/c/330406/ originally and i asked that he split it out
23:42:53 gmann ok
23:43:33 mriedem ah yeah https://review.openstack.org/#/c/330406/58/nova/api/openstack/compute/schemas/migrations.py@47
23:43:43 mriedem so in 2.55 he'd now restrict to the list of known parameters
23:44:01 gmann nice.
23:44:35 gmann and all pagination param are single_param.+1
23:51:50 openstackgerrit Ed Leafe proposed openstack/nova master: Add Selection objects https://review.openstack.org/499239
23:51:51 openstackgerrit Ed Leafe proposed openstack/nova master: Return Selection objects from the scheduler driver https://review.openstack.org/495854
23:51:51 openstackgerrit Ed Leafe proposed openstack/nova master: Modify select_destinations() to return objects and alts https://review.openstack.org/510159
23:51:52 openstackgerrit Ed Leafe proposed openstack/nova master: Change RPC for select_destinations() https://review.openstack.org/516707
23:51:52 openstackgerrit Ed Leafe proposed openstack/nova master: Move the claim_resources method to scheduler utils https://review.openstack.org/511357
23:51:53 openstackgerrit Ed Leafe proposed openstack/nova master: Make conductor pass and use host_lists https://review.openstack.org/511358
23:51:53 openstackgerrit Ed Leafe proposed openstack/nova master: Refactor the code to check for sufficient hosts https://review.openstack.org/520242
23:52:12 edleafe mriedem: ^^ refactored
23:53:38 efried jaypipes I find it very confusing that _Provider represents a node in a tree of providers, but ProviderTree actuall represents any number of such trees.
23:54:12 efried Rename ProviderTree to ProviderTrees. Or ProviderBush.
23:56:38 edleafe efried: ProviderVine?
23:57:06 efried edleafe ProvideMeVino
23:57:24 edleafe I like the way you think!
23:57:38 eandersson btw does placement take actual resources into account when scheduling vms on overcommited hosts?
23:59:04 eandersson In Mitaka we are hitting this when the compute is out of memory, but the scheduler still thinks there is room available
23:59:05 eandersson > qemu-kvm: cannot set up guest memory 'pc.ram': Cannot allocate memory
23:59:08 openstackgerrit Eric Fried proposed openstack/nova master: Iterator for a provider tree https://review.openstack.org/520243
23:59:28 efried jaypipes We're gonna need that ^
23:59:42 mriedem eandersson: are you setting the reserved space config?
23:59:51 eandersson Yea
#openstack-nova - 2017-11-16
00:00:00 eandersson This is with overcommit 1.5 on memory, which is fine in most cases for us
00:00:02 mriedem reserved_host_memory_mb
00:00:04 mriedem ^
00:00:15 eandersson The problem is that if one or two VMs are using all their memory
00:00:24 eandersson the overcommit ratio won
00:00:26 eandersson won't work
00:00:44 mriedem the reserved_host_memory_mb config option is used to tell placement that there is reserved space on a compute that can't be claimed by nova
00:00:49 mriedem and the scheduler will take that into account
00:01:30 eandersson Lets say you have 10GB memory, and 4 VMs with 5GB each and overcommit set to 2x (total 20GB)
00:02:00 eandersson or even better 12GB memory, 4 VMs with 5GB etc and reserved_host_memory_mb set to 2GB
00:02:29 eandersson If three of those VMs are actually using 10GB in total
00:02:37 eandersson The forth VM will fail to be created
00:03:32 eandersson but the scheduler will think that there is room on the VM, as it thinks the compute has 20+2GB memory
00:04:23 mriedem have you tried to recreate that in ocata?

Earlier   Later