Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-14
11:40:05 openstackgerrit Francesco Santoro proposed openstack/nova: os-vif: convert libvirt driver to use os-vif for fast path vhostuser https://review.openstack.org/385061
11:55:23 openstackgerrit Francesco Santoro proposed openstack/nova: os-vif-util: set vif_name for vhostuser ovs os-vif port https://review.openstack.org/410737
11:55:23 openstackgerrit Francesco Santoro proposed openstack/nova: os-vif: convert libvirt driver to use os-vif for fast path vhostuser https://review.openstack.org/385061
14:47:05 openstackgerrit OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/420345
15:24:45 weshay_mod anyone around who can review https://review.openstack.org/#/c/399735/
15:55:34 openstackgerrit Matt Riedemann proposed openstack/nova: [placement] Add an "Adding a Handler" section to placement_dev https://review.openstack.org/416304
15:55:35 openstackgerrit Matt Riedemann proposed openstack/nova: [placement] placement_dev info for microversion handling https://review.openstack.org/411946
15:55:35 openstackgerrit Matt Riedemann proposed openstack/nova: [placement] placement_dev info for testing and gabbi https://review.openstack.org/412600
15:55:36 openstackgerrit Matt Riedemann proposed openstack/nova: [placement] Add a bit about extraction plans to placement_dev https://review.openstack.org/416676
16:10:13 openstackgerrit Dirk Mueller proposed openstack/nova: [WIP] Testing latest u-c https://review.openstack.org/409077
18:09:48 openstackgerrit Davanum Srinivas (dims) proposed openstack/nova: Fixes to get all functional tests working on py35 https://review.openstack.org/419968
19:12:44 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:20:44 dansmith mriedem: I kinda think we should fix that nova_api_cell0 silliness, and I'm happy to work the two-step dance with devstack, but I left it the way it was until you comment
19:58:19 openstackgerrit Dan Smith proposed openstack/nova: Make api_samples tests use simple cell environment https://review.openstack.org/410473
19:58:19 openstackgerrit Dan Smith proposed openstack/nova: Multicell support for instance listing https://review.openstack.org/396775
19:58:23 openstackgerrit Dan Smith proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379
#openstack-nova - 2017-01-15
09:25:45 openstackgerrit OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/420398
17:19:41 mriedem dansmith: yeah i don't know why we default the cell0 db connection to the api db connection if one isn't provided in map_cell0, i never made that connection before you pointed it out
17:19:51 mriedem i guess we haven't gotten to the point of using it yet?
17:20:38 dansmith mriedem: yeah, we do, but devstack just has the same default of nova_api_cell0
17:20:55 dansmith so I will cook up a set of changes to fix that I guess
17:21:02 dansmith 'cause I think that needs to change
17:21:10 mriedem ot
17:21:15 mriedem it's also grenade http://git.openstack.org/cgit/openstack-dev/grenade/tree/projects/60_nova/from-newton/upgrade-nova#n8
17:21:51 mriedem if we're sending instances to cell0 already how are we not exploding when the instances table isn't in that schema?
17:21:57 mriedem or are we sending build_requests there?
17:22:45 dansmith no, that's the thing.. we call it "nova_api_cell0" but put main db schema into it
17:22:55 dansmith that's why we don't explode and that's why it's super confusing
17:26:07 mriedem oh i see http://git.openstack.org/cgit/openstack/nova/tree/nova/cmd/manage.py#n1104
17:26:45 mriedem that would have to pass database='api' to make it the api db schema
17:26:53 mriedem ok yeah so that's an easier fix
17:27:18 mriedem i think we can just rename the db in grenade/devstack since cell0 isn't used in newton anyway, and then fix the conf option we use in map_cell0
17:27:22 mriedem i'll open a bug
17:28:02 dansmith mriedem: the easiest thing to do would be to merge the create_cell change, then I can make devstack and grenade use the same exact code path, and change the db name at the same time
17:28:36 dansmith sorry,
17:28:48 dansmith I meant make devstack and grenade take the same code path for cellsv1 and cellsv2
17:30:19 openstack Launchpad bug 1656673 in OpenStack Compute (nova) "map_cell0 should use the main DB connection, not the API DB" [High,Confirmed]
17:30:19 mriedem https://bugs.launchpad.net/nova/+bug/1656673
17:31:17 mriedem sounds like i should also work on creating a list_cells CLI today
17:32:10 dansmith well, it's not required for any of this, but I agree we should have that for clarity
17:33:30 mriedem not for any of *this* right, but several commands take a cell uuid, like i discover_hosts
17:34:01 dansmith well, discover hosts doesn't need it, but yes, agreed
17:34:49 mriedem ok, i'm still unfed and unshowered and therefore unusable, so i'll do those in awhile
17:39:15 dansmith mriedem: those? or just list_cells?
17:39:28 dansmith I think I can do the devstack one without too much drama
17:39:36 dansmith and I am fed, caffeinated
17:57:29 mriedem sorry i meant i was going to work on list_cells
17:57:33 mriedem leaving you all of the hard work
17:57:44 dansmith okay, all good
17:57:46 mriedem oh i'm caffeinated,
17:57:49 mriedem that's top priority
17:58:02 dansmith heh
18:46:59 openstackgerrit Dan Smith proposed openstack/nova: Fix the generated cell0 default database name https://review.openstack.org/420439
18:46:59 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:02:34 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:02:35 openstackgerrit Dan Smith proposed openstack/nova: Fix the generated cell0 default database name https://review.openstack.org/420439
19:04:58 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:05:16 dansmith I think the devstack fail is related to the failure to talk to neutron, as I don't see any evidence that it's not finding the cell0 map, but we'll see on recheck
19:08:06 mriedem i'm surprised the devstack change doesn't require the grenade change
19:08:13 mriedem or vice-versa,
19:08:22 mriedem but that's just b/c i'm usually used to this being complicated
19:11:40 openstackgerrit Matt Riedemann proposed openstack/nova: Add nova-manage cell_v2 list_cells command https://review.openstack.org/420440
19:11:50 dansmith I dunno, I thought the point of the changes in grenade was to insulate them from devstack
19:12:09 dansmith remember the first version of one of those, I used the method from devstack and sdague said I should just put the commands themselves in there
19:12:17 dansmith I have to run off in a little bit, but I will check on these when I get back
19:18:01 openstackgerrit Matt Riedemann proposed openstack/nova: Remove nova-manage image from man pages https://review.openstack.org/420442
19:40:10 mriedem melwitt: dansmith: it occurs to me that we're going to need a HostMappingList object to get host mappings by cell mapping, to be used in a nova-manage cell_v2 delete_cell command to see if the cell mapping is empty or not
19:42:12 mriedem but i'm not sure if we're going to need to backport the delete_cell command to newton or not, if not then i can make it use a new HostMappingList object, otherwise i have to write it to query the DB API directly from the CLI
19:52:52 dansmith mriedem: I think it's fine to not have a full set of commands in newton
19:53:14 dansmith and like I said, I think it's okay if the simple command serves as the "upgrade from newton" command
19:53:48 mriedem ok
19:56:11 mriedem we might want to fix https://review.openstack.org/#/c/420439/ outside of your series
19:57:12 openstackgerrit Dan Smith proposed openstack/nova: Fix the generated cell0 default database name https://review.openstack.org/420439
19:57:12 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:57:13 mriedem since i'm assuming getting instance creation moved to conductor working is going to be a bit of a beast
19:57:23 dansmith mriedem: yeah, that one isn't actually dependent on anything
19:57:31 dansmith other than the devstack change but that's just for testing
19:57:39 dansmith once we see that work with and without this, we can remove that
19:57:46 dansmith I dunno why gerrit is showing it that way though,
19:57:51 dansmith oh wait
19:58:40 dansmith yeah, that was unintentional
19:58:44 openstackgerrit Dan Smith proposed openstack/nova: Add some cellsv2 setup docs https://review.openstack.org/420198
19:58:45 dansmith fixed ^
19:58:45 openstackgerrit Dan Smith proposed openstack/nova: Fix the generated cell0 default database name https://review.openstack.org/420439
19:58:53 dansmith gotta run now, back in a bit
20:04:13 mriedem ok i'm going to start on delete_cell, hopefully crank that out by 3:30 at which point i have to stop for football
21:13:32 mriedem something blowing up in grenade http://logs.openstack.org/39/420439/4/check/gate-grenade-dsvm-neutron-multinode-live-migration-nv/89fb1cf/logs/new/screen-n-api.txt.gz?level=TRACE#_2017-01-15_20_32_29_818
21:13:40 mriedem DBNonExistentDatabase: (pymysql.err.InternalError) (1049, u"Unknown database 'nova_cell0'")
21:14:01 mriedem which is why i think the devstack change will need to depend on the grenade change probably
21:36:20 openstackgerrit Matt Riedemann proposed openstack/nova: Add InstanceMappingList.get_by_cell_id https://review.openstack.org/420450
21:36:20 openstackgerrit Matt Riedemann proposed openstack/nova: Create HostMappingList object https://review.openstack.org/420449
21:36:21 mriedem dansmith: melwitt: delete cell command and friends ^
21:36:21 openstackgerrit Matt Riedemann proposed openstack/nova: Add nova-manage cell_v2 delete_cell command https://review.openstack.org/420451
21:47:07 dansmith "have to stop for football"
21:48:23 mriedem dansmith: dvr
21:48:31 mriedem dansmith: so i'm updating your nova cell0 rename thing right now
21:48:39 mriedem and adding a release note and depends-on the grenade change
21:49:01 mriedem dansmith: we also unfortunately have some existing renos that say the cell0 db name has to be the <api db name>_cell0
21:49:04 mriedem oops
21:49:48 dansmith okay just got back and reading stuff

Earlier   Later