| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-05-17 | |||
| 23:58:23 | cfriesen | _nw_info_get_ips and _nw_info_get_subnets both take noticeable time | |
| 23:58:32 | cfriesen | for each port | |
| 23:59:28 | mriedem | the first is going to iterate over all fixed_ips for a port and get floating IPs for each fixed_ip | |
| #openstack-nova - 2017-05-18 | |||
| 00:01:32 | mriedem | i wonder if we could bulk query the floating IPs by the list of fixed IPs and the port_id | |
| 00:01:46 | openstackgerrit | melanie witt proposed openstack/nova master: Cache database and message queue connection objects https://review.openstack.org/465042 | |
| 00:02:07 | mriedem | "When you specify multiple filters, the Networking API v2.0 returns only objects that meet all filtering criteria. The operation applies an AND condition among the filters." | |
| 00:02:17 | mriedem | so that wouldn't really work, | |
| 00:02:27 | mriedem | because fixed_ip1 AND fixed_ip2 wouldn't resolve to a floating IP | |
| 00:02:45 | cfriesen | mriedem: we get an event for each vif, and we have to get info for all vifs on each event, so when we have N vifs, the live-migration overhead goes up as N^2 | |
| 00:02:47 | mriedem | we could trim the fat on the fields we actually need to get back | |
| 00:03:07 | cfriesen | mriedem: it'd be better to have the vif specified in the network-changed event | |
| 00:03:24 | cfriesen | mriedem: and avoid having to scan all the vifs | |
| 00:03:37 | mriedem | cfriesen: yeah i'm not sure how feasible that is from the neutron side | |
| 00:03:46 | mriedem | i guess it would depend on how they calculate the network-changed event | |
| 00:03:52 | cfriesen | mriedem: gotta run, family is calling for supper | |
| 00:03:52 | edleafe | I've written a post to the dev list about continuing the discussion on claims in the scheduler: http://lists.openstack.org/pipermail/openstack-dev/2017-May/116949.html | |
| 00:04:00 | edleafe | ^^ cdent jaypipes dansmith bauzas johnthetubaguy | |
| 00:04:27 | edleafe | It has a link to my blog post and the patch series | |
| 00:04:36 | mriedem | edleafe: ack | |
| 00:04:37 | edleafe | And with that, I'm off to PyCon | |
| 00:04:42 | mriedem | have fun | |
| 00:04:46 | edleafe | See you all next week! | |
| 00:04:53 | edleafe | mriedem: oh, it's all work work work | |
| 00:15:21 | mriedem | cfriesen: here is something to look into later - the network-changed event in neutron does have a port in scope, it just doesn't put it in the external event request https://github.com/openstack/neutron/blob/master/neutron/notifiers/nova.py#L161 | |
| 00:15:30 | mriedem | but we should be able to scope that event to the single vif | |
| 00:16:03 | mriedem | cfriesen: it would be helpful if we had a bug to track some of this | |
| 00:41:08 | mriedem | cfriesen: https://bugs.launchpad.net/neutron/+bug/1691602 for tracking | |
| 00:41:08 | openstack | Launchpad bug 1691602 in OpenStack Compute (nova) "live migration generates several network-changed events which lock up refreshing the nw info cache" [Undecided,New] | |
| 00:59:53 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Add policy description for os-host https://review.openstack.org/465458 | |
| 01:18:23 | openstackgerrit | Alex Xu proposed openstack/nova master: Limit the min length of string for integer JSON-Schema https://review.openstack.org/415103 | |
| 01:21:53 | openstackgerrit | Zhenyu Zheng proposed openstack/nova master: Catch exception.OverQuota when create image for volume backed instance https://review.openstack.org/464548 | |
| 01:24:45 | openstackgerrit | Alex Xu proposed openstack/nova master: Limit the min length of string for integer JSON-Schema https://review.openstack.org/415103 | |
| 01:36:36 | dikonoor | pumaranikar: Hi | |
| 01:39:37 | dikonoor | pumaranikar: I am trying to understand if this new support added is still experimental > https://github.com/openstack/nova/blob/master/nova/conf/service_token.py#L43 | |
| 01:39:48 | dikonoor | Can it be turned on ? | |
| 01:40:47 | mriedem | dikonoor: yes it can be turned on for testing, we test with it on in the nova-next CI job | |
| 01:41:08 | mriedem | dikonoor: would actually appreciate someone to run with it in rally or longer-running operation scenarios like with live migration | |
| 01:41:18 | mriedem | and report back if it's helping | |
| 02:16:41 | openstackgerrit | Jay Pipes proposed openstack/nova master: Ensure image metadata when unshelving BFV instance https://review.openstack.org/347063 | |
| 02:22:55 | jaypipes | alex_xu: looks sensible to me: https://review.openstack.org/#/c/353862/23 | |
| 02:23:02 | jaypipes | alex_xu: XenAPI resource tracking fix. | |
| 02:34:46 | openstackgerrit | Alex Xu proposed openstack/nova master: Correct the error message for query parameter validation https://review.openstack.org/465785 | |
| 02:35:28 | openstackgerrit | Tao Li proposed openstack/nova master: Fix the display of updated_at time when using memcache driver. https://review.openstack.org/414463 | |
| 02:38:22 | mriedem | oh bug smashing glory | |
| 02:43:29 | mriedem | kevinbenton: looking at this https://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L331 | |
| 02:43:38 | mriedem | it looks like you can list resources by a list of ids | |
| 02:44:13 | kevinbenton | mriedem: yeah, i was actually just looking through the code that builds the filters | |
| 02:44:18 | kevinbenton | because i do remember that being a list somewhere | |
| 02:44:34 | kevinbenton | mriedem: i don't remember if it's just for IDs or for everything | |
| 02:44:50 | mriedem | ok, because i was wondering if this could be optimized https://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L2017 | |
| 02:45:09 | mriedem | that's called from a for loop https://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L2097 | |
| 02:45:54 | mriedem | if i could just do something like GET /floatingips?fixed_ip_address=['fixedip1', 'fixedip2', ...]&port_id=port_id | |
| 02:45:58 | mriedem | that would be the bulk query i'd need | |
| 02:46:10 | kevinbenton | mriedem: https://github.com/openstack/neutron/blob/master/neutron/db/_model_query.py#L191-L194 | |
| 02:46:10 | mriedem | but, | |
| 02:46:16 | kevinbenton | mriedem: yeah, it's an OR list | |
| 02:47:18 | mriedem | hmm, but the API ref says "OpenStack Networking does not offer an OR mechanism for filters." | |
| 02:47:45 | kevinbenton | mriedem: it wouldn't be the first time the docs were wrong :) | |
| 02:47:45 | mriedem | i figured the client would unpack the list and turn that into fixed_ip_address=1&fixed_ip_address=2&port_id=port_id | |
| 02:47:47 | mriedem | and AND them all | |
| 02:50:49 | mriedem | ok i can post a wip for that too and see what happens, | |
| 02:50:56 | mriedem | i don't know if there are any tests in tempest which would tickle that | |
| 02:52:50 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Handle network-changed event for a specific port https://review.openstack.org/465787 | |
| 02:52:51 | mriedem | kevinbenton: ^ is the kind of awful i was talking about earlier | |
| 02:53:12 | kevinbenton | mriedem: yeah, repeats turn into the list which becomes the OR in that first query link i sent | |
| 02:53:17 | kevinbenton | mriedem: here is the request parse https://github.com/openstack/neutron/blob/master/neutron/api/api_common.py#L46-L47 | |
| 02:53:40 | mriedem | cool | |
| 02:54:11 | kevinbenton | mriedem: yeah, not exactly a trivial change :) | |
| 02:54:23 | mriedem | it would really be split into 2 changes, the first being a refactor | |
| 02:54:25 | mriedem | but yeah | |
| 02:54:31 | mriedem | i'm not going for pretty at 10pm | |
| 02:55:01 | kevinbenton | mriedem: move to hawaii | |
| 02:55:10 | kevinbenton | more time in the day that way | |
| 02:55:12 | kevinbenton | :) | |
| 02:55:15 | mriedem | heh | |
| 02:55:27 | mriedem | you are a proven thought leader | |
| 03:00:32 | mriedem | apparently we handle GET /floatingips results as a list when we filter on fixed IP and port, | |
| 03:00:37 | mriedem | but wouldn't that really be 1:1? | |
| 03:00:49 | mriedem | can you have more than one floating IP associated to the same fixed ip and port? | |
| 03:00:56 | mriedem | i'd think not | |
| 03:01:26 | kevinbenton | no | |
| 03:01:34 | mriedem | ok good, this is just confusing nova code then | |
| 03:01:53 | kevinbenton | well let me confirm | |
| 03:04:59 | kevinbenton | mriedem: there actually may be a way to get there | |
| 03:05:06 | mriedem | ew | |
| 03:05:14 | kevinbenton | mriedem: two different routers | |
| 03:05:20 | mriedem | ok | |
| 03:05:21 | kevinbenton | mriedem: attached to two different ext nets | |
| 03:05:29 | kevinbenton | mriedem: both attached to same internal | |
| 03:05:37 | kevinbenton | mriedem: and both with a floating IP associated to the instance | |
| 03:06:09 | kevinbenton | mriedem: and it would actually work okay if you setup some routing rules on the instance | |
| 03:06:18 | kevinbenton | mriedem: so i think you'll have to preserve that use case :) | |
| 03:06:24 | mriedem | damn | |
| 03:06:24 | mriedem | ok | |
| 03:07:19 | kevinbenton | mriedem: the constraint for floating IP is external_network + IP + port_id https://github.com/openstack/neutron/blob/master/neutron/db/models/l3.py#L101 | |
| 03:09:42 | mriedem | i'll bake that into a comment in here | |
| 03:16:17 | mriedem | if we only have one fixed ip per port then this doesn't buy us much | |
| 03:16:19 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Bulk query floating IPs in _nw_info_get_ips https://review.openstack.org/465792 | |
| 03:16:24 | mriedem | but here is the thing ^ | |
| 03:17:15 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Bulk query floating IPs in _nw_info_get_ips https://review.openstack.org/465792 | |