Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2021-05-20
12:49:19 admin0 sean-k-mooney, how does the nova-compute process work ? it boots up, gathers a list of insances it has .. dumps to rabbitmq and then it gets populated to the db ?
12:49:19 admin0 sean-k-mooney, how does the nova-compute process work ? it boots up, gathers a list of insances it has .. dumps to rabbitmq and then it gets populated to the db ?
12:49:23 sean-k-mooney gibi: i need to step away for 10-20 mins but i can talk about htis when i get back or on the review
12:49:23 sean-k-mooney gibi: i need to step away for 10-20 mins but i can talk about htis when i get back or on the review
12:49:56 gibi sean-k-mooney: this is the comment thread https://review.opendev.org/c/openstack/nova-specs/+/785014/7/specs/xena/approved/qos-minimum-guaranteed-packet-rate.rst#98
12:49:56 gibi sean-k-mooney: this is the comment thread https://review.opendev.org/c/openstack/nova-specs/+/785014/7/specs/xena/approved/qos-minimum-guaranteed-packet-rate.rst#98
12:50:00 gibi sean-k-mooney: ack
12:50:00 gibi sean-k-mooney: ack
12:50:02 gibi thanks
12:50:02 gibi thanks
12:50:04 sean-k-mooney admin0: not quite the db is the souce of truth for what instance should be on the host
12:50:04 sean-k-mooney admin0: not quite the db is the souce of truth for what instance should be on the host
12:50:57 admin0 isn't there any nova command that can force update these ?
12:50:57 admin0 isn't there any nova command that can force update these ?
12:51:24 sean-k-mooney gibi lol ok i forgot YAGNI was not an irc name hehe i havent heard that term use since i last spoke to cdent
12:51:25 sean-k-mooney gibi lol ok i forgot YAGNI was not an irc name hehe i havent heard that term use since i last spoke to cdent
12:51:41 gibi sean-k-mooney: ohh, sorry, stephenfin used that now in the review
12:51:41 gibi sean-k-mooney: ohh, sorry, stephenfin used that now in the review
12:52:00 gibi but true it was cdent's terminology :)
12:52:00 gibi but true it was cdent's terminology :)
12:52:33 sean-k-mooney im familar with it just havent heard it in a while but yes stephen did use it
12:52:33 sean-k-mooney im familar with it just havent heard it in a while but yes stephen did use it
12:53:20 sean-k-mooney gibi: ok ill be back soon but did you read may question regardign provder.yaml last night?
12:53:20 sean-k-mooney gibi: ok ill be back soon but did you read may question regardign provder.yaml last night?
12:53:32 gibi sean-k-mooney: responded it this morning
12:53:32 gibi sean-k-mooney: responded it this morning
12:53:36 gibi in IRC
12:53:36 gibi in IRC
12:53:52 sean-k-mooney ah ok i have not check scroll back ill do that when i get back
12:53:53 sean-k-mooney ah ok i have not check scroll back ill do that when i get back
12:53:53 gibi 08:07 < gibi> sean-k-mooney: regarding provider.yaml; I think technically it is possible to
12:53:53 gibi 08:07 < gibi> sean-k-mooney: regarding provider.yaml; I think technically it is possible to
12:53:56 gibi report bw and pps inventories via that. As a downside of it the neutron agents
12:53:56 gibi report bw and pps inventories via that. As a downside of it the neutron agents
12:53:59 gibi already have the bridge_mapping and physical_device_mappings in their config
12:53:59 gibi already have the bridge_mapping and physical_device_mappings in their config
12:54:02 gibi file as we use that to figure out which physnet traits needs to be reported
12:54:02 gibi file as we use that to figure out which physnet traits needs to be reported
12:54:05 gibi 08:08 < gibi> sean-k-mooney: if you do the physnet trait reporting in provider.yaml then you
12:54:05 gibi 08:08 < gibi> sean-k-mooney: if you do the physnet trait reporting in provider.yaml then you
12:54:08 gibi manually need to make sure that it is in sync with the bridge mappings
12:54:08 gibi manually need to make sure that it is in sync with the bridge mappings
12:54:11 gibi 08:08 < gibi> sean-k-mooney: I don't know how ovn does the bridge mapping so there it might
12:54:11 gibi 08:08 < gibi> sean-k-mooney: I don't know how ovn does the bridge mapping so there it might
12:54:14 gibi not be a problem
12:54:15 gibi not be a problem
12:54:17 gibi 08:09 < gibi> the vnic_type trait has a similar consequence
12:54:17 gibi 08:09 < gibi> the vnic_type trait has a similar consequence
12:55:43 sean-k-mooney ack. bridge mapping are stored in the ovn south db casisi table in the external-ids field
12:55:43 sean-k-mooney ack. bridge mapping are stored in the ovn south db casisi table in the external-ids field
12:56:35 sean-k-mooney although in general with ovn you just have one bridge br-int
12:56:35 sean-k-mooney although in general with ovn you just have one bridge br-int
12:56:59 sean-k-mooney it might be something to test out i need to redeploy openstack soon and ill likely try doing it with ovn
12:57:00 sean-k-mooney it might be something to test out i need to redeploy openstack soon and ill likely try doing it with ovn
12:57:05 sean-k-mooney ok brb
12:57:05 sean-k-mooney ok brb
12:57:47 openstackgerrit Merged openstack/osc-placement master: Note env OS_PLACEMENT_API_VERSION support https://review.opendev.org/c/openstack/osc-placement/+/791356
12:57:47 openstackgerrit Merged openstack/osc-placement master: Note env OS_PLACEMENT_API_VERSION support https://review.opendev.org/c/openstack/osc-placement/+/791356
12:59:09 admin0 "admin0: not quite the db is the souce of truth for what instance should be on the host" -- will deleteing the rabbitmq and re-creating it help ?
12:59:09 admin0 "admin0: not quite the db is the souce of truth for what instance should be on the host" -- will deleteing the rabbitmq and re-creating it help ?
13:00:43 openstackgerrit Merged openstack/osc-placement master: Add openstackclient-plugin-jobs https://review.opendev.org/c/openstack/osc-placement/+/780598
13:00:43 openstackgerrit Merged openstack/osc-placement master: Add openstackclient-plugin-jobs https://review.opendev.org/c/openstack/osc-placement/+/780598
13:02:59 stephenfin gibi: done
13:02:59 stephenfin gibi: done
13:05:13 gibi stephenfin: thanks for the clarification, It is clear now
13:05:13 gibi stephenfin: thanks for the clarification, It is clear now
13:29:11 sean-k-mooney gibi: too bad we cant rename the bandwith resouce classes
13:29:11 sean-k-mooney gibi: too bad we cant rename the bandwith resouce classes
13:33:50 sean-k-mooney although the orginal approved specs used ingress and egress
13:33:50 sean-k-mooney although the orginal approved specs used ingress and egress
13:34:27 sean-k-mooney the were retoactivly updated to the short names https://github.com/openstack/neutron-specs/commit/2af53625e7a2e39e53db394f97328e6427cd9fb9
13:34:27 sean-k-mooney the were retoactivly updated to the short names https://github.com/openstack/neutron-specs/commit/2af53625e7a2e39e53db394f97328e6427cd9fb9
13:34:35 sean-k-mooney which is why i was surprised we were using them
13:34:35 sean-k-mooney which is why i was surprised we were using them
13:36:17 sean-k-mooney if it was not for the fact that we cant remove standard resouce class i woudl be very tempeted to treat this a a bug and change them
13:36:18 sean-k-mooney if it was not for the fact that we cant remove standard resouce class i woudl be very tempeted to treat this a a bug and change them
13:36:44 sean-k-mooney but reshaps are way to expensive to waste it on something as trivial as this
13:36:44 sean-k-mooney but reshaps are way to expensive to waste it on something as trivial as this
13:39:25 sean-k-mooney sigh ... https://review.opendev.org/c/openstack/nova/+/570847/10/nova/rc_fields.py#45
13:39:26 sean-k-mooney sigh ... https://review.opendev.org/c/openstack/nova/+/570847/10/nova/rc_fields.py#45
13:39:32 sean-k-mooney so that how that happend
13:39:32 sean-k-mooney so that how that happend
13:40:32 sean-k-mooney gibi: unless you 100% agaisnt the long names i would prefer to use ingress and egress expanded but its not like im going to block on that point
13:40:32 sean-k-mooney gibi: unless you 100% agaisnt the long names i would prefer to use ingress and egress expanded but its not like im going to block on that point
13:50:01 gibi sean-k-mooney: I really don't want to change the bandwidth naming, lets accept that as history. a well discussed history as we see from the original review
13:50:01 gibi sean-k-mooney: I really don't want to change the bandwidth naming, lets accept that as history. a well discussed history as we see from the original review
13:50:08 gibi sean-k-mooney: I'm not against long names
13:50:08 gibi sean-k-mooney: I'm not against long names
13:50:51 gibi it will be weird when the bw and the pps resources next to each outher in a resource_request has different spelling
13:50:51 gibi it will be weird when the bw and the pps resources next to each outher in a resource_request has different spelling
13:51:50 sean-k-mooney i dont agree that it was well disucsed since it did not happen in the spec. but yes i accetp we cant change it for bandwith
13:51:50 sean-k-mooney i dont agree that it was well disucsed since it did not happen in the spec. but yes i accetp we cant change it for bandwith
13:54:47 gibi fine, I can change the name :) maybe we wil re-discuss it in the os-resource-classes review :)
13:54:47 gibi fine, I can change the name :) maybe we wil re-discuss it in the os-resource-classes review :)
13:54:59 gibi I will spell out egress and ingress
13:54:59 gibi I will spell out egress and ingress
13:55:18 gibi I don't want to get hung up on that.
13:55:18 gibi I don't want to get hung up on that.
13:55:22 gibi it is just a name
13:55:22 gibi it is just a name

Earlier   Later