Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2021-05-12
10:07:01 sean-k-mooney why makeing osc using max is a huge breaking change
10:07:29 gibi sean-k-mooney: it is the CLI
10:07:29 gibi sean-k-mooney: it is the CLI
10:07:33 sean-k-mooney i hope that is going to be a major version bump
10:07:33 sean-k-mooney i hope that is going to be a major version bump
10:07:34 gibi use by humand
10:07:34 gibi use by humand
10:07:48 sean-k-mooney gibi: yes but it was intended to be stable for scripting
10:07:48 sean-k-mooney gibi: yes but it was intended to be stable for scripting
10:08:21 gibi if somebody scripts against the CLI instead of against the SDK then I'm a sad panda
10:08:21 gibi if somebody scripts against the CLI instead of against the SDK then I'm a sad panda
10:08:24 sean-k-mooney it predatae the sdk and not eveytone used python so saying use the sdk is not valid either
10:08:24 sean-k-mooney it predatae the sdk and not eveytone used python so saying use the sdk is not valid either
10:08:46 sean-k-mooney gibi: it was one of osc design goal to not do this
10:08:46 sean-k-mooney gibi: it was one of osc design goal to not do this
10:09:22 sean-k-mooney gibi: i would be fine with osc defaulting to the miniums micorverion required for the parmateres you passed
10:09:22 sean-k-mooney gibi: i would be fine with osc defaulting to the miniums micorverion required for the parmateres you passed
10:09:54 sean-k-mooney since that is still stable but follow the project client and going to latest is a regression in my view
10:09:54 sean-k-mooney since that is still stable but follow the project client and going to latest is a regression in my view
10:11:14 gibi OK, I will push a revert of the last patch that switch the default behaviopr
10:11:14 gibi OK, I will push a revert of the last patch that switch the default behaviopr
10:11:18 gibi stephenfin: ^^
10:11:18 gibi stephenfin: ^^
10:11:23 gibi and we can discuss
10:11:23 gibi and we can discuss
10:11:26 gibi in the review
10:11:26 gibi in the review
10:11:35 gibi as we might need tetsuro
10:11:35 gibi as we might need tetsuro
10:13:11 gibi I can imagine a solution where a flag or env variable can be used to enable auto negotiation instead
10:13:11 gibi I can imagine a solution where a flag or env variable can be used to enable auto negotiation instead
10:13:20 gibi I presonally would like to use auto negotiation
10:13:20 gibi I presonally would like to use auto negotiation
10:13:48 gibi as I always forget that openstack resource provider trait list needs 1.6 at least
10:13:48 gibi as I always forget that openstack resource provider trait list needs 1.6 at least
10:14:03 sean-k-mooney gibi: i would be fine with auto negociation but negociating the latest i think is not correct. also yes a flag for enabling it i think would be fine
10:14:03 sean-k-mooney gibi: i would be fine with auto negociation but negociating the latest i think is not correct. also yes a flag for enabling it i think would be fine
10:14:27 gibi I understand your point
10:14:27 gibi I understand your point
10:14:31 stephenfin sean-k-mooney: osc is slowly switching to using the latest as part of the migration to sdk
10:14:31 stephenfin sean-k-mooney: osc is slowly switching to using the latest as part of the migration to sdk
10:14:33 sean-k-mooney if we want to do it by default we coudl but that in my view would need a major version bump
10:14:33 sean-k-mooney if we want to do it by default we coudl but that in my view would need a major version bump
10:14:40 gibi I migth disagree that we need to support people scripting against the CLI
10:14:40 gibi I migth disagree that we need to support people scripting against the CLI
10:15:04 sean-k-mooney it was one of the desigin reason why it had its current behavior
10:15:04 sean-k-mooney it was one of the desigin reason why it had its current behavior
10:15:06 stephenfin scripting users should use the sdk now. They can keep using older versions of osc also
10:15:06 stephenfin scripting users should use the sdk now. They can keep using older versions of osc also
10:15:13 gibi * gibi wonders why the IRC log is not updated in the web :/
10:15:15 sean-k-mooney stephenfin: that assume they are using python
10:15:15 sean-k-mooney stephenfin: that assume they are using python
10:15:38 stephenfin yes
10:15:40 sean-k-mooney gibi: it lags there is a chorn script i think that runs evey few minutes
10:15:40 sean-k-mooney gibi: it lags there is a chorn script i think that runs evey few minutes
10:15:49 sean-k-mooney stephenfin: that is not a valid assumtion
10:15:49 sean-k-mooney stephenfin: that is not a valid assumtion
10:16:00 sean-k-mooney osc was create to be a stabel cli that is scriptable
10:16:00 sean-k-mooney osc was create to be a stabel cli that is scriptable
10:16:13 stephenfin maybe not, but we've got limited maintainers and the ones that are there want auto-negotiation
10:16:13 stephenfin maybe not, but we've got limited maintainers and the ones that are there want auto-negotiation
10:16:37 stephenfin osc was also created when we had many more maintainers
10:16:37 stephenfin osc was also created when we had many more maintainers
10:16:44 sean-k-mooney which is why we can channge that if we want to but it need to be a major version bump to singnale we are breakign backward compatiabiltiy
10:16:44 sean-k-mooney which is why we can channge that if we want to but it need to be a major version bump to singnale we are breakign backward compatiabiltiy
10:17:41 stephenfin yes, the last release was a major version bump iirc
10:17:41 stephenfin yes, the last release was a major version bump iirc
10:18:03 sean-k-mooney the release contiang the default change patch would have to be a major version bump
10:18:03 sean-k-mooney the release contiang the default change patch would have to be a major version bump
10:18:23 sean-k-mooney having the ablity to auto negociate is not a breaking change turning it on by default is
10:18:23 sean-k-mooney having the ablity to auto negociate is not a breaking change turning it on by default is
10:21:33 sean-k-mooney gibi: stephenfin if ye want to not revert and you make sure the next version of osc is a major version then i could proably live with that but i still think min required would be better that said i would like thinks like flavor validation to be on by default
10:21:33 sean-k-mooney gibi: stephenfin if ye want to not revert and you make sure the next version of osc is a major version then i could proably live with that but i still think min required would be better that said i would like thinks like flavor validation to be on by default
10:22:05 stephenfin this only affects osc-placement, not osc btw
10:22:05 stephenfin this only affects osc-placement, not osc btw
10:22:19 gibi * gibi lunches 
10:22:27 sean-k-mooney oh it does?
10:22:27 sean-k-mooney oh it does?
10:22:44 sean-k-mooney i tought it was all of osc
10:22:44 sean-k-mooney i tought it was all of osc
10:23:06 stephenfin no. That's happening but separately. The change above was just for osc-placement
10:23:06 stephenfin no. That's happening but separately. The change above was just for osc-placement
10:23:46 sean-k-mooney i still woudl prefer auto negociating the min reqruied not latest but maybe i have to just live with it being unstable
10:23:46 sean-k-mooney i still woudl prefer auto negociating the min reqruied not latest but maybe i have to just live with it being unstable
10:24:08 sean-k-mooney i rarely have need to script against osc so maybe it not worth avocating for those that do
10:24:08 sean-k-mooney i rarely have need to script against osc so maybe it not worth avocating for those that do
10:24:33 stephenfin tbf, those users could simply request an explicit base version manually
10:24:33 stephenfin tbf, those users could simply request an explicit base version manually
10:24:48 stephenfin '--os-compute-api-version 2.1' or whatever
10:24:48 stephenfin '--os-compute-api-version 2.1' or whatever
10:24:50 sean-k-mooney i just dont want verizon complaining that they cant create flavor when the insitent on setting hw:numa_mem_policy that does not exist
10:24:50 sean-k-mooney i just dont want verizon complaining that they cant create flavor when the insitent on setting hw:numa_mem_policy that does not exist
10:26:23 sean-k-mooney stephenfin: ya i know but many dont know what microverions are or care they just want the command they always used to work
10:26:23 sean-k-mooney stephenfin: ya i know but many dont know what microverions are or care they just want the command they always used to work
10:27:27 sean-k-mooney stephenfin: gibi if the desisin to move all of osc to latest has aleready been made then dont revert the placment patch
10:27:27 sean-k-mooney stephenfin: gibi if the desisin to move all of osc to latest has aleready been made then dont revert the placment patch
10:29:39 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Enable 'vmcoreinfo' feature by default https://review.opendev.org/c/openstack/nova/+/752912
10:29:40 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Remove 'create_fake_libvirt_mock' https://review.opendev.org/c/openstack/nova/+/790945
10:29:40 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Make vmcoreinfo dependent on feature flag https://review.opendev.org/c/openstack/nova/+/779342
10:29:41 openstackgerrit Stephen Finucane proposed openstack/nova master: WIP: libvirt: Make vmcoreinfo checks guest arch specific https://review.opendev.org/c/openstack/nova/+/790946

Earlier   Later