Thursday, 2024-05-30

*** bauzas_ is now known as bauzas00:10
*** bauzas_ is now known as bauzas00:18
*** bauzas_ is now known as bauzas01:32
*** bauzas_ is now known as bauzas02:23
*** bauzas_ is now known as bauzas02:30
*** bauzas_ is now known as bauzas04:01
*** bauzas_ is now known as bauzas04:21
*** bauzas_ is now known as bauzas05:19
opendevreviewRajesh Tailor proposed openstack/nova master: Handle neutron-client conflict  https://review.opendev.org/c/openstack/nova/+/91804806:04
*** mikal_ is now known as mikal06:52
mikalHey sean-k-mooney, I've been out of OpenStack a while, and I'm confused on what if any my next steps with https://review.opendev.org/c/openstack/nova-specs/+/915190 should be. Is it waiting on me somehow, or is it just in a process and will eventually happen? Thanks!06:54
sean-k-mooney[m]mikal:  a bit of both. its still needs review but also reminding us that i was there and ready for review. we had a spec review day 2? weeks ago i think but we never got to it so the ping is enough to remind us to re review07:27
mikalsean-k-mooney[m]: yeah fair enough. I updated it before the review day, but I didn't realise I needed to ping people. Is there an etherpad it should be on or something like that?07:36
sean-k-mooney[m]mikal:  yep https://etherpad.opendev.org/p/nova-dalmatian-status is our tracking etherpad for things that are ready for review08:01
mikalsean-k-mooney[m]: I see you just did a review. I'll churn through those and then re-ping here I suppose?08:02
sean-k-mooney[m]reviewing the spec quickly i still think it need some work. it could benefit form a ascii diagram showing the end user flow for how they would connect to kerbsdie08:02
sean-k-mooney[m]we pinging is not stricly required but its a good way to ensure it does not get lost in our email inbox or in the walll of other reviews we have in gerrit08:03
sean-k-mooney[m]its on my radar again to look at in the next few days but just an fyi im off tomorrow and monday so i likely wont get back to it until Tuesday08:04
mikalYeah, my quick take is that your comments seem reasonable to me, but clearly some parts of this you understand better than me. So for example I can try and write up how the HTML5 transcoding proxy works, but it might be faster for you to just provide a paragraph you consider accurate.08:04
mikalI'm not stressed about timelines as long as the window isn't going to close in the next couple of days or something.08:05
sean-k-mooney[m]it would be good to reach out to bauzas gibi or others to review08:05
bauzasI'm late for specs reviews but I'll try to look at some this week08:06
mikalYeah, I'm trying to work out what's considered hassling these days vs gently reminding. I'm hesitant to just smash out a bunch of pings if that will annoy people.08:06
sean-k-mooney[m]mikal:  the current deadline is milestone 2 for sepc appoval08:06
sean-k-mooney[m]so i think we have like 4 weeks or so08:06
sean-k-mooney[m]july 4th08:07
bauzashttps://releases.openstack.org/dalmatian/schedule.html08:07
bauzasno08:07
sean-k-mooney[m]mikal:  i have no issues with pings. if im busy i just wont reply08:07
bauzasspec freeze is on July 14th08:08
mikalOk cool. I think as long as it doesn't stall for ages that's workable. I'll address your comments over the next few days and then I'll re-ping you and bauzas here.08:08
sean-k-mooney[m]bauzas:  why the 14 we said it shold be milestone 208:08
bauzasie. 2 weeks after milestone-2, which itself is the spec review day08:08
bauzassean-k-mooney: we discussed this at the PTG and you reviewed the schedule change :)08:08
sean-k-mooney[m]right but that is not what we normally do  so why are we giving 2 addtional weeks08:09
bauzasI can't remember the why08:09
bauzaslook at the etherpad08:09
sean-k-mooney[m]fine i guess we can but if we are still at spec review time in the midle of july your spec is proably not going to get finsihed08:09
mikalAlso, I am open to horse trading. Like if someone would like me to review or assist with something else to earn karma for my stuff, then that's fine. I tried to review some other specs along those lines, but must admit I haven't in the last few weeks.08:10
sean-k-mooney[m]ya i missed a lot of the ptg this time because of time zones08:10
sean-k-mooney[m]i missed the reto and plannign disccustion and basiclly the firsr 90 mins  or so of every day since there was no upstream ping and i was off by 108:12
bauzasmikal: that's a kind offer but you shouldn't need it :)08:13
bauzaswe'll review your spec like others :)08:13
bauzas(I just need a way to work concurrently better :D )08:13
mikalOk. I know back in the day we used to tell people to go review some other stuff if they wanted their stuff reviewed. That is, not just the core team should be reviewing things.08:13
sean-k-mooney[m]mikal:  for what its worth since i didnt see an update on it for 30 days i assumd it had not changes since i last looked08:14
sean-k-mooney[m]but i guess i missed the post ptg update08:14
mikalsean-k-mooney[m]: "didn't see" means in email?08:15
mikalBut yeah, I just updated it and then went into a wait state.08:15
sean-k-mooney[m]well reviews are always wellcome but that really only worked/works in my view as a side efffect of “oh this person is doing review, wasnt i ment to review there stuff too” i.e. seeing your name in gerrit reminds me of the fact i proably should have gotten back to reviewing your spec/patch08:15
sean-k-mooney[m]didnt see in that on the spec review day i either looked before you updatd it or a few days after. so i had a spliting headache that day so i didnt do much review on that day and loopped back later in the week08:17
sean-k-mooney[m]and i only looked at specs that had recent updates08:17
sean-k-mooney[m]i must have missed yours08:17
sean-k-mooney[m]mikal:  do you have any devstack support or anything we can eaisiy setup to play with kerbside? when we get to reviewing the code it would be nice to eaisly deploy both08:19
mikalNo, I haven't done anything to integrate with devstack yet. I'm not opposed, but haven't done it. What I do have is a reasonably well documented process to build a set of Kolla-Ansible containers, as that's why my dev environment is using.08:20
mikals/why/what/08:20
mikalKerbside itself is pretty simple to deploy. The complicated bit is landing the 20ish patches across various OpenStack projects to get it all wired up and pretty.08:21
sean-k-mooney[m]ack, i like kolla-ansible although thats more useful for deploying kerbside then master for dev. that said i know they had an experimental dev mode that used git repos08:22
sean-k-mooney[m]i just never got around to using that08:22
mikalThe complete set of patches is at https://github.com/shakenfist/kerbside-patches for what its worth.08:22
mikalI've been working on publishing patched containers, but its taken longer than I'd like it to because I have a moral issue with running docker registry on my limited hosting resources and the complete set of Kolla-Ansible containers is pretty big.08:23
sean-k-mooney[m]its about 8 GB of disksapace if i recall08:23
sean-k-mooney[m]its been a while since i built them once you account for all the layer sharing08:24
mikalYeah, its 60+gb with unshared layers, about 6 IIRC with shared layers.08:24
sean-k-mooney[m]prebuilt images isnt a requirement08:24
mikalThe readme in the patch repository documents by build process.08:24
mikalRebasing the patches is also a pretty good way to find mypy breakages in openstacksdk it turns out.08:25
sean-k-mooney[m]hehe08:26
sean-k-mooney[m]im partly surpsied you are not just keeping forked branches of each of the project08:26
sean-k-mooney[m]and maintianin patches instead08:26
mikalI am trying very hard to not be a bad person.08:26
mikalI also backport all the patches to a bunch of openstack versions because I am fun at parties.08:27
sean-k-mooney[m]why would having a fork be bad everyone has local forks :)08:27
mikalI spent a lot of time arguing against forks in my youth.08:27
sean-k-mooney[m]what you really need to aovid is getting on the hook with this in production for protracted periods of time without upstreaming08:28
mikalYeah agreed, that would suck.08:28
sean-k-mooney[m]well normally i would either rjust submit the patches to gerrit or host them in a gitbub branch08:28
sean-k-mooney[m]preferbally the former08:28
sean-k-mooney[m]since that makes them aviable for review and eaislly downloadable in a working git tree08:29
mikalI'd be happy to do that. I had kind of assumed people would find that annoying until the spec is approved. That said, a lot of the nova patches are fairly vanilla -- add a sound card, add some USB passthrough devices, enforce network encryption, etc etc.08:29
mikalI also haven't really chased down any of the other openstack projects because there's not a lot of point until nova agrees to land the key bit.08:30
sean-k-mooney[m]so it depend on if you follow the letter of the law or are pratical. our docs say you should not start writign the code until the specc is merged but practically speaking most people at least do a poc. and as long as your are not rebasing them alot and using lots of ci resouce there is noharm in pushing them08:31
mikalYeah, those patches haven't changed in a couple of months so there wouldn't be a lot of CI load.08:32
mikalI'll add pushing them to my todo list.08:32
sean-k-mooney[m]by the way kolla only builds debian images now as far as im aware so you probaly dont need the rpm supprot anymore in your kolla series08:33
sean-k-mooney[m]not that that saves much since its only 4 lines or so but just tought i would mention it08:34
mikalOh interesting. I only build debian images anyways because RHEL dropped SPICE support which is kind of awkward for me right now.08:36
sean-k-mooney[m]ya… redhat strop all development of spice and removed supprot for it in our version of qemu when the only person that work on spice on our virt/qemu team  left08:39
sean-k-mooney[m]they may of left because of the decsion to drop spice im not sure08:39
sean-k-mooney[m]but in rhel 9 it got compiled out08:39
sean-k-mooney[m]i dont know of any plans to every change that going forward08:40
sean-k-mooney[m]if kerbside takes off who knows08:40
sean-k-mooney[m]mikal:  kolla still support installing on other distos they just use the debian image for that now08:41
sean-k-mooney[m]i used to use ubuntu-source as my build target for kolla in the past, i never had issues with that on ubuntu or centos08:42
sean-k-mooney[m]anyway im going to go grab a coffeee and move from my ipad to my desk o/08:43
mikalNo worries, thanks for the chat.08:47
sean-k-mooneygibi: bauzas:  when ye have time can ye re reivew my ci fix for the emulation job https://review.opendev.org/c/openstack/nova/+/91846408:51
bauzasack08:51
* mikal wanders off because 12 hours at a computer is enough for one day...08:52
opendevreviewMarlin Cremers proposed openstack/nova master: feat: set rotation rate on disks based on rotation rate from volume  https://review.opendev.org/c/openstack/nova/+/92081809:32
opendevreviewMarlin Cremers proposed openstack/nova master: libvirt: set rotation rate on disks based on rotation rate from volume  https://review.opendev.org/c/openstack/nova/+/92081809:34
opendevreviewMarlin Cremers proposed openstack/nova master: libvirt: set rotation rate on disks based on rotation rate from volume  https://review.opendev.org/c/openstack/nova/+/92081809:43
marlincHello there, I'm trying to contribute a small change to introduce support for the libvirtd rotation_rate flag for volumes. Now I'm wondering if I have to also create a spec alongside a blueprint for this09:49
marlincAnd I'm also wondering if I should also do this for cinder as it uses a new field in the connection_info from cinder09:50
marlincI have already written a small blueprint for this https://blueprints.launchpad.net/nova/+spec/rotation-rate09:50
*** bauzas_ is now known as bauzas09:55
opendevreviewMarlin Cremers proposed openstack/nova master: libvirt: set rotation rate on disks based on rotation rate from volume  https://review.opendev.org/c/openstack/nova/+/92081809:55
*** bauzas_ is now known as bauzas11:43
*** bauzas_ is now known as bauzas11:56
sean-k-mooneymarlinc feature by default are tracked as bluepirnt, we only require specs if there is an upgrade impact which this may or may not have when it comes to live migraton 12:26
sean-k-mooneywe suppoirt compute nodes of up to n-2 or speciicly the previous slurp release12:27
sean-k-mooneyin thei case that n-112:27
sean-k-mooneyso we need to supprot live migrating form master to the previous relese12:27
sean-k-mooneymarlinc i see your irc client disconnected but ill capture it here in teh logs. what we woudl need to confirm is is there a live migration impact. i think yes. then we need to determin how to adress that.12:28
sean-k-mooneythe case that i am concerned about breaking is live migratine form n to n-1 and then back to n12:29
sean-k-mooneyif that work then the only other thing we should need to conifm is when was rotation_rate added and is it supported in our min libvirt version12:30
sean-k-mooneythose are the two main upgrade concerns since this is a dirver only change12:31
sean-k-mooney"rotation_rate" attribute value since 7.3.012:31
sean-k-mooneyour min supported libvirt is 7.0.012:32
sean-k-mooneyso the patch also need to accomidate for that.12:32
sean-k-mooneynova will not allow you to live migrate to an older libvirt12:32
sean-k-mooneyso you just need that check on the source node (the upgraded node)12:33
sean-k-mooneyill metion thsi in the reveiw12:33
*** bauzas_ is now known as bauzas13:05
*** bauzas_ is now known as bauzas13:33
*** bauzas_ is now known as bauzas14:43
*** bauzas_ is now known as bauzas15:07
opendevreviewArnaud Morin proposed openstack/nova master: libvirt: Log exception if failure checking block job  https://review.opendev.org/c/openstack/nova/+/90772015:11
opendevreviewArnaud Morin proposed openstack/nova master: libvirt: Add retry on device job operation timeout  https://review.opendev.org/c/openstack/nova/+/90773015:30
*** bauzas_ is now known as bauzas16:13
*** bauzas_ is now known as bauzas18:35
*** bauzas_ is now known as bauzas19:15
*** bauzas- is now known as bauzas19:32
*** clarkb is now known as Guest808919:49
*** elodilles is now known as elodilles_ooo20:04
*** Guest8089 is now known as clarkb20:04
*** bauzas_ is now known as bauzas20:49
*** clarkb is now known as Guest809421:15
*** Guest8094 is now known as clarkb21:19
*** bauzas_ is now known as bauzas22:45
*** bauzas_ is now known as bauzas23:02
*** bauzas_ is now known as bauzas23:29

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!