08:04:33 <bauzas> #startmeeting nova_extra
08:04:33 <opendevmeet> Meeting started Thu Oct  7 08:04:33 2021 UTC and is due to finish in 60 minutes.  The chair is bauzas. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:04:33 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:04:33 <opendevmeet> The meeting name has been set to 'nova_extra'
08:04:44 <bauzas> good moning folks, I have forgotten to open it :)
08:04:53 <bauzas> sorry for having it delayed
08:05:05 <bauzas> I don't have agenda for this meeting
08:05:20 <bauzas> it's more something like "if you want to discuss with us, let us know"
08:05:34 <bauzas> do we have Asian contributors here ?
08:06:07 <bauzas> or any other folks wanting to discuss ?
08:07:27 <gibi> I can rant about the rainy weather
08:07:40 <gibi> ;)
08:08:42 <sean-k-mooney[m]> just looking at the patch submitted above by jingsheng zhang https://review.opendev.org/c/openstack/nova/+/812796
08:08:54 <sean-k-mooney[m]> is that something we even support
08:09:01 <sean-k-mooney[m]> ipxe boot of nova vms
08:09:20 <bauzas> gibi: hah
08:09:28 <bauzas> we have a good weather here :)
08:09:29 <sean-k-mooney[m]> i dont know if they are here and could explain there motivation for that
08:09:45 <bauzas> sean-k-mooney: good question
08:09:52 * bauzas looks at the change
08:10:25 <sean-k-mooney[m]> my understandign is the only way to use ipxe is to enable the qemu boot menu
08:10:33 <sean-k-mooney[m]> and manually select it
08:10:49 <bauzas> I'm not really sure it's a nova question
08:10:50 <sean-k-mooney[m]> also this feels like a spec not a bug since it has  object changes
08:10:52 <bauzas> maybe more an image
08:11:37 <bauzas> I mean, if you want to have instances calling PXE, then you need to have an image booting by it
08:11:57 <sean-k-mooney[m]> not really
08:12:05 <sean-k-mooney[m]> that is provided by the qemu seabios image on the host
08:12:08 <bauzas> sean-k-mooney: but agreed, this isn't a bug, more a new feature
08:12:13 <sean-k-mooney[m]> its not provide by the glance image
08:12:17 <gibi> I agree that it feels like a feature
08:12:42 <sean-k-mooney[m]> we dont officaly support ipxe boot which is why i raised it
08:12:51 <gibi> when I see ipxe boot with nova that was with a special image booted
08:12:59 <bauzas> sean-k-mooney: oh, you mean that the host needs to support iPXE ?
08:13:17 <sean-k-mooney[m]> gibi you can use an ipxe image form glance
08:13:34 <sean-k-mooney[m]> but seabios i belive support ipex/network boot too
08:13:46 <bauzas> is this about having guests calling a pxe server or having a way to have a host supporting pxe ?
08:14:18 <gibi> sean-k-mooney[m]: I see so there is multiple ways
08:14:33 <sean-k-mooney[m]> its about the guest pxe booting and what happens if you live migrate that guest between hosts
08:15:03 <sean-k-mooney[m]> if the image was from glance then presumable it would not change between hosts when you live migrate
08:15:20 <sean-k-mooney[m]> which is why im assumje this is related to teh seabios image form the host
08:15:32 <sean-k-mooney[m]> it may not be however
08:16:50 <sean-k-mooney[m]> https://bugs.launchpad.net/nova/+bug/1946298
08:17:00 <sean-k-mooney[m]> this is the relevent bug report
08:17:18 <sean-k-mooney[m]> anyway it looks like they are not present to talk about it in the meeting
08:17:27 <bauzas> yeah
08:17:27 <sean-k-mooney[m]> so we can disucss it later
08:17:44 <bauzas> sean-k-mooney: you should review the change and tell me to at least create a blueprint
08:17:53 <bauzas> to exactly explain their needs
08:18:39 <bauzas> it looks to me we never supported ipxe booting with libvirt
08:19:03 <bauzas> libvirt had the knob, but we never exposed it on nova
08:19:32 <bauzas> so this is fortunate to work, but indeed for move operations, you have to do extra things
08:19:39 <sean-k-mooney[m]> correct we dont offically support it
08:20:10 <sean-k-mooney[m]> looking at the bug report then issue is related to the option rom of the network interface which can have different size on diferent hosts
08:20:23 <sean-k-mooney[m]> but i dd not think we supported enabling that either
08:20:34 <gibi> I agree with generic sentiment above, so this is a bp at least but most probably a spec
08:21:28 <bauzas> that's probably a spec indeed, but let's just ask for the blueprint first
08:21:41 <bauzas> just to understand the needs
08:22:25 <bauzas> sean-k-mooney: I can triage the bug as Invalid/Wishlist and explain them to open a blueprint
08:22:38 <bauzas> agreed ?
08:22:56 <sean-k-mooney[m]> sure
08:23:58 <gibi> agreed
08:24:14 <bauzas> ok, anything else to discuss ?
08:24:53 <bauzas> I'll put a PTG session proposal to discuss the opportunity of the Asian timeslot
08:25:12 <bauzas> I don't exactly want to kill it, but I wonder how to make it productive if we keep it
08:25:35 <bauzas> meetings for the sake of meeting seems a bit irrelevant
08:25:47 <gibi> bauzas: I had the meeting mentioned in the retro part of the etherpad
08:25:52 <bauzas> cool
08:25:57 <bauzas> thanks
08:26:18 <bauzas> it's more or less the question to reach the contributors who don't usually get into IRC
08:26:18 <gibi> L119
08:26:51 <bauzas> at least, if we continue, I'll officially name "office hour"
08:27:01 <bauzas> this sounds more reasonable
08:27:08 <bauzas> as we don't really have an agenda
08:27:40 <bauzas> ok, anything to add before I close ?
08:29:09 <gibi> -
08:32:13 <bauzas> #endmeeting