15:00:33 <sbonazzo> #startmeeting Virt SIG
15:00:33 <centbot> Meeting started Tue May 15 15:00:33 2018 UTC.  The chair is sbonazzo. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:33 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:00:44 <sbonazzo> #chair szarkos
15:00:44 <centbot> Current chairs: sbonazzo szarkos
15:00:56 <sbonazzo> anthonyper: gwd: rafaelmartins: virt sig?
15:01:00 <sbonazzo> lsm5:
15:01:13 <sbonazzo> alynpost:
15:01:20 <sbonazzo> jirkade:
15:01:42 * lsm5 here
15:01:49 <sbonazzo> #chair lsm5
15:01:49 <centbot> Current chairs: lsm5 sbonazzo szarkos
15:01:52 <anthonyper> sbonazzo: hi
15:01:57 <sbonazzo> #chair anthonyper
15:01:57 <centbot> Current chairs: anthonyper lsm5 sbonazzo szarkos
15:02:28 <sbonazzo> I'm pretty sure I missed a couple of people to ping
15:03:02 <sbonazzo> but let's start
15:03:13 <sbonazzo> anthonyper: do oyu want to start?
15:03:36 <anthonyper> sbonazzo: Ok
15:03:47 <alynpost> Howdy.
15:03:55 <sbonazzo> #chair alynpost
15:03:55 <centbot> Current chairs: alynpost anthonyper lsm5 sbonazzo szarkos
15:04:12 <gwd> sbonazzo: Hey, I'm here
15:04:19 <anthonyper> #info Xen 4.8 and 4.6 packages updates with latests XSAs
15:04:21 <sbonazzo> #chair gwd
15:04:21 <centbot> Current chairs: alynpost anthonyper gwd lsm5 sbonazzo szarkos
15:04:41 <sbonazzo> gwd: giving back the lead to you :-)
15:05:13 <anthonyper> #info seabios and libvirt have been rebuild with newer versions, so there are newer than the one in CentOS 7.5
15:05:53 <anthonyper> But I'm nott sure how many of this packages are in release, there are at least in testing.
15:06:17 <gwd> anthonyper: I think I pushed seabios and libvirt to release yesterday
15:06:22 <gwd> I haven't checked to see if they managed to get signed yet.
15:06:52 <anthonyper> I've got also a Xen 4.10 packages in CBS, just waiting for repo for testing.
15:07:06 <anthonyper> gwd: Did I miss anything?
15:07:23 <gwd> #info kernel-4.9.99 build as candidate
15:08:10 <gwd> We may want to have a chat about dealing with the whole seabios override thing, but maybe in the AOB section
15:08:26 <gwd> That's all I can think of. :-)
15:09:23 <sbonazzo> gwd anthonyper anything else on xen?
15:09:27 <gwd> alynpost: Out of curiosity, has any further progress been made on those kernel fixes srn_prgmr made?
15:09:40 <gwd> DSBD and... mellanox driver, was it?
15:09:49 <gwd> (Hope I haven't mixed you up with someone else)
15:10:11 <alynpost> gwd: I don't think there has been any movement upstream.  We've on v2/v3 of the patch--the difference being the commit message.
15:10:24 <gwd> OK, thanks.
15:10:34 <gwd> sbonazzo. you want to go next?
15:10:51 <sbonazzo> gwd ok
15:11:03 <sbonazzo> #topic oVirt updates
15:11:32 <sbonazzo> #info oVirt 4.2.3 has been released upstream, packages are being built for Virt SIG
15:11:54 <sbonazzo> #info qemu-kvm-ev has been rebased to 2.10 for CentOS 7.5
15:12:30 <sbonazzo> #info after CentOS 7.5 release we experienced several bugs in particular:
15:13:07 <sbonazzo> #info libguestfs required qemu-kvm-ma instead of qemu-kvm on altarch, causing dependency resolution errors on ppc64le and aarch64
15:13:43 <sbonazzo> #info centos-qemu-ev-release and centos-ovirt-42-release are using the new contentdir variable
15:14:03 <sbonazzo> #info this allow to install the packages on ppc64le and aarch64 (as a good news)
15:14:40 <sbonazzo> #info but this breaks on x86_64 due to a bug in centos-release (and here's the bad news) on new CentOS 7.5 installation, see https://lists.centos.org/pipermail/centos-virt/2018-May/005797.html for more info
15:15:07 <sbonazzo> #info hopefully kbsingh / Arrfab / bstinson / hughesjr / Evolution can look into that and fix quickly
15:16:26 <sbonazzo> #info started tagging packages for oVirt 4.3 consumption, development of 4.3 started using them
15:16:40 <sbonazzo> I think that's mostly it for ovirt
15:16:58 <bstinson> sbonazzo: we're looking into the centos-release issue
15:17:05 <sbonazzo> bstinson: thanks
15:17:17 <gwd> OK -- lsm5?
15:17:31 <lsm5> #topic container update
15:17:43 <lsm5> #info mostly rebuilding rawhide packages on CBS
15:17:48 <lsm5> #info podman bumped to 0.5.3
15:17:55 <lsm5> #info buildah bumped to v1.0
15:18:02 <lsm5> i guess that's about it from my side
15:18:06 <lsm5> gwd: back to you
15:18:17 <gwd> szarkos?
15:18:46 <szarkos> #topic azure support
15:19:09 <sbonazzo> szarkos: welcome
15:19:22 <szarkos> per my mail to the list, I'm looking to publish packages with additional Hyper-V/Azure support
15:19:28 <szarkos> sbonazzo: thanks :)
15:19:48 <sbonazzo> szarkos: if I understood correctly you're also willing to provide a patched kernel
15:20:23 <szarkos> we have two sets of packages - an existing kmod package (with userspace components too), and a newer kernel package that integrates those modules into a standard CentOS kernel
15:20:24 <gwd> szarkos: So the idea would be, someone installs CentOS + your project's repos in a VM running on Azure
15:20:58 <szarkos> gwd: yes, or it could be integrated into the Azure images built by CentOS
15:21:10 <gwd> Right
15:21:38 <szarkos> the kmod packages have some downsides - upgrade issues, kabi breakages that we sometimes cannot avoid
15:22:14 <szarkos> a standard CentOS kernel with updated drivers will be more seamless, but maintain everything else standard in the CentOS 6/7 kernel
15:22:27 <gwd> szarkos: Yeah, having a separate kernel based on one of the stable releases (4.9) works really well for Xen
15:22:54 <gwd> Or that works too, if you don't have driver requirements
15:23:09 <gwd> And you can port your patches to the base CentOS kernel
15:23:27 <szarkos> gwd: we could do either, but we have backports already for the standard CentOS 2.6.32/3.10 kernels
15:23:36 <sbonazzo> szarkos: any chance the patches to the kernel can be merged in the upstream kernel?
15:23:49 <szarkos> maybe less surprising for customers who want to use standard CentOS
15:24:05 <szarkos> sbonazzo: everything we would put in this kernel will already be upstream
15:24:29 <TrevorH> I suspect upstream in this case means RHEL
15:24:35 <szarkos> sbonazzo: or I suppose there may be cases where those patches are in flight
15:24:42 <sbonazzo> szarkos: ok
15:25:05 <sbonazzo> TrevorH: well upstream in this case may also mean RHEL, but I was thinking about kernel.org :-)
15:26:16 <szarkos> TrevorH: ah, i see - yes, upstream RHEL will take these patches, but at some slower cadence.  There are cases in RHEL6 where some features/patches won't make it because Red Hat wants to freeze the lifecycle of that release for only critical fixes
15:26:54 <szarkos> for example - PCI passthrough in CentOS 6
15:27:11 <sbonazzo> szarkos: got it. make sense to me.
15:27:19 <szarkos> this hasn't made it sadly, but many could use it
15:28:18 <sbonazzo> gwd: scope looks well defined to me, ok on my side for adding szarkos to virt sig
15:28:31 <gwd> Yes, I agree.
15:28:38 * gwd digs out the URL
15:29:17 <gwd> szarkos: do you have an account on accounts.centos.org?
15:29:43 <gwd> Most of what you'll need is on this page: https://wiki.centos.org/SIGGuide
15:29:52 <szarkos> gwd: no, shall I create one?
15:29:54 <gwd> But it's not always obvious what parts apply. :-)
15:30:06 <gwd> Yes
15:30:38 <gwd> So you'll need an account there, and then a CBS account
15:30:57 <gwd> Then we need the appropriate tags in CBS created (that's one section of the document)
15:31:25 <gwd> And after that's set up, we'll need the appropriate directories set up on buildlogs and mirrors for distributing content (and that's a different section of the document)
15:33:05 <szarkos> gwd: alright, I created an account and will work through the document
15:33:09 <gwd> CBS tags normally have this template: <signame><centos_version>-<project>-<release_string>-{candidate,release,testing}
15:33:44 <gwd> So virt6-xen-46-candidate is Virt SIG, CentOS 6, Xen, version 4.6
15:33:57 <rbowen> If folks have feedback about how the document can be improved, please let us know, or feel free to go ahead and make those improvements. I know that doc can be very intimidating.
15:34:36 <gwd> "Candidate" is what it starts at after just being built; it's hosted on the CBS server so not suitable to point users at
15:34:51 <gwd> "testing" is synced out to "buildlogs", which is suitable for users to use, but not signed
15:34:56 <szarkos> gwd: can I create an account on CBS or must that be done for me?
15:35:27 <gwd> "release" ends up getting signed and pushed out to mirrors
15:35:36 <gwd> szarkos: CBS> Honesly I forget. :-/
15:36:25 <gwd> szarkos: I *think* once you have an account on accounts.centos.org you should be able to follow the steps under "Community Buiild System" of that document
15:37:33 <gwd> A bunch of these things involve submitting bug reports asking the person responsible for functionality to do something for you, and having someone  responsible for the SIG (like me) give approval
15:38:54 <szarkos> gwd: thanks - looks like kb just needs to approve
15:39:15 <szarkos> thank you for all your help. should I just post to the virt ml if I run into questions?
15:39:47 <gwd> szarkos: Yes, or chase people here
15:41:08 <gwd> #topic AoB
15:41:10 <gwd> Anything else?
15:41:26 <gwd> alynpost: I feel like I should have asked you if you had anything to bring up during the Xen update section.
15:42:25 <gwd> One issue I'd like to bring up is the issue we're having with seabios and libvirt
15:42:54 <gwd> Basically, at the moment CentOS's seabios breaks when running under Xen; and Centos's libvirt doesn't have Xen drivers
15:43:23 <gwd> So Xen4CentOS build our own version of seabios and libvirt, with versions higher than those in the core repo
15:43:58 <gwd> But every time we get an update from RHEL, they bump their version numbers too; and we have a brief window where if people do an update they end up with a broken system
15:44:02 <gwd> until we can build newer versions again.
15:44:35 <gwd> People have suggested we use yum priorities or something like this to work around this issue.  Any other opinions out there?
15:44:46 <alynpost> gwd: I'm glad you asked about the kernel patch.  As I reported ~last week we were looking at upgrades but have focused more recently on non-Xen code so you didn't miss anything not asking after us.
15:47:50 <sbonazzo> gwd: alynpost: if you always need to have an higher version than the base image, why not bumping epoch?
15:48:15 <sbonazzo> gwd: alynpost: so you'll always be at a higher version than the base packages.
15:49:00 <gwd> sbonazzo: I don't know anything about the epoch
15:49:34 <sbonazzo> gwd have a look at https://fedoraproject.org/wiki/Packaging:Versioning
15:50:37 <gwd> Well, that looks like what I need!
15:51:03 <gwd> Thanks. :-)
15:51:48 <gwd> OK, anything else to bring up?
15:53:16 <gwd> OK, thanks everyone!
15:53:18 <gwd> #endmeeting