fedora_atomic_wg
LOGS
17:00:09 <ksinny> #startmeeting fedora_atomic_wg
17:00:09 <zodbot> Meeting started Wed Aug  2 17:00:09 2017 UTC.  The chair is ksinny. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:09 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
17:00:09 <zodbot> The meeting name has been set to 'fedora_atomic_wg'
17:00:26 <ksinny> #topic Roll Call
17:00:47 <dustymabe> .hello dustymabe
17:00:48 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dustymabe@redhat.com>
17:00:48 <jbrooks> .fas jasonbrooks
17:00:49 <ksinny> .hello sinnykumari
17:00:49 <roshi> .hello roshi
17:00:50 <zodbot> jbrooks: jasonbrooks 'Jason Brooks' <JBROOKS@REDHAT.COM>
17:00:52 <maxamillion> .hello maxamillion
17:00:53 <zodbot> ksinny: sinnykumari 'Sinny Kumari' <ksinny@gmail.com>
17:00:56 <zodbot> roshi: roshi 'Mike Ruckman' <roshi@mykolab.com>
17:00:59 <zodbot> maxamillion: maxamillion 'Adam Miller' <maxamillion@gmail.com>
17:01:04 <scollier> .hello scollier
17:01:04 <zodbot> scollier: scollier 'Scott Collier' <emailscottcollier@gmail.com>
17:01:39 * dustymabe waves at scollier
17:01:44 <scollier> hyea
17:02:00 <ksinny> #chair dustymabe ksinny jbrooks roshi maxamillion scollier
17:02:00 <zodbot> Current chairs: dustymabe jbrooks ksinny maxamillion roshi scollier
17:02:54 <ksinny> #topic previous meeting action items
17:03:07 <ksinny> * jberkus strigazi to continue moving kube issues to new kube-sig
17:03:07 <ksinny> tracker
17:03:07 <ksinny> * jberkus to write blog post for fedora magazine/pa.io on changes
17:03:07 <ksinny> * dusty to write a comprehensive blog post on updating f25->f26 and will
17:03:09 <ksinny> include update about f25 updates ref
17:03:10 <ksinny> * dustymabe jberkus to write doc on docs vfad with all resources linked
17:03:13 <ksinny> * jbrooks to create kube-sig ticket about versions discussion
17:03:15 <ksinny> * dustymabe to communicate out issue with ref in atomic cloud images
17:03:17 <ksinny> * dustymabe to create ticket to track RFE for rawhide based containers
17:03:19 <ksinny> to be made available from registry.fedoraproject.org
17:03:22 <jbrooks> ksinny, mine is here https://pagure.io/atomic/kubernetes-sig/issue/6
17:03:39 <dustymabe> jbrooks: want to #info that?
17:03:44 <miabbott> .hello miabbott
17:03:44 <jbrooks> #info https://pagure.io/atomic/kubernetes-sig/issue/6
17:03:44 <zodbot> miabbott: miabbott 'Micah Abbott' <miabbott@redhat.com>
17:04:02 <ksinny> jbrooks: thanks
17:04:15 <ksinny> #chair miabbott
17:04:15 <zodbot> Current chairs: dustymabe jbrooks ksinny maxamillion miabbott roshi scollier
17:04:18 <dustymabe> jbrooks: yeah but a summary + the link in the #info would be useful
17:04:28 <jbrooks> ok
17:04:29 <dustymabe> so #undo and then #info
17:04:46 <jbrooks> Will that undo what I did or undo the latest chairing?
17:04:56 <dustymabe> not sure let's see
17:04:57 <dustymabe> #undo
17:04:57 <zodbot> Removing item from minutes: INFO by jbrooks at 17:03:44 : https://pagure.io/atomic/kubernetes-sig/issue/6
17:04:57 <jbrooks> #undo
17:04:57 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x3bc3bc90>
17:05:04 <dustymabe> :)
17:05:04 <ksinny> cool
17:05:13 <dustymabe> jinx
17:05:22 <jbrooks> #info  kube-sig ticket about versions discussion https://pagure.io/atomic/kubernetes-sig/issue/6
17:05:36 <dustymabe> ok i'll do mine
17:06:07 <dustymabe> #info jberkus wrote a blog post on the irc/list changes for pa.io and is working on getting a blog out on the fedora comm blog http://www.projectatomic.io/blog/2017/07/changes-to-fah-mailing-lists/
17:07:00 <dustymabe> #info dusty wrote a comprehensive blog post on upgrading f25->f26 atomic host - it got merged but has not yet posted to the pa.io/blog: https://github.com/projectatomic/atomic-site/pull/458
17:07:37 <jbrooks> Hmm, some broken fu there
17:07:41 <dustymabe> # dusty communicated about the ref issue in atomic host https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-July/msg00043.html
17:07:50 <dustymabe> #info dusty communicated about the ref issue in atomic host https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-July/msg00043.html
17:08:02 <dustymabe> I did not do this one
17:08:07 <dustymabe> #action dustymabe to create ticket to track RFE for rawhide based containers
17:08:12 <dustymabe> #unfo
17:08:15 <dustymabe> #undo
17:08:15 <zodbot> Removing item from minutes: ACTION by dustymabe at 17:08:07 : dustymabe to create ticket to track RFE for rawhide based containers
17:08:22 <dustymabe> #action dustymabe to create ticket to track RFE for rawhide based containers to be made available from registry.fedoraproject.org
17:08:47 <dustymabe> did jberkus and strigazi_OFF do this one: jberkus strigazi to continue moving kube issues to new kube-sig tracker ?
17:09:29 <jbrooks> I don't think so?
17:09:32 <dustymabe> ksinny: re-action that one
17:09:51 <ksinny> #action jberkus strigazi to continue moving kube issues to new kube-sig tracker
17:09:54 <dustymabe> oh - and
17:10:23 <dustymabe> #info dustymabe jberkus came up with etherpad for us to use during the VFAD http://etherpad.osuosl.org/fedora_atomic_docs_vfad
17:10:29 <dustymabe> ok done :)
17:10:42 <ksinny> thanks :)
17:11:04 <ksinny> All action item from last meeting seems to be covered
17:11:30 <ksinny> we don't have any issue tagged for meting discussion
17:11:50 <dustymabe> ksinny: let's ask people for topics
17:11:51 <ksinny> does anyone have anything to discuss before we move to open floor?
17:11:53 <dustymabe> i have one
17:12:07 <ksinny> go on dustymabe
17:12:27 <dustymabe> #topic experiment with modularity
17:12:30 <maxamillion> I just dropped the usb port to my keyboard into a cup of coffee trying to re-arrange something but that's not exactly meeting pertitant
17:12:43 <sayan> .hello sayanchowdhury
17:12:44 <zodbot> sayan: sayanchowdhury 'Sayan Chowdhury' <sayan.chowdhury2012@gmail.com>
17:12:47 <dustymabe> #link https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-August/msg00002.html
17:13:00 <ksinny> #chair sayan
17:13:00 <zodbot> Current chairs: dustymabe jbrooks ksinny maxamillion miabbott roshi sayan scollier
17:13:30 <dustymabe> so the email sums most of this up. but mainly we want to experiment with modularity (you know, that thing people have been talking about for a few years) with Fedora Atomic Host
17:13:49 <walters> i wrote up a "why we should do this" here: https://gist.github.com/cgwalters/c69bf2091eab7c1af316d0d7dd41f530
17:14:07 <dustymabe> Some of the reasoning behind using a module is actually derived from us wanting to have better CI/CD for atomic host
17:14:39 <dustymabe> and we want to see how hard it is to derive atomic host from a module (as a POC)
17:14:50 <walters> #link  https://gist.github.com/cgwalters/c69bf2091eab7c1af316d0d7dd41f530
17:15:25 <dustymabe> does anyone think that experimenting with modules is bad or can think of a reason why it's not something we want to try out for atomic host?
17:16:29 <walters> a quick note; some of you may wonder whether this means `rpm-ostree install` would learn about modules; that's a distinct later step
17:16:47 <jbrooks> I don't think it's bad
17:16:55 <jbrooks> I'd like to see something concrete
17:17:16 <jbrooks> I've been reading various modularity things and I don't know that I fully get it
17:17:18 <dustymabe> jbrooks: yeah - we actually want to engage the modularity team to see if they will help us create the first module
17:17:43 <dustymabe> we think modularity will help us for the CI stuff, but we'll need to talk to them to find out for sure
17:17:43 <jbrooks> I looked at their google form and they sort of lost me w/ their questions
17:18:02 <jbrooks> I'd like to see someone write a little demo or here's what's cool, specificallyt
17:18:30 <dustymabe> jbrooks: for modularity in general or for this specific topic of how it pertains to atomic host
17:18:38 <walters> jbrooks, this is mostly build side, which is distinct from the modules-on-client side
17:18:41 <jbrooks> Both
17:18:43 <jbrooks> Either
17:18:49 <dustymabe> jbrooks: :)
17:18:59 <jbrooks> I've seen lots of hand waving and I'd like to see something specific
17:18:59 <dustymabe> there is a youtube channel for modularity
17:19:01 <walters> where thinks will blend a bit as i note in my doc is things like "where do i report a bug"
17:19:16 <dustymabe> so maybe some stuff in there
17:19:18 <jbrooks> I've watched a video on their channel before
17:19:33 <dustymabe> walters: IIUC everything rpm-ostree compose and down, is mostly unchanged
17:19:52 <dustymabe> so really we would only be changing where we build from and how the builds get triggered
17:19:55 <dustymabe> right?
17:20:00 <walters> yeah i think so
17:20:03 <dustymabe> cool
17:20:06 <jbrooks> My take is -- whatever, I trust you guys
17:20:25 <dustymabe> jbrooks: :) - we still have to convince ourselves
17:20:26 <jbrooks> But I think modularity in general needs some messaging
17:20:45 <dustymabe> this is more or less, "hey, let's experiment with it"
17:20:56 <dustymabe> I think the biggest one of those requirements from the email that we are interested in is:
17:21:01 <jbrooks> Yeah, I'd love to see someone's experiment
17:21:06 <dustymabe> 3  A way to revert a package in the Atomic Host compose to an earlier version.
17:21:30 <jbrooks> That's coming in rpm-ostree though, right?
17:21:33 <dustymabe> this has burned us for a while and is something we'd like to be able to have more control over
17:21:54 <dustymabe> jbrooks: well i think you are talking about overrides
17:21:59 <dustymabe> we are referring more to server side
17:22:11 <walters> yeah though that's another crossover point
17:22:16 <dustymabe> say we build kernel-1.2.3 and it has a bug
17:22:23 <jbrooks> We want to compose from a specific bucket of rpms
17:22:29 <walters> having the rpm-md repo versioned and sync'd the same way as the host is really useful
17:22:32 <jbrooks> We could keep specific buckets of rpms
17:22:36 <dustymabe> we could simply change our configs to point back to kernel-1.2.2 while we wait for kernel-1.2.4
17:22:59 <dustymabe> right
17:23:08 <jbrooks> I'm guessing that modularity makes that easy?
17:23:18 <jbrooks> I really can't tell from what I've seen of it so far
17:23:27 <dustymabe> jbrooks: indeed. i've been told this
17:23:32 <dustymabe> but i haven't looked into it either
17:23:42 <dustymabe> so essentially "let's try it out" :)
17:23:53 <jbrooks> I think we need to hear from some of the modularity peeps
17:23:53 <dustymabe> EOM
17:24:15 <dustymabe> jbrooks: invite them to our next meeting?
17:24:26 <jbrooks> Yeah, invite them to sell it to us
17:24:29 <jbrooks> ;)
17:24:30 <dustymabe> do you have specific questions? or maybe we can make a list of questions for them
17:24:51 <roshi> a list of questions would be good I think
17:25:10 <dustymabe> jbrooks: can I put you up for that?
17:25:23 <dustymabe> feel free to grab me if you need help coming up with a list
17:25:43 <jbrooks> dustymabe, Well, I need to read up on it more
17:25:51 <dustymabe> jbrooks: yep, same here
17:25:54 * roshi hasn't read much about the modularity bits
17:26:02 <jbrooks> I don't know why I should care at all yet, honestly
17:26:11 * ksinny also needs to get more faimilar with modularity
17:26:15 <jbrooks> And to me, an older version of a package, it's dead, man ;)
17:26:34 <dustymabe> jbrooks: yeah, unless the newer version of the package with the fix isn't available yet
17:26:42 <dustymabe> then you want the older version don't you
17:27:03 <jbrooks> Then I install the newer one, or give karma, or deploy back to the working version
17:27:26 <dustymabe> 'deploy back to the working version' - that's the part we can't do
17:27:37 <dustymabe> we can have our users rollback to previous version
17:27:38 <jbrooks> atomic host deploy foo
17:27:38 <dustymabe> sure
17:27:53 <dustymabe> but what if the fix isn't available for a week and we had a scheduled release
17:28:06 <jbrooks> I haven't personally encountered a situation where that didn't suffice
17:28:08 <dustymabe> and in the mean time all of the other updates can't be properly tested?
17:28:21 <jbrooks> I test with a testing ref
17:28:40 <jbrooks> But maybe that ref is broken, is what you're saying?
17:28:55 <dustymabe> it's happened to us several times in atomic host, we had to delay the release of atomic host because a newer package wasn't available yet
17:28:56 <jbrooks> I just can't really call to mind a lot of instances of that
17:29:03 <dustymabe> and we didn't want to release with broken stuff
17:29:28 <jbrooks> So you'd have released, but w/ a package override?
17:29:38 <dustymabe> absolutely
17:29:44 <jbrooks> Bodhi is a problem for sure
17:30:01 <dustymabe> override problem package to older version, run tests, ship!
17:30:07 <jbrooks> I'd maintain another repo that we control and override w/ that
17:30:20 <dustymabe> that's kinda what modularity is giving us
17:30:33 <dustymabe> it's essentially giving us control of a repo
17:30:44 <dustymabe> at least that's how i understand it
17:30:46 <jbrooks> It seems much simpler to just create that repo
17:31:23 <dustymabe> #action jbrooks dustymabe to come up with list of questions for modularity folks and invite them to our atomic weekly meeting in 1 or 2 weeks
17:31:29 <dustymabe> ^^ sound good?
17:31:44 <jbrooks> OK, I'll look into it more
17:31:46 <roshi> ack
17:32:02 <dustymabe> i'll open two tickets for this
17:32:24 <dustymabe> 1 - a ticket for experimenting with modularity - where we can track progress on an atomic host module POC
17:32:42 <dustymabe> 2 - a ticket for questions for them for us to curate with a meeting tag
17:32:52 <dustymabe> sound good?
17:33:03 <ksinny> perfect!
17:33:40 <ksinny> dustymabe: should we get these two tickets as action item?
17:33:47 <dustymabe> sure I'll action them
17:33:55 <dustymabe> jbrooks: does this sound reasonable?
17:33:58 <maxamillion> dustymabe: +1
17:34:02 <jbrooks> dustymabe, +1
17:34:09 <dustymabe> #action dustymabe to create a ticket for experimenting with modularity - where we can track progress on an atomic host module POC
17:34:21 <dustymabe> #action dustymabe to create a ticket for questions for them for us to curate with a meeting tag
17:35:13 <dustymabe> ok next topic anyone ?
17:35:23 <ksinny> Any other topic before we move to open floor?
17:35:45 <ksinny> dustymabe: ah, you already mentioned :)
17:35:52 <dustymabe> :)
17:36:11 <ksinny> seems not
17:36:20 <_ari_> dustymabe: can you add me to the ticket for module experimentation
17:36:36 * roshi has something for open floor - it can be the last thing though (just an announcement)
17:36:53 <ksinny> #topic Open Floor
17:37:00 <dustymabe> _ari_: sure - want me to also add you to the list of people to ping for this meeting?
17:37:14 <_ari_> dustymabe: yes please thanks
17:37:16 <dustymabe> ksinny: you should have something for open floor :)
17:37:37 <ksinny> dustymabe: If you insist :)
17:38:08 <ksinny> #topic status of building Atomic cloudimages on multi-arches
17:38:13 <ksinny> #link https://pagure.io/atomic-wg/issue/299
17:39:09 <dustymabe> ksinny: want to give us a summary of the progress you've made?
17:39:11 <ksinny> I am working on enabling Atomic CloudImages on ppc64le and aarch64, encountered few issues which is mentioned in ticket.
17:39:34 <dustymabe> nice work!
17:39:54 <dustymabe> ksinny: are there any other architectures we should care about ?
17:39:55 <dustymabe> walters: ^^
17:39:59 <ksinny> dustymabe: thanks!
17:40:16 <walters> maybe 32 bit arm but...eh
17:40:35 <dustymabe> walters: pbrobinson would like that :)
17:40:52 <dustymabe> walters: any obvious road blocks for the 32bit arm case?
17:41:05 <walters> i don't know offhand
17:41:15 <walters> i mean none that aren't general fedora
17:41:32 <dustymabe> k, i just wanted to make sure there wasn't something obvious that was blocking it from working
17:41:47 <maxamillion> I think pbrobinson has had a armv7hl atomic image working as a PoC at some point
17:41:52 <maxamillion> for IoT stuff*
17:42:06 <dustymabe> maxamillion: cool yeah I think he said that in his talk last year
17:42:30 <dustymabe> anyway, thanks ksinny - keep pinging us and we'll try to find solutions for the issues you found
17:42:31 <ksinny> If I can get access to an armv7hl machine, can give it a try
17:42:48 <dustymabe> maxamillion: can you point ksinny to one of our fedora arm machines?
17:43:21 <dustymabe> any other open floor items?
17:43:24 <maxamillion> dustymabe: I don't think that will help
17:43:32 <maxamillion> dustymabe: those are for packaging, you can't install an OS on them
17:43:35 * roshi has one
17:43:56 <roshi> Friday is my last day at RH, so you all will likely be seeing less of me
17:44:07 <dustymabe> roshi: :(
17:44:09 <roshi> I'll still be around, just in a limited capacity
17:44:11 <maxamillion> roshi: best wishes! you will be missed!
17:44:22 <maxamillion> roshi: don't be a stranger :)
17:44:28 <ksinny> I will check with pwhalen, he had help me with getting aarch64
17:44:34 <roshi> I won't :)
17:44:43 <roshi> since I still use fedora for all of my stuff
17:44:50 <maxamillion> +1
17:45:02 <roshi> and I still work from home, so I'll still be on IRC most of the time I imagine :D
17:45:09 <dustymabe> roshi: yeah, it won't be the same if we can't fire adamw together
17:45:09 <maxamillion> rocking
17:45:10 <maxamillion> +1
17:45:11 <dustymabe> .fire _ari_
17:45:11 <zodbot> adamw fires _ari_
17:45:13 <dustymabe> sigh
17:45:16 <dustymabe> .fire adamw
17:45:16 <zodbot> adamw fires adamw
17:45:19 <maxamillion> heh
17:45:24 <roshi> lol
17:45:25 <maxamillion> how meta
17:45:27 <ksinny> all the best with roshi :)
17:45:32 <roshi> thanks :)
17:45:38 <sayan> roshi: all the best! :)
17:45:39 <maxamillion> roshi: all the best indeed
17:45:46 <maxamillion> alright, I need to step afk
17:46:39 <roshi> EOF :)
17:47:50 <ksinny> any additional topic from anyone?
17:47:58 <dustymabe> nope
17:49:01 <Drakonis_> btrfs support dropped from rhel7 yesterday
17:49:22 <dustymabe> Drakonis_: is that a meeting topic
17:49:25 <jbrooks> Interesting
17:49:36 <Drakonis_> hmm no, its a open floor
17:49:59 <ksinny> :)
17:50:23 <dustymabe> Drakonis_: right, just wanted to make sure you knew we were in the meeting
17:50:40 <dustymabe> any questions related to it?
17:51:45 <dustymabe> Drakonis_: ^^ ?
17:52:11 <Drakonis_> ostree has support for btrfs, i suppose xfs support will come once it receives CoW and dedupe
17:52:33 <Drakonis_> its not the focus right now, is it?
17:52:41 <dustymabe> isn't ostree agnostic of the filesystem ?
17:53:09 <dustymabe> we have used ext4 and xfs for prebuilt images for atomic host in the past
17:53:11 <Drakonis_> it has support for btrfs features
17:53:35 <Drakonis_> ostree will use if they're available
17:53:44 <dustymabe> didn't know that
17:53:55 <dustymabe> walters must have baked those in
17:54:07 <dustymabe> or maybe some other contributors added support there
17:54:26 <walters> the only place they currently intersect is when we copy /usr/etc -> /etc, we use CoW if available
17:54:35 <walters> actually i need to update this to use the new standard copy_file_range syscall
17:54:41 <dustymabe> either way, what is officially supported by RHEL doesn't affect Fedora as much
17:54:44 <walters> since that will get us e.g. NFS server copy offload
17:55:02 <walters> which isn't really relevant for ostree itself but userspace should use it in general
17:55:11 <dustymabe> it will mean that RH engineers don't get much time to work on specific btrfs things, but doesn't mean we can't try to enable btrfs when possible
17:55:44 <dustymabe> i created an issue in our tracker for trying to get btrfs support enabled
17:55:46 <dustymabe> https://lists.projectatomic.io/projectatomic-archives/atomic-devel/2017-August/msg00002.html
17:55:49 <dustymabe> sigh - wron't link
17:55:52 <dustymabe> https://pagure.io/atomic-wg/issue/306
17:55:55 <dustymabe> ^^ that one
17:56:08 <dustymabe> so let's have discussions about btrfs issues there?
17:56:22 <jbrooks> I added https://pagure.io/atomic-wg/issue/306#comment-451542
17:56:31 <pbrobinson> maxamillion: yes, sort of, it was a bit glued together, still working towards atomic images for iot across armv7/aarch64/x86_64
17:56:31 <Drakonis_> xfs had a update back in 4.8 that lays the foundations for zfs/btrfs features, but that's probably a bit ahead in the future
17:56:55 <Drakonis_> added reverse mapping
17:57:29 <Drakonis_> http://lkml.iu.edu/hypermail/linux/kernel/1608.0/04662.html the pull request in question
17:57:44 <Drakonis_> alright, i think i'll step away, i have some work to do
17:58:10 <dustymabe> and we're out of time
17:58:15 <ksinny> Any info item here?
17:58:53 <dustymabe> ksinny: i don't think so
17:59:02 <dustymabe> #info btrfs discussions for atomic host in https://pagure.io/atomic-wg/issue/306
17:59:08 <dustymabe> but there you go :)
17:59:14 <ksinny> Okay, last 1 minute before we run out of time
17:59:18 <ksinny> dustymabe: thx
18:00:04 <ksinny> Thanks everyone for joining the meeting :)
18:00:08 <ksinny> #endmeeting