19:01:59 <kushal> #startmeeting Fedora Cloud SIG 19:01:59 <zodbot> Meeting started Wed Jan 21 19:01:59 2015 UTC. The chair is kushal. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:59 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:02:09 <kushal> #meetingname Fedora Cloud SIG 19:02:09 <zodbot> The meeting name has been set to 'fedora_cloud_sig' 19:02:19 <kushal> #topic Roll Call 19:02:43 <oddshocks> .hellomynameis oddshocks 19:02:44 <zodbot> oddshocks: oddshocks 'David Gay' <dgay@redhat.com> 19:03:00 <roshi> .hello roshi 19:03:01 <zodbot> roshi: roshi 'Mike Ruckman' <mruckman@redhat.com> 19:03:13 <kushal> .hellomynameis kushal 19:03:14 <zodbot> kushal: kushal 'Kushal Das' <kushaldas@gmail.com> 19:04:54 <kushal> roshi, people vanished :) 19:05:34 <roshi> so it seems :) 19:05:40 * gholms appears 19:05:42 <lsm5> .hellomynameis lsm5 19:05:44 <zodbot> lsm5: lsm5 'Lokesh Mandvekar' <lsm5@redhat.com> 19:05:46 * dustymabe got held up 19:05:54 <dustymabe> .hellomynameis dustymabe 19:05:54 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dusty@dustymabe.com> 19:06:11 <kushal> #chair dustymabe oddshocks roshi lsm5 gholms 19:06:11 <zodbot> Current chairs: dustymabe gholms kushal lsm5 oddshocks roshi 19:06:33 <jzb> .hellomynameis jzb 19:06:34 <zodbot> jzb: jzb 'Joe Brockmeier' <jzb@redhat.com> 19:06:45 <kushal> #chair jzb 19:06:45 <zodbot> Current chairs: dustymabe gholms jzb kushal lsm5 oddshocks roshi 19:06:59 <scollier> hello 19:07:02 <kushal> #chair scollier 19:07:02 <zodbot> Current chairs: dustymabe gholms jzb kushal lsm5 oddshocks roshi scollier 19:07:44 <kushal> wondering if anyone else will join!! 19:07:55 <kushal> anyway 19:08:10 <kushal> #topic Action items from last meeting 19:08:17 <kushal> * jzb propose the Vagrant feature for F22 19:08:22 * jsmith is finally here 19:08:31 <kushal> #chair jsmith 19:08:31 <zodbot> Current chairs: dustymabe gholms jsmith jzb kushal lsm5 oddshocks roshi scollier 19:08:39 <dustymabe> fyi https://meetbot.fedoraproject.org/fedora-meeting-1/2015-01-14/fedora_cloud_sig.2015-01-14-19.02.log.html 19:08:51 <dustymabe> I don't know if that ever made it to the list 19:09:05 <kushal> dustymabe, nope, I missed that. 19:09:19 <kushal> Including another task :( 19:09:54 <kushal> So who all submitted feature proposals for Fedora 22? 19:10:09 <jzb> oooh, me! 19:10:26 * kushal is going discuss those links/proposals before going to tickets if everyone is okay :) 19:10:26 <lsm5> kushal: the one assigned to me (rocket) is still pending :( ...I'll get that done over this week 19:10:30 <kushal> * going to 19:10:42 <kushal> lsm5, have you submitted the proposal? 19:10:55 <lsm5> kushal: not yet ..is it too late ? 19:11:05 <kushal> lsm5, yes :( 20th was the last date. 19:11:11 <lsm5> woah 19:11:16 <jzb> kushal: does Rocket actually require a feature request? 19:11:43 <kushal> jzb, lsm5 we can just have it, but adding that as a feature proposal means we can market it better. 19:11:56 * roshi did his too 19:11:57 <kushal> In this case we will make sure that we market that well anyways :D 19:11:59 <lsm5> kushal: so should I still go ahead and file it? 19:12:02 <jzb> kushal: exactly 19:12:45 <kushal> lsm5, I guess, you can just keep working on the real technical work and update that proposal wiki as you find time. Because they will not accept any new proposal. 19:12:57 <lsm5> kushal: alright 19:13:26 <kushal> I have submitted one proposal to use tunir to test Fedora Cloud images automatically. 19:13:53 <kushal> jzb, How many proposals from you this time? 19:14:17 <kushal> I guess langdon did a proposal on vagrant, not sure though if it went in on time. 19:14:51 <roshi> I think I saw the announce 19:14:54 <jzb> kushal: two 19:15:16 <jzb> kushal: Vagrant (along with imcleod and langdon) and bare metal (with imcleod) 19:15:35 <kushal> That means at least 4 proposals from the Cloud SIG, good :) 19:15:41 <jzb> https://fedoraproject.org/wiki/Changes/Vagrant_Box_Atomic 19:15:54 <jzb> https://fedoraproject.org/wiki/Changes/Bare_Metal_Atomic 19:15:55 <kushal> https://fedoraproject.org/wiki/Changes/tunir 19:16:22 <jzb> https://fedoraproject.org/wiki/Changes/Local_Test_Cloud 19:16:26 <jzb> that one's roshi's 19:16:36 <kushal> jzb, Thanks :) 19:16:50 <kushal> Moving on then, we can discuss more on these at the open floor. 19:16:58 <jsmith> Sounds good... 19:17:11 <kushal> https://fedorahosted.org/cloud/report/9 current meeting tickets 19:17:27 <kushal> #topic support HVM instances in EC2 (#68) 19:18:10 <kushal> oddshocks, do you know if the site has the new AMI ID(s) ? 19:19:32 <kushal> Can anyone else please confirm that on the ticket later on? 19:19:58 <dustymabe> confirm that the site has the new AMIs? 19:20:07 <oddshocks> kushal: I don't 19:20:17 <kushal> dustymabe, yes 19:20:17 <oddshocks> I sent out that email with the list, and said to contact me if I could do anything to get them on the site 19:20:20 <kushal> oddshocks, Okay 19:20:29 <oddshocks> AFAIK robyduck used to be the one to do that 19:20:32 <oddshocks> prior to F21 at least 19:20:39 <oddshocks> he's the one that asked me for them to put them on the official site 19:20:40 <jzb> oddshocks: I think we need to open a ticket with the web team? 19:20:42 <dustymabe> the site just has AMIs listed 19:20:48 <dustymabe> doesn't specify PV vs HVM 19:20:50 <oddshocks> If I get access to the site, I'm more than happy to be the one to update them from now on 19:20:56 <oddshocks> yeah those are all PV for base and HVM for atomic I bet 19:21:23 <kushal> #action oddshocks should get access to the website to update AMI ID(s) 19:21:31 <kushal> oddshocks, ^^ is that okay? 19:21:33 <oddshocks> IDK the workflow for updating the site, but I am totally happy to pick it up 19:21:36 <oddshocks> kushal: yes :) 19:21:50 <kushal> oddshocks, Thanks :) 19:21:51 <oddshocks> it'd probably make things a lot smoother since I'm usually the first person to notice new fedimg AMIs 19:21:52 <oddshocks> sure! 19:21:59 <kushal> oddshocks, hehe. 19:22:02 <jzb> oddshocks: I believe that you need to grab the site via git and send a patch 19:22:09 <oddshocks> EZ PZ 19:22:15 <jzb> oddshocks: not sure that they give access to update directly right away 19:22:25 <oddshocks> that's fine, I can send patches no prob 19:22:34 <oddshocks> I'd have someone review changes first anyway. don't wanna be breaking anything 19:22:42 <jzb> oddshocks: https://git.fedorahosted.org/git/fedora-web.git 19:22:47 * oddshocks clones 19:22:48 <jzb> oddshocks: I *believe* that's right 19:23:03 <kushal> Moving to next ticket then. 19:23:12 <kushal> #topic publicize fedora-dockerfiles #84 19:23:21 <oddshocks> jzb: hopefully you're right! Once you clone something, you can't unclone. It's the topic of many sci-fi stories 19:24:15 <kushal> Anyone any tips on #84? 19:24:51 <jzb> I've made no progress since last week 19:25:04 <kushal> Okay, moving on then. 19:25:17 <kushal> #topic Update Timezone for the docker images #91 19:25:23 <kushal> This one is stuck on me. 19:25:38 <kushal> we tested the new patch which worked, I should be able to commit it this week. 19:26:02 <kushal> #action kushal should commit the new patch for docker kickstart and update #91 19:26:41 <kushal> #topic Getting sha256sum published for the cloud images #93 19:26:48 <kushal> This is something new. 19:27:06 <kushal> The proposal actually came from systemd upstream in a mail 19:27:39 <kushal> I have talked with nirik on IRC, he has asked me to open a corresponding rel-eng ticket. 19:27:43 <dustymabe> http://mirror.pnl.gov/fedora/linux/releases/21/Cloud/Images/x86_64/ 19:27:59 <dustymabe> the sums are there they just aren't *easy* to find 19:28:08 <dustymabe> from the download page 19:28:09 <kushal> dustymabe, Yeah, we release the info in a different way. 19:28:13 <nirik> kushal: note also there's an existing ticket about that same thing from the virt folks. There is apparently already a standard they use... 19:28:27 <kushal> nirik, Okay. 19:28:28 <jsmith> Standards are nice :-) 19:28:52 <kushal> nirik, I think we should ask systemd to follow that standard instead :) 19:29:14 * nirik nods. 19:29:36 <kushal> nirik, can you please paste link to that other ticket here? I have misplaced the link as it seems. 19:29:53 <nirik> I need to go find it... in a fesco meeting, will take me a few. 19:30:14 <kushal> nirik, okay, thanks 19:30:29 <kushal> Anyone wants to add something to this ticket? 19:31:10 <kushal> Moving on then. 19:31:52 <kushal> I am skipping the openshift commons ticket as it is stalled. 19:32:11 <kushal> There are two more tickets on Dockerfiles 19:32:18 <jzb> kushal: still in process 19:32:20 <kushal> Fedora Magazine Post: How to Use Dockerfiles #87 19:32:29 <jzb> kushal: I think we can remove the meeting tag from "social media" there 19:32:36 <kushal> Social Media for Dockerfiles #88 19:32:38 <kushal> jzb, Okay 19:33:07 <kushal> jzb, done. 19:33:34 <kushal> #topic Decide whether to update cloud image to not include ruby #09 19:33:37 <kushal> #topic Decide whether to update cloud image to not include ruby #90 19:33:51 <kushal> Ruby is not there anymore. 19:33:58 <kushal> Can we just close this ticket? 19:34:06 <roshi> works for me 19:34:08 <jzb> kushal: +1 19:34:13 <roshi> if it's done, just close it :) 19:34:15 <dustymabe> kushal: have the new images been pulbished? 19:34:20 <dustymabe> published? 19:34:31 <kushal> dustymabe, Nope, need to discus that in open floor. 19:34:41 <dustymabe> kushal: good :) I wanted to discuss that 19:35:39 <kushal> Closed that ticket. 19:35:47 <kushal> #topic Open floor 19:35:52 <kushal> I have two 19:36:07 <kushal> 1. We will have to release updated cloud images. 19:36:16 * langdon back if you need me about vagrant 19:36:20 <kushal> dustymabe, ^^ you wanted to say something? 19:36:33 <kushal> langdon, you can update us with the status :) 19:36:53 <dustymabe> kushal: yep I was just wondering when we were going to start doing that 19:37:01 <kushal> langdon, can you please create a ticket to the fedora-cloud track? It will be easier to keep tab then :) 19:37:02 <langdon> kushal, uh... i believe jzb wrote the change (or most of it).. not sure what is next.. 19:37:07 <dustymabe> I'm working with Digital Ocean and want to make sure they keep up :) 19:37:28 <langdon> jzb, you or me? 19:37:37 <jzb> langdon: I can create the trac ticket 19:37:38 <langdon> kushal, he or i will create the ticket 19:37:45 <langdon> jzb, cool thanks 19:37:51 <jzb> kushal: just FYI - I think the bits we need are ready 19:38:03 <jzb> kushal: it's going to be a matter of imcleod syncing with rel-eng 19:38:13 <jzb> to get everything working in Koji 19:38:14 <langdon> kushal, the imagefactory parts are.. i think it is as you say jzb 19:38:44 <langdon> kushal, "someone" should also decide if there should be anything in the image besides a clone of cloud & atomic 19:38:58 <kushal> jzb, I thought we are missing the repo structure part. 19:39:11 <langdon> my suggestion would be just to clone cloud's / atomic's and wait for the bug report 19:39:14 <kushal> langdon, for 21, we had those two and docker. 19:39:27 <kushal> langdon, +1 19:39:29 <langdon> kushal, oh.. docker + cloud? 19:39:32 <langdon> so do all 3? 19:39:36 <jzb> docker? 19:39:48 <jzb> docker image? in Vagrant? 19:39:49 <kushal> langdon, docker filesystem anyway does not have much 19:39:50 * jzb is puzzled 19:39:57 <kushal> jzb, image. 19:40:14 <kushal> langdon, I would prefer to release updated cloud and atomic images. 19:40:14 * langdon opening download site :) 19:40:23 <langdon> kushal, updated? 19:40:33 <kushal> langdon, with latest packages. 19:40:37 <langdon> i assume they should be at the same cadence as isos or qcows 19:40:42 <jzb> I'm not sure we're all talking about the same thing atm. 19:40:47 <langdon> like why do something special 19:41:13 * kushal is talking about releasing updated cloud and atomic images in Fedora 21 land 19:41:36 <jzb> kushal: OK. I think Langdon and I were talking about Vagrant ;-) 19:41:48 <kushal> jzb, :) 19:42:10 * dustymabe was on kushal's wavelength 19:42:24 <jzb> yes, we should release updated images 19:42:24 <langdon> kushal, so.. i don't see a "docker" image.. just cloud and atomic (which adverts docker) here: https://getfedora.org/en/cloud/download/ 19:42:27 <jzb> well, wait 19:42:31 <langdon> kushal, dustymabe im still confused :) 19:42:38 <jzb> we are releasing updates to Atomic 19:42:47 * langdon recognizes this as a normal state 19:42:49 <kushal> langdon, docker image comes from docker hub. 19:43:11 <dustymabe> jzb: yeah we are.. but I think they are talking about releasing a new "starting" point for atomic 19:43:13 <kushal> jzb, Okay 19:43:16 <langdon> kushal, ohhh.. "docker base image".. not "docker on fedora on an iso" 19:43:20 <dustymabe> on the same cadence as the fedora cloud image 19:43:24 <kushal> langdon, Yeah :) 19:43:41 <dustymabe> "starting point" == new atomic image with updates baked in 19:43:51 <kushal> dustymabe, jzb or in case if you do not want that, we can note down that too. 19:44:21 <jzb> kushal: if we want to release updated images, I don't mind, but I'm not sure it's necessary 19:44:27 <jzb> kushal: we don't release updated ISOs for instance 19:44:43 <langdon> so.. for vagrant, I wanted to release a vagrant box of fedora-cloud and fedora-atomic everytime a new iso/qcow is released.. with the same bits... 19:44:53 <kushal> jzb, yeah, but for cloud we decided that we will be releasing updated images. 19:45:11 <jzb> kushal: and, I could be wrong here, an Atomic update that brings you from F21 -> F21 current would be less bandwidth than stock F21 -> F21 Current 19:45:17 <kushal> langdon, Okay, how to create a box? can we do that from a raw file or qcow2? 19:45:31 <roshi> jzb: it was one of the features for F21 cloud product 19:45:31 <kushal> jzb, Okay. 19:45:39 <langdon> kushal, ks file -> imagefactory 19:45:44 <roshi> releasing updated cloud images throughout the release 19:46:54 <jzb> roshi: yeah, I know 19:47:20 <dustymabe> I think the question comes down to "What are the benefits of releasing updated images?" 19:47:36 <roshi> security fixes for one 19:47:37 <dustymabe> then you ask yourself: "do those benefits apply to both the cloud image and atomic?" 19:47:41 <kushal> dustymabe, I think that was discussed during F21 release :) 19:47:43 <langdon> kushal, jzb rebasing atomic: personally, i think you have to rebase the cloud images (regular and atomic) because no one yum updates/ostree update on a cloud server .. i don't think atomic is any different (in practice) for running in the cloud.. if it is running on bare metal though.. that would be an argument not to rebase... 19:47:50 <kushal> dustymabe, atomic, I am not sure. 19:48:32 <langdon> cause, as a consumer, i know x.y.z-atomic is working on the other copy of my hardware.. and i want to add a new hardware.. 19:48:41 <langdon> and i can upgrade with rollbacks 19:48:44 <jzb> langdon: fair 19:49:54 <kushal> langdon, Okay. 19:49:57 <langdon> so.. my comment is.. sorta.. cloud-wg should only be doing cloud-atomic where it should be the same as cloud-normal... and server-wg should be doing server-atomic where the rebasing follows server ... just to make a mess :) 19:50:18 <jzb> langdon: wat 19:50:54 <jzb> langdon: I think that'd be more cat wrangling than is needed 19:50:55 <langdon> like... fedora-cloud is not meant to be used on bare metal.. just "in the cloud"... the problem with fedora-atomic is it can/should be used either in the cloud or on bare metal 19:51:28 <langdon> and the update process is probably different for cloud-instances vs bare metal instances 19:51:47 <dustymabe> langdon: they both use the same tree upstream right? 19:51:48 <jzb> langdon: the division is really "what's legacy server" and "what's cloud-model" 19:51:50 * langdon pretends vms are not important :) 19:51:50 <dustymabe> or is that incorrect? 19:52:01 <jzb> langdon: not so much "what runs in cloud" and "what runs on bare metal" 19:52:15 * jzb has flashbacks to a similar discussion mid-2014 19:52:35 <langdon> dustymabe, correct.. just "in the cloud" you rebuild/reinstall.. you don't rpmostree update (or whatever it is) 19:52:38 * misc point people to the latest stuff in systemd to show that cloud image may soon become container ready 19:54:06 <dustymabe> ok.. so in my opinion running atomic in the cloud (and yes I do run atomic in the cloud) you do run rpmostree upgrade 19:54:27 <dustymabe> but when I choose to rebuild my "cattle" if I could start at a newer image I would certainly choose to do that 19:54:39 <dustymabe> so there is less to update when I run rpmostree upgrade 19:55:25 <dustymabe> does that make sense? 19:55:51 <jzb> dustymabe: makes sense, but I agree with langdon that the use case should be just standing up a new image 19:55:52 <langdon> dustymabe, it does to me.. but.. if your "rebuild stuff" was fully automated, why would you bother to do updates? 19:56:18 <dustymabe> langdon: you have a good point 19:56:34 <langdon> by "fully automated" i include bringing up new servers, shifting traffic, killing old 19:56:36 <jzb> but on the third hand 19:56:52 <jzb> are we going to respin images every time there's an update or just monthly 19:56:55 <jzb> ? 19:57:07 <dustymabe> jzb: I would do it on the same cadence as the cloud image 19:57:13 <dustymabe> too much overhead otherwise 19:57:14 <roshi> the plan was monthly, I thought 19:57:26 <jzb> monthly + security, correct? 19:57:28 <kushal> roshi, that is what I understood. 19:57:32 <dustymabe> jzb: correct 19:57:38 <jzb> e.g. if a shellshocked comes in the day after a respin. 19:57:42 <dustymabe> hopefully we don't have too many heartbleeds or shellshocks 19:57:44 <langdon> propose a compromise? the main big button is always the rebased.. but there is a "history" link that gives you every one back to last major release.. or something like thta 19:58:04 <jzb> langdon: not sure we need the old images, do we/ 19:58:05 <jzb> er? 19:58:18 <langdon> jzb, i think you do for the "pet" case 19:58:22 <kushal> jzb, we do actually. 19:58:39 <jsmith> langdon: Do we really need the history of *every* previous image? 19:59:05 <langdon> jsmith, maybe not.. maybe it is just current and original? 19:59:14 <langdon> you can always upgrade to a point in between 19:59:35 <kushal> langdon, jsmith we have to get an idea from the fedora-legal too. 19:59:55 <dustymabe> jsmith: langdon when you say image? is there any chance you are referring to the upstream ostree? or are you actually talking about the qcow or raw.xz? 20:00:22 <langdon> dustymabe, i mean the "qcow" (insert image format here) 20:00:34 <langdon> but you also need the ostree somewhere too 20:00:50 <jzb> hmm. I wonder if we're saving the ostree trees 20:01:17 <dustymabe> jzb: not sure but we probably should be 20:01:20 * langdon wonders if those are sustainably farmed ostrees... 20:01:52 <dustymabe> I think if we ever get to the point where ostree is as easy to use and manipulate as git.. that history would be useful 20:02:13 <dustymabe> langdon: nice 20:02:24 * langdon bows 20:02:25 <dustymabe> langdon: they are definitely cage-free 20:02:32 <langdon> dustymabe, lol 20:02:56 <dustymabe> ok so where do we stand? on the updates issue? 20:03:15 <dustymabe> same cadence as cloud image? once a month + updates? 20:03:27 <dustymabe> s/updates/security updates/ 20:04:21 <kushal> For the cloud image: once a month + security fixes. 20:04:22 <jsmith> I think that sounds reasonable 20:04:31 <kushal> What about the atomic then? 20:05:34 <dustymabe> ^^ same cadence as cloud image 20:05:44 <dustymabe> I was referring to atomic 20:05:45 <roshi> I don't know enough about atomic to weigh in on this one 20:05:46 <kushal> Okay. 20:06:10 <kushal> dustymabe, can you please open a ticket for the next update release and note down this there? 20:06:17 * jzb is pulled into another meeting 20:06:18 <dustymabe> kushal: yep 20:06:25 <kushal> dustymabe, Thank you. 20:06:30 <dustymabe> we also need to talk about at what point those image updates stop 20:06:40 <dustymabe> but we can save that 20:06:43 <dustymabe> for another day 20:06:46 <kushal> I guess we can close this meeting now 20:06:55 <kushal> It is 1:36am here. 20:07:00 <dustymabe> kushal +1 20:07:10 <jsmith> +1 from me 20:07:35 <jsmith> I'm happy to help coodinate with the Fedora Security Team to decide when security updates warrant a rebuild 20:07:47 * gholms also gets dragged off 20:07:58 <kushal> jsmith, +1 :) 20:08:03 <roshi> sgtm jsmith 20:08:10 <kushal> Closing the meeting now. 20:08:15 <kushal> #endmeeting