17:02:35 <roshi> #startmeeting cloud_wg 17:02:35 <zodbot> Meeting started Wed Jun 17 17:02:35 2015 UTC. The chair is roshi. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:02:35 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 17:02:44 <roshi> #meetingname Cloud WG 17:02:44 <zodbot> The meeting name has been set to 'cloud_wg' 17:02:52 <roshi> #topic Roll Call 17:02:55 <dustymabe> .hellomynameis dustymabe 17:02:55 <roshi> .hello roshi 17:02:56 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dusty@dustymabe.com> 17:03:00 <zodbot> roshi: roshi 'Mike Ruckman' <mruckman@redhat.com> 17:03:15 <kushal> .hellomynameis kushal 17:03:16 <zodbot> kushal: kushal 'Kushal Das' <kushaldas@gmail.com> 17:04:28 <roshi> anybody else? 17:04:55 <roshi> gholms jsmith jzb imcleod ^^ 17:05:23 <dustymabe> number80 ^^ 17:05:35 <number80> .hello hguemar 17:05:36 <zodbot> number80: hguemar 'Haïkel Guémar' <karlthered@gmail.com> 17:05:41 <kushal> the internet ^^^ 17:05:45 <roshi> heh 17:06:45 * gholms is here 17:06:54 <roshi> sweet 17:07:04 <roshi> #topic Previous Meeting Follow-up 17:07:06 <roshi> nzwulfin Blog: Build systemd unit file to start containerized cockpit as service kushal will update https://fedorahosted.org/cloud/ticket/94 about releasing a new image in the end of june 17:07:23 <roshi> that was supposed to be two lines... 17:07:28 <roshi> nzwulfin Blog: Build systemd unit file to start containerized cockpit as service 17:07:34 <roshi> kushal will update https://fedorahosted.org/cloud/ticket/94 about releasing a new image in the end of june 17:07:46 <dustymabe> I just pinged him in #atomic 17:07:55 <roshi> I saw :) 17:08:05 <dustymabe> I think he did blog though 17:08:12 <roshi> I have it set to highlight every time someone mentions "cloud" 17:08:19 <kushal> roshi, :) 17:08:28 <dustymabe> http://www.projectatomic.io/blog/2015/06/running-cockpit-as-a-service/ 17:08:31 <kushal> dgilmore, has a new tree 17:08:52 <kushal> We can ask him to create a new image if the rest is ready, then we can test it. 17:09:28 <roshi> regarding this is ticket 1444 17:09:29 <roshi> https://fedorahosted.org/fesco/ticket/1444 17:09:52 <imcleod> .hello imcleod 17:09:53 <zodbot> imcleod: imcleod 'Ian McLeod' <imcleod@redhat.com> 17:09:54 <roshi> decision is still being made regarding what all is required of an updated release media 17:09:59 <roshi> o/ imcleod 17:10:29 <imcleod> \o roshi 17:10:39 * maxamillion is here ... sorry for being late 17:10:46 <kushal> maxamillion, no problem :) 17:11:19 <roshi> no worries 17:11:30 <roshi> so, are we ready to test the images when they're ready? 17:11:41 <roshi> we're kinda blazing the trail to ensure testing gets done 17:11:43 <kushal> roshi, Yes, tunir can help in. 17:12:11 <roshi> regarding tunir, have you talked to tim to get things working with taskotron or helping with disposable clients? 17:12:18 <roshi> tim = tflink 17:12:23 <kushal> roshi, and if people have bandwidth, I can add in the docker related tests. 17:13:08 <kushal> roshi, Nope, but I am keeping my tests in such a way so that even taskotron can use them. I remember tim saying wait for the new taskotron, he wanted to design it in his way. 17:13:17 <roshi> it seems to me that tunir and taskotron should mesh well, and like there's a duplication of work happening 17:13:23 * tflink doesn't recall saying that 17:13:57 <kushal> tflink, I mean you were designing it at that time and have asked to see how that goes. 17:14:18 <kushal> My first line was not your exact words, sorry for that. 17:15:14 <roshi> I just figure since we have people working on automation, it makes sense for it to be the same automation :) 17:15:46 <kushal> roshi, Yes, that is why the actual tests are just python unittest cases, anyone or any system can use them. 17:16:01 <roshi> good to know 17:16:08 <dgilmore> kushal: I am working to enable daily trees 17:16:34 <kushal> dgilmore, Oh great news. 17:16:39 <roshi> can you sync with tim and figure out anything you could help with regarding testing within the taskotron framework? 17:16:45 <kushal> #info dgilmore is working to enable daily trees. 17:16:59 <roshi> #chair dustymabe kushal imcleod gholms 17:16:59 <zodbot> Current chairs: dustymabe gholms imcleod kushal roshi 17:17:01 <kushal> roshi, I will, but after fudcon :) 17:17:05 <roshi> ok 17:17:23 <roshi> #action kushal to sync with tflink regarding taskotron after fudcon 17:17:33 <kushal> roshi, as a separate update, I will be too busy till 29th, and may stay dead for 1-2 days that the event. 17:17:40 <tflink> it's been a while since we talked about it but cloud image testing is one of the use cases we're designing for 17:17:46 <kushal> * after that. 17:17:52 <kushal> tflink, Yes. 17:18:14 <roshi> sweet, that'll be good 17:18:18 * tflink needs to look at the test cases for tunir to see if they'd work with the design decisions we ended up making for taskotron 17:18:32 <tflink> assuming kushal isn't a zombie after fudcon, anyways 17:18:47 <tflink> :) 17:18:48 <kushal> tflink, hehe :) 17:18:53 <roshi> at this point we need to make sure any updated image gets tested, and the results documented somewhere so QA can sign off 17:19:03 <kushal> roshi, get us a wiki page. 17:19:05 <roshi> since there's no SOP for updated release media ATM 17:19:13 <roshi> for the testing? 17:19:16 <roshi> will do 17:19:32 <roshi> #action roshi to create a test matrix for testing updated images 17:19:57 <kushal> Yes 17:20:30 <roshi> onto the agenda items! 17:20:35 <roshi> of which, we have a lot 17:20:48 <gholms> Heh 17:21:06 <roshi> #topic Missing Cockpit RPMS (#105) 17:21:09 <roshi> #link https://fedorahosted.org/cloud/ticket/105 17:21:35 <dustymabe> sounds like our previous decision isn't going to be what we go with? 17:21:49 <dustymabe> we are pushing forward with the containerize it approach? 17:22:08 <dustymabe> kushal: this is the testing you were doing right ^^ 17:22:16 * roshi hasn't been tracking this one 17:22:46 <kushal> dustymabe, I thought we all agreed to have it as rpms 17:23:35 <dustymabe> kushal: yeah we did.. but I don't think there is any way to enforce that 17:24:33 <dustymabe> and there is now a blog post about how to use the container 17:24:42 <dustymabe> http://www.projectatomic.io/blog/2015/06/running-cockpit-as-a-service/ 17:24:48 <kushal> dustymabe, what about changing the kickstart? 17:25:24 <dustymabe> do make it be enabled by default? 17:25:26 <dustymabe> like as a service? 17:25:33 <kushal> yes 17:25:36 <maxamillion> I was under the impression that it was agreed to not to rpms ... but it's possible I'm mistaken 17:25:40 <dustymabe> yeah I don't know about that 17:26:10 <dustymabe> maxamillion: yeah we decided in last meeting to do rpms but I think others disagree 17:26:22 <dustymabe> ones that have more control over it 17:26:43 <maxamillion> dustymabe: ohhhh ok 17:27:00 <dustymabe> i'm not sold on either way really. I just want the user to have a good experience 17:27:18 <maxamillion> I think both upstream projectatomic and cockpit is recommending it be run as a container for whatever that's worth 17:27:19 <dustymabe> and since we botched to "let's move to containers" for this release we decided to just go back to the rpm approach 17:27:30 <maxamillion> let me try to find documentation on that 17:27:40 <dustymabe> maxamillion: yeah I think that is the long term 17:27:54 <dustymabe> so here's the current situation 17:27:59 <maxamillion> yeah --> http://cockpit-project.org/running.html click on "Fedora Atomic" 17:28:15 <dustymabe> we either double down and decide rpms are what we want and go try to enforce it 17:28:27 <dustymabe> or we embrace the new approach and try to make the user experience better 17:28:43 <dustymabe> kushal, roshi thoughts? 17:29:04 <kushal> dustymabe, I am in for rpms, container means more to download 17:29:08 <maxamillion> maybe related to the discussion but we're trying to come up with a testing-phase approach for https://fedoraproject.org/wiki/Changes/Layered_Docker_Image_Build_Service ... proposal should be going to the releng and/or infra lists in the near future 17:29:23 <dustymabe> obviously in comment 11 they were ignoring what I wrote in comment 7 17:29:24 <dustymabe> https://fedorahosted.org/cloud/ticket/105#comment:11 17:30:27 <roshi> honestly, I'm out of the loop enough with atomic to not trust anything I'd say regarding it 17:30:40 <dustymabe> I guess I can just comment in the ticket and try to push the discussion along 17:31:02 <roshi> speaking blindly, whatever is best for the users - but I realize that means nil because of how loosely 'user' is defined 17:31:09 <roshi> aiui, anyways 17:31:40 <dustymabe> ok. let's move on 17:31:58 <roshi> #action dustymabe to comment in ticket and keep the conversation moving 17:32:03 <roshi> that fine dustymabe ? 17:32:06 <dustymabe> sure 17:32:22 <roshi> #topic Maintaining Docker images 17:32:25 <roshi> #link https://fedorahosted.org/cloud/ticket/97 17:33:38 <roshi> do we have any hard liaison with the Base WG? 17:33:55 <roshi> I mean, docker isn't strictly cloud, I guess, but it's pretty close 17:34:05 <kushal> roshi, in my count it is cloud :) 17:34:08 <kushal> roshi, so lsm5 will update the repo with the steps on creating an official image on the github repo. 17:34:22 <kushal> too many repo in one sentence. 17:34:31 <roshi> can you update the ticket with that info? 17:34:48 * roshi realizes he doesn't even know who's all on the Base WG 17:34:56 <kushal> roshi, I will. 17:35:07 <kushal> roshi, even I don't know all the names :) 17:35:44 <roshi> #action kushal to update ticket #97 with lsm5 activities 17:36:15 <roshi> #topic Investigate systemd-networkd (#14) 17:36:18 <roshi> #link https://fedorahosted.org/cloud/ticket/14 17:36:20 <kushal> There goes my dream of staying out of action. 17:36:24 <roshi> lol 17:36:31 <roshi> comes with the territory :p 17:36:43 <kushal> roshi, So I did investigated with running instances, worked well. 17:37:02 <kushal> roshi, Now I will have to write the change proposal, and work on it through the ks file. 17:37:19 <kushal> roshi, I will update that ticket as it goes. 17:37:28 <kushal> roshi, now add another action :p 17:37:49 <roshi> lol, will do 17:38:10 <roshi> #action kushal to write networkd change proposal and update the kickstart 17:38:34 <roshi> #topic Care and Feeding of Dockerfiles (#84) 17:38:35 <kushal> update the ticket with status updates :) 17:38:36 <roshi> #link https://fedorahosted.org/cloud/ticket/84 17:38:46 <kushal> roshi, I guess we can skip that. 17:39:15 <roshi> I don't know of any work done to split out the dockerfiles 17:39:24 <roshi> we should probably have some tests for each of them too 17:39:54 <kushal> roshi, Yes, more tests are always welcome ;) 17:40:14 <roshi> of course, you're talking to a QA guy :p 17:40:46 <roshi> kk, we'll skip it for now 17:41:05 <roshi> #topic Flyer Text (#108) 17:41:07 <roshi> #link https://fedorahosted.org/cloud/ticket/108 17:41:59 <roshi> I think a flyer could be good 17:42:52 <roshi> it might be good to get some testimonials from people who run the cloud image for their production instances 17:43:13 <kushal> Yes. 17:43:16 <dustymabe> roshi: we got some testimonials for F21 release 17:43:29 <roshi> can we use those for this? 17:43:37 <kushal> roshi, I guess os. 17:43:44 <roshi> who wants to chase those down and write up a draft? 17:43:56 * kushal volunteers dustymabe :p 17:44:00 * dustymabe ducks 17:44:08 <kushal> hah 17:44:10 <kushal> haha 17:44:20 <dustymabe> i think they are on the fedora cloud download page 17:44:23 <dustymabe> or at least they were 17:44:27 <dustymabe> let me see if I can find 17:44:56 <dustymabe> roshi: yeah here is the one we got from Major 17:44:57 <roshi> looks like there needs to be about 6 paragraphs and links to our docs/etc 17:44:58 <dustymabe> V 17:44:59 <dustymabe> https://getfedora.org/en/cloud/ 17:45:07 <dustymabe> that was short though 17:45:25 <roshi> judging fromthe server flyer 17:46:32 <dustymabe> roshi: yeah I don't want to sign up for anything extra right now 17:46:41 <dustymabe> summit next week is going to consume life for a while 17:47:07 <roshi> I'm pretty swamped as is, so if it falls to me it's going to take a while 17:47:15 <roshi> makes sense dustymabe 17:47:27 <kushal> dustymabe, yes yes, that was not that serious :) 17:48:16 <dustymabe> open floor? 17:48:37 <roshi> we've got 3 more items 17:48:42 <dustymabe> oh ok 17:48:47 <roshi> we'll punt on this and move on I guess 17:48:50 * dustymabe has to leave on the hour 17:49:12 <roshi> #topic Updated images (#94) 17:49:14 <roshi> #link https://fedorahosted.org/cloud/ticket/94 17:49:28 <roshi> this is all still in the discussion phase at a higher level 17:49:34 <roshi> as we talked about a bit earlier 17:49:50 <roshi> the best we can do, I think is get a RC ready, run the tests and submit to QA 17:49:57 <dustymabe> roshi: ok 17:50:03 <roshi> once that happens we can take it to whoever and then see if it's good enough 17:50:09 <roshi> then the process can evolve from there 17:50:12 <dustymabe> how are we on the issue that I listed 17:50:16 <dustymabe> issues 17:50:31 <roshi> can't say off the top of my head 17:50:44 <roshi> v2 qcows is a non-trivial fix, aiui 17:50:59 <kushal> Yes 17:51:00 <roshi> didn't nfs-utils go stable a while back? 17:51:01 <dustymabe> roshi: really? 17:51:08 <dustymabe> roshi: yes nfs did 17:51:12 <roshi> yeah, it's something deep in koji 17:51:14 <roshi> aiui 17:51:33 <dustymabe> I really wish we could just run the qemu-img command and be done with it 17:51:50 <roshi> honestly, it might be faster to write a patch for openstack to do the conversion if it needs it 17:51:59 <roshi> ie, has an older version of libvirt installed 17:52:11 <roshi> that's me just guessing though 17:52:21 <dustymabe> roshi: can the *old* openstack do the conversion 17:52:31 <roshi> dustymabe: want to jump to open floor before you leave? 17:52:34 <roshi> that, I don't know 17:52:45 <dustymabe> roshi: no that's fine 17:52:54 <dustymabe> Id like to talk a bit more about these issues 17:52:55 <kushal> dustymabe, I think nope. 17:52:56 <roshi> ok, just checking 17:53:13 <dustymabe> so fix selinux policy 17:53:15 <roshi> I don't know if there is an easy fix, aside from people converting it themselves 17:53:17 <dustymabe> BZ#1227484 17:53:27 <roshi> .bug 1227484 17:53:31 <zodbot> roshi: Bug 1227484 install of packages during cloud-init boot fails with dnf selinux avc denials - https://bugzilla.redhat.com/1227484 17:53:40 <dustymabe> supposedly fix is out but I don't see any builds of the rpm 17:53:48 <dustymabe> I asked in the bug but no response 17:53:54 <roshi> I haven't either (but I haven't looked) 17:53:59 <dustymabe> does anyone know how we can get a copy and test it 17:54:21 <roshi> have to check if there's a koji build of it, then spin up a customized image to test 17:54:30 <roshi> and I haven't been able to get local builds to work for me 17:54:33 <roshi> always fails out 17:55:07 <dustymabe> https://admin.fedoraproject.org/updates/selinux-policy doesn't list the new rpm 17:55:21 <dustymabe> i guess I'll keep asking 17:55:24 <dustymabe> ok 17:55:24 <kushal> dustymabe, ouch. 17:55:46 <roshi> not much we can do if we're not a maintainer for the package, I don't think 17:55:59 <roshi> download the source, do a custom copr build and test with that? 17:56:01 <roshi> I dunno 17:56:35 <dustymabe> it's fine.. I just want to get this in before we release new images 17:56:41 <roshi> yeah, for sure 17:56:49 <roshi> the silver lining is that other tools work 17:57:04 <roshi> so at least you're not totally blocked by it 17:57:10 <dustymabe> yeah 17:57:13 <roshi> still needs fixed and is annoying 17:57:15 <dustymabe> ok that's it for me 17:57:20 <roshi> kk 17:57:23 <roshi> gl hf! 17:57:26 <roshi> thanks dustymabe 17:57:31 <dustymabe> ok going to run now 17:57:37 <dustymabe> here is one thing before I leave 17:57:39 <dustymabe> http://www.fpaste.org/233213/43456328/ 17:57:55 <dustymabe> I gain some space back if I just from fstrim on rootfs of our image 17:58:01 <dustymabe> might be worth looking at 17:58:05 <kushal> dustymabe, Okay 17:58:22 <kushal> We need to update the PRD. 17:58:23 <roshi> good to know 17:58:42 <roshi> #topic 32 bit image (#106) 17:58:44 <roshi> #link https://fedorahosted.org/cloud/ticket/106 17:59:05 <roshi> I think that sans stats on users/downloads, the only way we can know if it's being used 17:59:12 <roshi> is if we pull it and people scream 17:59:27 <roshi> so, I'm +1 that we remove it, provided we can bring it back if we need to 18:00:10 <roshi> thoughts? 18:00:26 <roshi> (we can continue discussion in #fedora-cloud regarding the PRD) 18:00:54 <kushal> roshi, I did a +1 18:01:15 <roshi> I'll vote in the ticket 18:01:24 <roshi> #topic Open Floor 18:02:14 * gholms notes that EC2 introduced a new 32-bit instance type today 18:02:26 <gholms> (Not that that necessarily means anyone cares) 18:02:58 <roshi> interesting 18:03:08 <roshi> I didn't figure there'd be anything new for 32 bit 18:03:14 * roshi sets the fuse 18:03:15 <roshi> 3... 18:03:23 <kushal> We have to update the PRD 18:03:27 <gholms> Every instance type with 4G of memory or less supports it. 18:03:32 * gholms shrugs 18:03:39 <roshi> 2... 18:03:44 <kushal> gholms, Yes. 18:03:52 <roshi> #info PRD discussion to continue in #fedora-cloud 18:03:56 <roshi> 1... 18:04:02 <roshi> thanks for coming folks! 18:04:10 <roshi> #endmeeting