Tuesday, 2018-04-03

« Monday, 2018-04-02 Index Wednesday, 2018-04-04 »
*** annabelleB has quit IRC00:07
*** annabelleB has joined #kata-dev00:14
*** justJanne has quit IRC00:14
*** justJanne has joined #kata-dev00:14
*** justJanne has quit IRC00:19
*** justJanne has joined #kata-dev00:19
*** justJanne has quit IRC00:30
*** justJanne has joined #kata-dev00:30
*** justJanne has quit IRC00:51
*** justJanne has joined #kata-dev00:52
*** mcastelinoo has joined #kata-dev00:58
*** zerocoolback has joined #kata-dev00:59
*** mcastelino has quit IRC01:01
*** annabelleB has quit IRC01:05
*** mcastelinoo has quit IRC01:08
*** annabelleB has joined #kata-dev01:16
*** annabelleB has quit IRC01:31
*** annabelleB has joined #kata-dev01:32
*** annabelleB has quit IRC01:42
*** liujiong has joined #kata-dev02:49
*** fuentess1 has quit IRC03:21
kata-dev-irc-bot<zhangwei555>  Regarding bare metal server for Metrics, I checked Huawei public cloud, we do have bare metal service, but all server resides in China so I’m not sure if it will be too slow for US and EU to visit. Price is about $9300/year, hardware is 20cores+128G mem+2*10G net card03:51
kata-dev-irc-bot<zhangwei555> I’m trying to get some machine for kata for free, but didn’t get any luck so far :(03:52
kata-dev-irc-bot<eric.ernst> @laijs - are you able to continue discussion on vm-templating this (your evening, our tomorrow morning)?04:05
kata-dev-irc-bot<eric.ernst> Some folks on our side want to be part of the discussion but cannot make it at 7am pacific; are you able to meet on slack at 6apm pacific?04:06
kata-dev-irc-bot<eric.ernst> Wanted to see if you could be available an hour earlier.  Let me know.  @bergwolf @xu @zhangwei555, heads up.04:07
kata-dev-irc-bot<eric.ernst> @zhangwei555 thanks for the details.04:10
*** sjas_ has joined #kata-dev04:22
*** eernst has joined #kata-dev04:22
*** sjas has quit IRC04:25
*** eernst has quit IRC04:27
kata-dev-irc-bot<xu> How about you guys @laijs @bergwolf04:46
kata-dev-irc-bot<bergwolf> fine with me04:48
kata-dev-irc-bot<xu> Is it too early for you guys @eric.ernst? What's the time for us @bergwolf04:52
kata-dev-irc-bot<bergwolf> it's 9pm for us in China04:53
kata-dev-irc-bot<bergwolf> usually it's 10pm04:53
*** liujiong has quit IRC05:11
*** jodh has joined #kata-dev06:34
*** jodh has quit IRC06:34
*** jodh has joined #kata-dev06:34
*** sjas_ is now known as sjas06:52
*** diga has joined #kata-dev07:08
*** diga has quit IRC07:19
*** sameo has joined #kata-dev07:24
*** gwhaley has joined #kata-dev08:12
*** FL1SK has quit IRC08:22
*** cdent has joined #kata-dev10:19
*** FL1SK has joined #kata-dev10:32
*** gwhaley has quit IRC11:01
*** gwhaley has joined #kata-dev12:12
*** devimc has joined #kata-dev12:25
kata-dev-irc-bot<niteshkonkar007> @bergwolf: I followed the instructions mentioned in this pull request for the initrd approach: https://github.com/kata-containers/documentation/pull/45  However at the end of it , I got stuck at install: cannot stat 'kata-containers-initrd.img': No such file or directory  Here are the steps I followed https://paste.fedoraproject.org/paste/yJXZ7kAxKZkH7hq0DgDjmA12:26
kata-dev-irc-bot<niteshkonkar007> Did I miss anything or the PR needs to be amended for more info in the doc?12:27
*** fuentess1 has joined #kata-dev12:35
gwhaleyhi niteshkonkar007 - I think the step missing there is probably the change to the .toml kata config file to set which type of image you are using12:35
gwhaleyyou probably just need to comment out the .img file line, and leave the initrd line iirc, in your .toml file12:35
gwhaleythere is some 'sed magic' in one of the PRs to do that which allows the CI systems to work (sorry, I don't have a link to hand though)12:36
* gwhaley tries to translate irc to slack - let's add an @niteshkonkar007 ^^ for good luck ;-)12:36
kata-dev-irc-bot<niteshkonkar007> File uploaded https://katacontainers.slack.com/files/U8CB48V61/FA02H4U68/sed_magic.png / https://slack-files.com/T86U7NQTT-FA02H4U68-248b8a098e - <@U87L8TV34>: This one?12:39
gwhaley@niteshkonkar007 I think you actually need the first line, and not the second. The CI runs with the .img, and you want it the other way around12:41
*** fuentess1 has quit IRC12:41
gwhaleyso, you need to comment out the .img line, and uncomment the initrd line (I believe that is the opposite of the sed?) - @jodh, can you take a peek pls?12:42
gwhaley@niteshkonkar007 - maybe post the few lines of your .toml file around that area so we can see those?12:42
kata-dev-irc-bot<niteshkonkar007> [hypervisor.qemu] path = "/usr/bin/qemu-system-x86_64" kernel = "/usr/share/kata-containers/vmlinuz.container" initrd = "/usr/share/kata-containers/kata-containers-initrd.img" # image = "/usr/share/kata-containers/kata-containers.img" machine_type = "pc"12:42
kata-dev-irc-bot<niteshkonkar007> The above lines are from /usr/share/defaults/kata-containers/configuration.toml file12:43
kata-dev-irc-bot<niteshkonkar007> post sed magic12:44
gwhaley@niteshkonkar007 - (on my irc view of that, which is all one line ;-) - that looks right to me - the initrd is there, and the .img line is commented out.12:44
gwhaleyAh, @niteshkonkar007 - now I see (once I stared at your error harder) - that config is right, but it looks like the inird.img file does not exist in that path then?12:44
gwhaleyI'm guessing then that the osbuilder step to build and then copy/install the inird.img file did not work correctly12:45
* gwhaley has not gone through this process yet - so that is probably about as much help as I can offer you right now12:45
kata-dev-irc-bot<niteshkonkar007> @graham.whaley: Sorry for that view on IRC. Yes the file does not exists on that path.  Thanks for your inputs ...12:48
kata-dev-irc-bot<james.o.hunt> @niteshkonkar007 - if you run "kata-runtime kata-env" it will tell you if any paths are missing.12:49
kata-dev-irc-bot<bergwolf> @niteshkonkar007 do you have the initrd image file at `/usr/share/kata-containers/kata-containers-initrd.img`?13:00
kata-dev-irc-bot<bergwolf> From the error you posted, it looks like the initrd image is not there. Please follow the steps in `Create an initrd image - OPTIONAL` section to create it.13:01
kata-dev-irc-bot<sebastien.boeuf> @eric.ernst @manohar.r.castelino @laijs @bergwolf @samuel.ortiz shall we begin ?13:04
*** mcastelino has joined #kata-dev13:05
kata-dev-irc-bot<manohar.r.castelino> @sebastien.boeuf sure13:06
*** mcastelinoo has joined #kata-dev13:08
*** mcastelino has quit IRC13:08
kata-dev-irc-bot<bergwolf> looks like @laijs is offline. shall we start with other topics first?13:09
*** mcastelinoo has quit IRC13:09
*** mcastelino has joined #kata-dev13:09
kata-dev-irc-bot<sebastien.boeuf> @bergwolf sure, do you have any in mind ?13:11
kata-dev-irc-bot<sebastien.boeuf> @bergwolf BTW the NVDIMM is by default passing the rootfs on the host as read only13:11
kata-dev-irc-bot<bergwolf> thanks. then I'm at peace :slightly_smiling_face:13:12
kata-dev-irc-bot<sebastien.boeuf> I've seen that you've submitted PRs for built-in shim and proxy13:13
kata-dev-irc-bot<sebastien.boeuf> I'll take a look today13:13
kata-dev-irc-bot<bergwolf> thanks. I'll appreciate it13:13
kata-dev-irc-bot<sebastien.boeuf> in the meantime, could you look into https://github.com/kata-containers/runtime/pull/136 and https://github.com/kata-containers/agent/pull/19613:13
kata-dev-irc-bot<sebastien.boeuf> oh BTW, what do you think about the network namespace monitoring process ? C or Go ?13:14
kata-dev-irc-bot<bergwolf> I'll take a look at them13:15
kata-dev-irc-bot<bergwolf> I don't have any language preference13:15
kata-dev-irc-bot<sebastien.boeuf> @xu is there a way you could ping @laijs?13:16
kata-dev-irc-bot<sebastien.boeuf> @bergwolf well I ask because the footprint is gonna be very different13:16
kata-dev-irc-bot<sebastien.boeuf> if it was not about the footprint I would use Go since it would be easier to implement13:16
kata-dev-irc-bot<sebastien.boeuf> but I am trying to keep things low...13:17
kata-dev-irc-bot<bergwolf> yeah, I agree. It'll be one per netns process13:17
kata-dev-irc-bot<bergwolf> but docker use case is not our first priority right? So I'm fine with either way you guys decide13:18
kata-dev-irc-bot<sebastien.boeuf> @bergwolf well initially we thought it was only for Docker, but ZTE needs this for k8s too. This means we could have several containers, and that's the reason why we cannot implement this into the shim.13:21
kata-dev-irc-bot<sebastien.boeuf> but we can make this configurable I think,13:21
kata-dev-irc-bot<manohar.r.castelino> @sebastien.boeuf what is the k8s use case13:22
kata-dev-irc-bot<sebastien.boeuf> this way we can decide to use this monitoring process or not13:22
kata-dev-irc-bot<bergwolf> How is it interacting with k8s?13:23
kata-dev-irc-bot<sebastien.boeuf> @manohar.r.castelino not sure13:23
kata-dev-irc-bot<sebastien.boeuf> when I say k8s use case, I mean they have several containers per pod and they still need to plug more network13:24
kata-dev-irc-bot<sebastien.boeuf> I think this is the docker implementation of pods13:24
*** mcastelinoo has joined #kata-dev13:26
*** mcastelino has quit IRC13:26
kata-dev-irc-bot<bergwolf> Do you mean they want to add new containers to a running pod (thus hotplug more network)?13:27
kata-dev-irc-bot<sebastien.boeuf> @bergwolf well no, otherwise this would be part of the normal lifecycle, meaning you add a network everytime you add a container13:28
kata-dev-irc-bot<eric.ernst> I think it is similar to how others do it, but they prefer not to have a custom CNI13:29
*** annabelleB has joined #kata-dev13:29
kata-dev-irc-bot<eric.ernst> (Like Huawei)13:29
kata-dev-irc-bot<eric.ernst> Adding multiple interfaces per pod, etc, sometimes after the pod has been started.13:30
kata-dev-irc-bot<manohar.r.castelino> @eric.ernst so instead of something like multus. they will create the POD and then run additional CNI plugins once the POD is up?13:32
kata-dev-irc-bot<sebastien.boeuf> yes13:33
kata-dev-irc-bot<eric.ernst> And or not just static13:33
kata-dev-irc-bot<manohar.r.castelino> In that case I wonder what the resolv.conf should look like (as that is also driven from CNI output). the rest of the params are in the network ns13:33
kata-dev-irc-bot<eric.ernst> They’ll want to attach to already running systems also13:33
kata-dev-irc-bot<eric.ernst> Perhaps bind mount that?  @miao.yanqiang would know13:34
kata-dev-irc-bot<bergwolf> @eric.ernst you mean share netns across pods?13:34
kata-dev-irc-bot<sebastien.boeuf> @bergwolf no, they are sharing netns across containers, not pods13:35
kata-dev-irc-bot<eric.ernst>  Not necessarily ; I mean adding additional network interfaces.13:35
kata-dev-irc-bot<eric.ernst> I’m not sure, but perhaps ZTE guys will be on later13:36
*** mylinux_ has joined #kata-dev13:37
kata-dev-irc-bot<bergwolf> @sebastien.boeuf I thought all containers in a pod would always share the same netns. Do ZTE want to split it and make some of them share a different netns inside the pod?13:37
kata-dev-irc-bot<bergwolf> I hope ZTE can write something to illustrate their use case so we get better understanding of the requirement13:38
kata-dev-irc-bot<bergwolf> BTW, can anyone help setup Jenkins jobs? https://github.com/kata-containers/tests/issues/19013:40
*** eernst has joined #kata-dev13:40
kata-dev-irc-bot<eric.ernst> @bergwolf no no, the pod semantics remain the same.13:40
kata-dev-irc-bot<eric.ernst> shouldn't be any confusion there.13:40
gwhaley@fuentess may have thoughts on jenkins @bergwolf ^^13:41
kata-dev-irc-bot<eric.ernst> As I said, my understaning is this is about adding additional network interfaces to a pods existing netns.13:41
kata-dev-irc-bot<bergwolf> ok, so they want to modify a running pod, correct?13:48
kata-dev-irc-bot<sebastien.boeuf> yes13:48
kata-dev-irc-bot<eric.ernst> Yeah, that's my undersatdning.  They didn't want to have to modify their existing CNI plugins, though, so the more native solution (calling directly via custom CNI plugin) was not attractive.13:48
kata-dev-irc-bot<bergwolf> hmm, so it does make sense by allowing kata runtime to integrate more easily with existing CNI plugins. I guess it is not a general requirement to modify a running pod, right?13:56
*** clarkb has joined #kata-dev13:57
kata-dev-irc-bot<zhangwei555> @sebastien.boeuf About program language for net watcher, how about making a golang implementation first ? It will be much more easier with netlink repo’s help13:58
kata-dev-irc-bot<zhangwei555> I mean the golang netlink repo13:58
kata-dev-irc-bot<sebastien.boeuf> @zhangwei555 well this is not a big deal to implement this in C I think, and I don't want to say (let's implement this first in Go and we'll move to C later, I'd rather go with the right implementation first)14:00
kata-dev-irc-bot<zhangwei555> For ZTE’s requirements, I think @eric.ernst ‘s guess could be the reason. We also moved from the “watcher” solution to custom CNI solution. And I love the latter one :)14:02
kata-dev-irc-bot<eric.ernst> Yep.  Exactly.14:02
kata-dev-irc-bot<sebastien.boeuf> @zhangwei555 this is not a guess, that's what ZTE told us14:02
kata-dev-irc-bot<eric.ernst> this could be iterative, but I understand the starting point.14:02
kata-dev-irc-bot<zhangwei555> @sebastien.boeuf That’s good. I’m fine with the language, dont have any preference:)14:04
kata-dev-irc-bot<sebastien.boeuf> ok14:06
kata-dev-irc-bot<eric.ernst> cool - sounds like we can get started there, then, sebastien.14:07
kata-dev-irc-bot<eric.ernst> @zhangwei555, @bergwolf - zte engineers are planning to get started with the virtcontainer changes to support hotplug.14:07
kata-dev-irc-bot<sebastien.boeuf> @eric.ernst yeah I'll work on this today14:08
kata-dev-irc-bot<eric.ernst> **network hotplug!14:08
kata-dev-irc-bot<eric.ernst> @bergwolf -- I assigned the sandbox API updates to you; I thought that was on your list last we talked.  Is that correct, you think?14:09
kata-dev-irc-bot<bergwolf> cool. are they targeting the network hotplug APIs? or just the monitor process?14:09
kata-dev-irc-bot<eric.ernst> *just* the APIs14:09
kata-dev-irc-bot<eric.ernst> Sebastien is doing the monitoring process.14:09
kata-dev-irc-bot<eric.ernst> We'll probably handle updates to the agent (if they become necessary)14:09
kata-dev-irc-bot<bergwolf> yeah, that's next on my list, right after the builtin shim/proxy14:10
kata-dev-irc-bot<eric.ernst> okay.  What about volume plugin?14:10
kata-dev-irc-bot<eric.ernst> Is that on someone's list?14:10
kata-dev-irc-bot<eric.ernst> (see the email I sent regarding 1.0 release)14:10
kata-dev-irc-bot<eric.ernst> I kind of threw a dart at the calendar ;)14:10
kata-dev-irc-bot<bergwolf> I'm not sure14:10
kata-dev-irc-bot<bergwolf> I can take volume **hotplug** if no one volunteers to, but I need to finish the sandbox API first (and that's kinda big per se)14:12
kata-dev-irc-bot<eric.ernst> Yes, I understand and don't expect you to, necssarily.14:13
kata-dev-irc-bot<sebastien.boeuf> @bergwolf yeah please focus on the sandbox API14:13
kata-dev-irc-bot<eric.ernst> I just want to make sure we have owners for each major item now.14:13
kata-dev-irc-bot<eric.ernst> Anyway, I'm curious if you have feedback regarding release-gating issues, etc.14:13
kata-dev-irc-bot<eric.ernst> I don't see an issue necessarily for the builtin shim/proxy14:13
kata-dev-irc-bot<eric.ernst> want to make sure its all up there so we have an accurate view of what's "left"14:14
kata-dev-irc-bot<bergwolf> I agree with you on release-gating. I need to know what blocks the v1.0 release and start to add backlogs14:14
kata-dev-irc-bot<eric.ernst> yep.14:14
kata-dev-irc-bot<anne> would it make sense to have that as its own label until we reach 1.0?14:15
kata-dev-irc-bot<eric.ernst> To clarify, you mean the label "release-gating" ?14:16
kata-dev-irc-bot<eric.ernst> I've started throwing that on issues already (which were obviously release gating)14:17
kata-dev-irc-bot<anne> ah, no, sorry. a label that something is specifically blocking 1.0 (vs nice to haves)14:17
kata-dev-irc-bot<eric.ernst> That was the intention of release-gating.14:18
kata-dev-irc-bot<eric.ernst> Throw that on items which are specifically blocking 1.014:18
kata-dev-irc-bot<eric.ernst> snippet from my (too long) email: release-gating: feature or bug which must work in our 1.0 release.  We should make sure there is agreement from the architecture committee on what is marked release-gating14:18
kata-dev-irc-bot<anne> ah, perfect. i'll go get more coffee and straighten out my thoughts...14:19
*** gabyc_ has joined #kata-dev14:19
kata-dev-irc-bot<eric.ernst> :)  Same here.14:20
kata-dev-irc-bot<eric.ernst> Similarly, I think "help wanted" and "good first issue" are great ways to enable more people to get involved (hello everyone out there!)14:21
kata-dev-irc-bot<anne> are others around to comment on the June 1 date proposal?14:22
kata-dev-irc-bot<sebastien.boeuf> I think it's pretty tight, but doable14:24
kata-dev-irc-bot<sebastien.boeuf> not even 2 months14:24
kata-dev-irc-bot<sebastien.boeuf> but better we target earlier than later, this will force people to work faster :slightly_smiling_face:14:25
kata-dev-irc-bot<anne> gives a community goal :slightly_smiling_face:14:27
kata-dev-irc-bot<sebastien.boeuf> yeah and if we go over the target date, we can start hitting people with wooden sticks14:30
kata-dev-irc-bot<sebastien.boeuf> ...14:30
kata-dev-irc-bot<eric.ernst> Yeah.  I think we can probably make good use of time at OSS.14:30
kata-dev-irc-bot<eric.ernst> for finishing up issues as well.14:30
kata-dev-irc-bot<eric.ernst> we'll see.  Yes, wooden sticks, sure.14:30
kata-dev-irc-bot<eric.ernst> anyway, I'd be curious to hear from @xu on it.14:32
kata-dev-irc-bot<eric.ernst> and others if they have feedback.  @zhangwei555?14:33
kata-dev-irc-bot<xu> Jun 1st is pretty tight, but we do need a release soon.14:33
kata-dev-irc-bot<anne> I'm okay with, "tight but feasible"14:33
kata-dev-irc-bot<eric.ernst> Yeah, I had to pick a first date :slightly_smiling_face:14:33
kata-dev-irc-bot<eric.ernst> as a first suggestion.14:33
kata-dev-irc-bot<eric.ernst> I think its feasible, but pretty tight.14:34
kata-dev-irc-bot<eric.ernst> so long as we're pretty well focused and have clear ownership/expectations for features, etc.14:34
*** mylinux_ has quit IRC14:34
kata-dev-irc-bot<anne> any other arch committee members around? @kata @tallclair @samuel.ortiz? opinions on aiming for a june 1 release?14:35
*** mylinux has joined #kata-dev14:37
*** mcastelinn0 has joined #kata-dev14:45
*** mcastelinoo has quit IRC14:45
kata-dev-irc-bot<zhangwei555> June 1 looks good, let’s try our best to reach it though tight14:47
kata-dev-irc-bot<eric.ernst> I think an important next step is to make sure we're aligned on the minimal functional / testing / documentation criteria for a 1.0 release, and make sure we have this tracked with owners.  I was going to follow up on this (my email yesterday was already too long)14:49
*** cdent has left #kata-dev14:50
kata-dev-irc-bot<anne> sounds good!14:50
*** sameo has quit IRC14:53
*** mcastelinn0 has quit IRC15:19
*** mylinux has quit IRC15:21
*** mylinux has joined #kata-dev15:22
*** mylinux has quit IRC15:22
*** mylinux has joined #kata-dev15:22
*** eernst has quit IRC15:23
kata-dev-irc-bot<kata> june 1 sgtm15:35
*** eernst has joined #kata-dev15:36
*** fiddletwix has joined #kata-dev15:50
*** mcastelino has joined #kata-dev15:56
kata-dev-irc-bot<eric.ernst> @archana.m.shinde @sebastien.boeuf and anyone else in Portland who likes free pizza: https://www.meetup.com/Cloud-Native-PDX/events/249368520/?refund_policy=true&rv=wcs1&_xtd=gatlbWFpbF9jbGlja9oAJDg3OTI4NTlhLTJkYTctNDRlNC05N2MzLWJlYjNhYmQzMDNiNw&_af=event&_af_eid=249368520&response=3&action=rsvp&fromEmail=24936852016:13
kata-dev-irc-bot<archana.m.shinde> @eric.ernst Great! Will definitely attend.16:16
gwhaleyhmmm, pizza. A long way to come for it though...16:19
kata-dev-irc-bot<eric.ernst> usually beer too :slightly_smiling_face:16:25
kata-dev-irc-bot<sebastien.boeuf> @eric.ernst you're not presenting ??16:26
kata-dev-irc-bot<eric.ernst> @sebastien.boeuf just failed a reading test :slightly_smiling_face:16:26
kata-dev-irc-bot<eric.ernst> Here16:27
kata-dev-irc-bot<eric.ernst> Ensuite, Eric Ernst fera une démonstration de KataContainers, un moyen de combiner la virtualisation matérielle avec des conteneurs Kubernetes pour des services plus isolés et sécurisés sans sacrifier les temps de démarrage.16:27
kata-dev-irc-bot<sebastien.boeuf> Oh yeah I needed the French version16:30
kata-dev-irc-bot<sebastien.boeuf> Sorry I had read only the beginning16:30
kata-dev-irc-bot<sebastien.boeuf> Of course I won't be there, don't want to support you16:31
*** gwhaley has quit IRC16:32
*** zerocoolback has quit IRC16:39
kata-dev-irc-bot<eric.ernst> @sebastien.boeuf, or anyone else that recalls -- when using kubectl, do we need to take into account the cri-socket specifically like we do in kubeadm?16:53
kata-dev-irc-bot<eric.ernst> I heard report of pod delete failures that sounded much like what I'd see when we do kubeadm reset without specifying the CRIO cri socket16:54
kata-dev-irc-bot<sebastien.boeuf> @eric.ernst don't know16:56
kata-dev-irc-bot<raravena80> @eric.ernst don't think so, I believe kubectl only talks to the api server17:01
kata-dev-irc-bot<eric.ernst> @tallclair - I was looking the signode Q2 planning - is "Prototype and support monitoring metrics for 3rd party device plugins monitoring" along the lines of standardizing particular metrics, and something we should look to support in Kata?17:01
kata-dev-irc-bot<eric.ernst> @raravena80 - okay.  Perhaps kubeadm init wasn't done with the explicit --cri-socket, in this case then.17:02
kata-dev-irc-bot<eric.ernst> That makes sense....17:02
kata-dev-irc-bot<raravena80> right, makes sense since kubeadm would setup the kubelet (k8s node)17:03
*** jodh has quit IRC17:03
*** gwhaley has joined #kata-dev17:14
kata-dev-irc-bot<tallclair> I'm actually not sure exactly what that means, I'll sync up with David later. For Kata, there are already a set of core metrics in the CRI that should be supported.17:15
kata-dev-irc-bot<tallclair> https://github.com/kubernetes/kubernetes/blob/master/pkg/kubelet/apis/cri/runtime/v1alpha2/api.proto#L83-L8717:16
kata-dev-irc-bot<eric.ernst> Okay, thanks Tim.  I'm curious about the debug pod as well -- I'll just keep an eye on it.17:19
kata-dev-irc-bot<eric.ernst> And, I'll open an issue re: the existing metrics.  I need to better understand who that'll work with respect to CRI-containerd/CRI-O, and how that'll end up getting down to Kata.17:20
kata-dev-irc-bot<eric.ernst> FYI: https://github.com/kata-containers/osbuilder/issues/7717:28
kata-dev-irc-bot<eric.ernst> curious if others (@jonolson, @manohar.r.castelino) have thoughts on this17:28
*** eernst has quit IRC17:39
*** eernst has joined #kata-dev17:40
kata-dev-irc-bot<jonolson> I don't have a lot of thoughts (our image team is mostly responsible for packaging up user-mode and environmental stuff -- we rely on distros for he kernel) -- the limited experience I do have here is packaging a custom kernel network driver against a variety of distro/upstream kernels -- for that we maintain a set of coccinelle patches17:56
kata-dev-irc-bot<jonolson> For vendor drivers I'd generally expect them to have done the porting work to target a variety of kernels, so I think beyond automation for running the builds, you'd really only need patches for anywhere the Kata kernel diverges (which should hopefully be thin in the long term, assuming things make it upstream, but I imagine may be more substantial earlier on)17:57
kata-dev-irc-bot<jonolson> Although, based on my experience with the last Linux box I built from hardware and the current state of AMD video drivers, the above really sounds hopelessly optimistic :slightly_smiling_face:17:58
kata-dev-irc-bot<eric.ernst> Yeah, I think for the most part, we are just using an upstream kernel, and that is the continued intention.  Just need to provide that pipeline for "how to" use DKMS against our image.17:59
kata-dev-irc-bot<eric.ernst> :slightly_smiling_face:17:59
kata-dev-irc-bot<eric.ernst> I think DKMS is "supposed" to solve these kinds of problems.  Most of the time.17:59
kata-dev-irc-bot<eric.ernst> I think we just need to ensure DKMS is available and can be used in the kernel and image build process (ie, for developers to easily get their configuration in)18:00
kata-dev-irc-bot<eric.ernst> Having said that, I hadn't gotten it working locally yet, but on paper it sounds good, right?  :slightly_smiling_face:18:01
kata-dev-irc-bot<raravena80> @eric.ernst fyi, cri-containerd just got renamed to just cri https://github.com/containerd/cri . They basically merged the containerd and cri-containerd bits together for k8s.18:07
kata-dev-irc-bot<eric.ernst> Fun times -- that shouldn't add to any confusion, should it?18:08
kata-dev-irc-bot<eric.ernst> I knew they were pulling the bits in into containerd.  Name might be slightly overloaded already.18:09
kata-dev-irc-bot<raravena80> I know, cri the api interface and cri the runtime :,)18:09
kata-dev-irc-bot<eric.ernst> but its one of "those runtimes" that uses a runtime.18:10
kata-dev-irc-bot<eric.ernst> Don't confuse it with the runtime the runtime uses.18:10
gwhaleyoh, for CRI'ing out loud ;-) (Eric made me do it ;-) )18:10
kata-dev-irc-bot<raravena80> we'll just rename the thing that was called the runtime the 'runanytime'18:11
*** cdent_ has joined #kata-dev18:38
*** mugsie_ has joined #kata-dev18:41
*** ninjanne has joined #kata-dev18:43
*** stefanha` has joined #kata-dev18:45
*** bgmccollum_ has joined #kata-dev18:46
*** justJanne has quit IRC18:46
*** ninjanne is now known as justJanne18:47
*** mordred has quit IRC18:47
*** bgmccollum has quit IRC18:47
*** stefanha has quit IRC18:47
*** mugsie has quit IRC18:47
*** cdent_ is now known as cdent18:47
*** mordred has joined #kata-dev18:55
*** annabelleB has quit IRC18:56
*** justJanne has quit IRC18:58
*** justJanne has joined #kata-dev18:59
*** annabelleB has joined #kata-dev18:59
*** eernst has quit IRC19:27
*** gwhaley has quit IRC19:31
*** annabelleB has quit IRC19:43
*** stefanha` is now known as stefanha19:47
*** eernst has joined #kata-dev20:52
*** annabelleB has joined #kata-dev21:11
*** sameo has joined #kata-dev21:12
*** cdent has quit IRC21:18
*** devimc has quit IRC21:34
*** annabelleB has quit IRC21:49
*** annabelleB has joined #kata-dev21:53
*** gabyc_ has quit IRC21:56
*** mylinux has quit IRC22:32
*** mylinux has joined #kata-dev23:05
*** mylinux has quit IRC23:10
*** annabelleB has quit IRC23:36
*** mylinux has joined #kata-dev23:39
*** mylinux has quit IRC23:44
*** mylinux has joined #kata-dev23:58
« Monday, 2018-04-02 Index Wednesday, 2018-04-04 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!