| *** mylinux has quit IRC | 00:00 | |
| *** zerocoolback has quit IRC | 00:02 | |
| *** annabelleB has joined #kata-dev | 00:15 | |
| *** sameo has quit IRC | 00:18 | |
| *** oikiki has quit IRC | 00:26 | |
| *** oikiki has joined #kata-dev | 00:26 | |
| *** mylinux has joined #kata-dev | 00:31 | |
| *** mylinux has quit IRC | 00:36 | |
| *** sameo has joined #kata-dev | 00:38 | |
| *** sameo has quit IRC | 01:12 | |
| *** mylinux has joined #kata-dev | 01:16 | |
| *** eernst has quit IRC | 01:20 | |
| *** mylinux has quit IRC | 01:21 | |
| *** annabelleB has quit IRC | 01:55 | |
| *** mylinux has joined #kata-dev | 01:56 | |
| *** oikiki has quit IRC | 01:58 | |
| *** mylinux has quit IRC | 02:00 | |
| *** annabelleB has joined #kata-dev | 02:00 | |
| *** mcastelinn0 has joined #kata-dev | 02:23 | |
| *** mcastelinoo has quit IRC | 02:26 | |
| *** eocardon has left #kata-dev | 02:27 | |
| *** zerocoolback has joined #kata-dev | 02:32 | |
| *** mcastelinn0 has quit IRC | 02:32 | |
| *** zerocoolback has quit IRC | 02:32 | |
| *** zerocoolback has joined #kata-dev | 02:33 | |
| kata-dev-irc-bot | <anne> Reminder for the LinuxCon/ContainerCon/CloudOpenCon happening in China in June. CFP is still open if anyone wants to submit! I've sent in a community onboarding session, maybe @xu would want to send something?! http://lfasiallc.com/events/lc3-2018/ | 02:34 |
|---|---|---|
| kata-dev-irc-bot | <anne> closes march 11th IIRC | 02:35 |
| kata-dev-irc-bot | <xu> the hyper team has submit some topics, and @zhangwei555 and I together with a China based intel stuff have submitted a panel topic. | 02:36 |
| kata-dev-irc-bot | <xu> @anne I though it was closed on 4th originally | 02:37 |
| kata-dev-irc-bot | <anne> got extended for a few more days :slightly_smiling_face: | 02:37 |
| kata-dev-irc-bot | <xu> cool, a cfp extension is not really common in container world :,) | 02:39 |
| kata-dev-irc-bot | <anne> it's not! it's everyone's lucky day! | 02:40 |
| *** annabelleB has quit IRC | 03:02 | |
| *** annabelleB has joined #kata-dev | 03:07 | |
| *** zerocoolback has quit IRC | 03:26 | |
| *** annabelleB has quit IRC | 03:47 | |
| *** annabelleB has joined #kata-dev | 03:50 | |
| *** eernst has joined #kata-dev | 03:58 | |
| *** eernst has quit IRC | 04:05 | |
| *** zerocoolback has joined #kata-dev | 04:20 | |
| *** mylinux has joined #kata-dev | 04:56 | |
| *** pabelanger has quit IRC | 04:58 | |
| *** mylinux has quit IRC | 05:01 | |
| *** sjas has joined #kata-dev | 05:57 | |
| *** sjas_ has quit IRC | 06:00 | |
| *** zerocoolback has quit IRC | 06:08 | |
| *** zerocoolback has joined #kata-dev | 06:08 | |
| *** zerocoolback has quit IRC | 06:10 | |
| *** zerocoolback has joined #kata-dev | 06:11 | |
| *** zerocoolback has quit IRC | 06:14 | |
| *** zerocoolback has joined #kata-dev | 06:15 | |
| *** zerocoolback has quit IRC | 06:19 | |
| *** dims has quit IRC | 06:24 | |
| *** zerocoolback has joined #kata-dev | 06:27 | |
| *** zerocoolback has quit IRC | 06:30 | |
| *** zerocoolback has joined #kata-dev | 06:30 | |
| *** dims has joined #kata-dev | 06:30 | |
| kata-dev-irc-bot | <sebastien.boeuf> @laijs hey | 06:32 |
| kata-dev-irc-bot | <laijs> hey | 06:32 |
| kata-dev-irc-bot | <sebastien.boeuf> thanks for the comment here https://github.com/kata-containers/agent/pull/163 | 06:33 |
| kata-dev-irc-bot | <sebastien.boeuf> what do you think about having a field `Rootfs bool` which would remove any confusion | 06:33 |
| kata-dev-irc-bot | <laijs> you are welcome | 06:33 |
| *** annabelleB has quit IRC | 06:33 | |
| kata-dev-irc-bot | <laijs> I don’t think so | 06:34 |
| kata-dev-irc-bot | <laijs> in host we do have storage drivers like aufs, devmapper, btrfs,… | 06:34 |
| *** zerocoolback has quit IRC | 06:34 | |
| kata-dev-irc-bot | <laijs> pb.Storage is designed to partially simulate it | 06:35 |
| kata-dev-irc-bot | <sebastien.boeuf> I get it but in that case the whole current implementation of agent is wrong | 06:36 |
| kata-dev-irc-bot | <sebastien.boeuf> I was more thinking about Storage like the description of what need to be mounted | 06:36 |
| kata-dev-irc-bot | <sebastien.boeuf> and Device about what actually represent a device | 06:36 |
| kata-dev-irc-bot | <sebastien.boeuf> and basically your block device is gonna be both | 06:37 |
| kata-dev-irc-bot | <laijs> I guest you want to split Device out from storage | 06:37 |
| kata-dev-irc-bot | <sebastien.boeuf> what do you mean ? | 06:37 |
| kata-dev-irc-bot | <laijs> pb.Storage handles all storage things likes host’s aufs, devmapper, btrfs. but it also need to reslove the device which is duplicated with pb.Device | 06:39 |
| kata-dev-irc-bot | <sebastien.boeuf> yes exactly, that's why I need both in this case to solve such a case | 06:40 |
| kata-dev-irc-bot | <sebastien.boeuf> I thought the split was clear when I have introduced `Device` | 06:40 |
| kata-dev-irc-bot | <sebastien.boeuf> cause if I follow your logic, then everything should be described in Storage | 06:41 |
| kata-dev-irc-bot | <sebastien.boeuf> but I was really thinking that Storage was responsible for everything related to the filesystem, meaning a mount basically, while device was responsible for all the details related to the device itself | 06:42 |
| kata-dev-irc-bot | <laijs> yes | 06:44 |
| kata-dev-irc-bot | <sebastien.boeuf> for a devicemapper rootfs, we have a hotplugged block device which will show up inside the VM, this needs to be taken care of by `Device`. Once this device has been properly detected inside, we rely on Storage to know where this device should be mounted and with which filesystem | 06:44 |
| *** annabelleB has joined #kata-dev | 06:46 | |
| kata-dev-irc-bot | <laijs> actually, device can be bind mounted from vm_path into container_path via pb.Storage instead of `mknod`, It is allowed by oci-spec, shall we try this. | 06:48 |
| kata-dev-irc-bot | <laijs> so that we can focus on pb.Storage | 06:50 |
| kata-dev-irc-bot | <sebastien.boeuf> hmm, I get your point. So we would use `Device` for any device passed through the VM, while we would use `Storage` for rootfs and any volume | 06:52 |
| kata-dev-irc-bot | <sebastien.boeuf> And both `Device` and `Storage` would actually call into the same functions inside the agent to wait for a block device for instance | 06:54 |
| kata-dev-irc-bot | <laijs> we would use `Storage` for rootfs and any volume, and we can also use Storage for any device passed through the VM(use bind mount), so we can implement the driver in one place. | 06:54 |
| kata-dev-irc-bot | <laijs> At least, I insist that we would use `Storage` for rootfs and any volume. | 06:55 |
| kata-dev-irc-bot | <sebastien.boeuf> This makes sense, but this goes further than my SCSI PR then | 06:57 |
| kata-dev-irc-bot | <sebastien.boeuf> we need to rework the way it works right now for blk | 06:57 |
| kata-dev-irc-bot | <sebastien.boeuf> But even if we do that, it'd be easier to have a flag to identify we're dealing with a rootfs. I am saying this because for rootfs, there will be no corresponding entry into `oci.Device` or `oci.Mount` | 07:00 |
| kata-dev-irc-bot | <sebastien.boeuf> @laijs ok here is the proposal. I can rework the current block device handling so that we actually use Storage if it's a rootfs or a volume passed as a device. `Storage` will only be related to `oci.Mount` and `Device` to `oci.Device`. This way we need one function `AddStorages()` and one `AddDevices()`, both called from `CreateContainer()` in `grpc.go`. They can both be called interchangeably, because they are not tied by | 07:11 |
| kata-dev-irc-bot | defining the same things. And I really want to add a `RootFs bool` field to the `Storage` structure so that we can cleanly identify it. | 07:11 |
| kata-dev-irc-bot | <laijs> > there will be no corresponding entry into `oci.Device` or `oci.Mount` but there will be ocispec.Root.Path which can also help with it to use the rootfs | 07:13 |
| kata-dev-irc-bot | <laijs> use pb.Storage to mount the rootfs on some place and use ocispec.Root.Path to point to it. | 07:14 |
| *** zerocoolback has joined #kata-dev | 07:17 | |
| kata-dev-irc-bot | <sebastien.boeuf> but you need to make the diff with a volume | 07:20 |
| kata-dev-irc-bot | <sebastien.boeuf> @laijs I gotta go, but please explain on the PR https://github.com/kata-containers/agent/pull/163 how you see things to work exactly, and I'll try to implement it tomorrow | 07:21 |
| *** jodh has joined #kata-dev | 07:37 | |
| *** jodh has quit IRC | 07:37 | |
| *** jodh has joined #kata-dev | 07:37 | |
| *** zerocoolback has quit IRC | 07:47 | |
| *** annabelleB has quit IRC | 08:14 | |
| *** annabelleB has joined #kata-dev | 08:43 | |
| *** annabelleB has quit IRC | 08:56 | |
| *** gwhaley has joined #kata-dev | 09:13 | |
| *** mylinux has joined #kata-dev | 09:18 | |
| *** zerocoolback has joined #kata-dev | 09:38 | |
| *** pabelanger has joined #kata-dev | 09:41 | |
| *** cdent has joined #kata-dev | 09:47 | |
| *** mylinux has quit IRC | 09:50 | |
| *** cdent has quit IRC | 10:03 | |
| *** zerocoolback has quit IRC | 10:05 | |
| *** mylinux has joined #kata-dev | 10:21 | |
| *** mylinux has quit IRC | 10:27 | |
| *** mylinux has joined #kata-dev | 10:53 | |
| *** mylinux has quit IRC | 10:58 | |
| *** mylinux has joined #kata-dev | 11:24 | |
| *** cdent has joined #kata-dev | 11:27 | |
| *** mylinux has quit IRC | 11:30 | |
| *** mylinux has joined #kata-dev | 11:58 | |
| *** gwhaley has quit IRC | 12:00 | |
| *** mylinux has quit IRC | 12:03 | |
| *** zerocoolback has joined #kata-dev | 12:04 | |
| *** pabelanger_ has joined #kata-dev | 12:27 | |
| *** pabelanger has quit IRC | 12:29 | |
| *** zerocoolback has quit IRC | 12:49 | |
| *** annabelleB has joined #kata-dev | 13:09 | |
| *** pabelanger_ is now known as pabelanger | 13:10 | |
| *** gwhaley has joined #kata-dev | 13:22 | |
| *** zerocoolback has joined #kata-dev | 13:24 | |
| *** mylinux has joined #kata-dev | 13:24 | |
| *** mylinux has quit IRC | 13:29 | |
| stefanha | What is the advantage of USE_DOCKER=true rootfs.sh? It's not obvious to me why I would want or not want to use Docker to build the rootfs. | 13:30 |
| *** fuentess has joined #kata-dev | 13:32 | |
| *** fuentess has quit IRC | 13:37 | |
| *** devimc has joined #kata-dev | 13:38 | |
| kata-dev-irc-bot | <ydjainopensource> U may use docker to avoid installing dependencies directly on your computer | 13:42 |
| kata-dev-irc-bot | <ydjainopensource> With use_docker a container with dependencies required to build a rootfs is created which in turn generates the rootfs | 13:43 |
| gwhaley | exactly - and the script only supports some of the dependancies for some distros I think - so we use USE_DOCKER on a fedora system for instance | 13:46 |
| *** annabelleB has quit IRC | 14:03 | |
| stefanha | ydjainopensource & gwhaley: Thanks! | 14:14 |
| *** rcw has joined #kata-dev | 14:15 | |
| *** eocardon has joined #kata-dev | 14:35 | |
| *** eernst has joined #kata-dev | 15:21 | |
| *** mylinux has joined #kata-dev | 15:24 | |
| *** mylinux has quit IRC | 15:28 | |
| *** mylinux has joined #kata-dev | 15:49 | |
| *** zerocoolback has quit IRC | 16:18 | |
| *** eernst has quit IRC | 16:19 | |
| *** mylinux has quit IRC | 16:21 | |
| *** zerocoolback has joined #kata-dev | 16:24 | |
| *** mylinux has joined #kata-dev | 16:28 | |
| *** zerocoolback has quit IRC | 16:29 | |
| *** justJanne is now known as justjanneTest | 16:31 | |
| *** justjanneTest is now known as justJanne | 16:31 | |
| *** mylinux has quit IRC | 16:33 | |
| *** mylinux has joined #kata-dev | 16:34 | |
| *** mylinux has quit IRC | 16:38 | |
| *** devimc has quit IRC | 16:52 | |
| *** eernst has joined #kata-dev | 16:52 | |
| *** mcastelino has joined #kata-dev | 17:00 | |
| *** mcastelino has quit IRC | 17:00 | |
| *** mcastelino has joined #kata-dev | 17:00 | |
| *** gabyc_ has joined #kata-dev | 17:16 | |
| *** oikiki has joined #kata-dev | 17:42 | |
| *** jodh has quit IRC | 18:04 | |
| *** devimc has joined #kata-dev | 18:07 | |
| *** gabyc_1 has joined #kata-dev | 18:08 | |
| *** gabyc_ has quit IRC | 18:08 | |
| *** manjeets has joined #kata-dev | 18:20 | |
| *** gwhaley has quit IRC | 18:46 | |
| *** cdent has quit IRC | 18:56 | |
| *** oikiki has quit IRC | 19:41 | |
| *** eernst has quit IRC | 20:10 | |
| *** eernst has joined #kata-dev | 20:10 | |
| *** eernst has quit IRC | 20:54 | |
| *** eernst has joined #kata-dev | 20:54 | |
| *** rcwoolley_ has joined #kata-dev | 21:07 | |
| *** rcw has quit IRC | 21:10 | |
| *** rwoolley__ has joined #kata-dev | 21:14 | |
| *** rwoolley__ has quit IRC | 21:15 | |
| *** rcwoolley_ has quit IRC | 21:16 | |
| *** oikiki has joined #kata-dev | 21:29 | |
| *** oikiki has quit IRC | 21:31 | |
| *** oikiki has joined #kata-dev | 21:32 | |
| *** oikiki has quit IRC | 21:39 | |
| *** oikiki has joined #kata-dev | 22:05 | |
| *** oikiki has quit IRC | 22:13 | |
| *** oikiki has joined #kata-dev | 22:13 | |
| *** eernst has quit IRC | 22:21 | |
| *** eernst has joined #kata-dev | 22:22 | |
| *** devimc has quit IRC | 22:27 | |
| *** annabelleB has joined #kata-dev | 22:32 | |
| *** annabelleB has quit IRC | 22:33 | |
| *** annabelleB has joined #kata-dev | 22:34 | |
| *** gabyc_1 has quit IRC | 22:36 | |
| *** oikiki has quit IRC | 23:11 | |
| *** oikiki has joined #kata-dev | 23:19 | |
| *** annabelleB has quit IRC | 23:24 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!