Linode上的Docker

评论 (17)

1. 
   Scott January 3, 2014 at 2:25 pm

   Awesome! Note that Docker is for 64bit only I believe. Someone else may confirm?

   Reply
2. 
   Les Aker January 3, 2014 at 2:29 pm

   That’s correct. From my understanding, they’re currently imposing the 64 bit restriction because it keeps their code cleaner and the benefits of having a 32 bit “host” are minimal.

   Our 32 bit kernel has all the right bits flipped as well, so if they do start supporting 32 bit down the line, it’ll work with our kernel too.

   Reply
3. 
   Nick Stinemates January 3, 2014 at 2:38 pm

   I’ll be switching my linode instances today.

   Thanks for the awesome work, guys.

   Reply
4. 
   Nikola January 3, 2014 at 3:28 pm

   I’m trying to understand why people pushing to use docker on cloud instead of dedicated server. In the cloud it adds additional unnecessary level of abstraction.

   Reply
5. 
   Leo January 3, 2014 at 3:59 pm

   @Nikola

   Ease of deployment.

   Reply
6. 
   Brian January 3, 2014 at 4:41 pm

   Ease of Deployment, Manageability, portability and reduces the security surface while able to use less hardware to support the larger VM’s. Dedicated VM’s are in the past.

   Reply
7. 
   NathanielInKS January 3, 2014 at 5:15 pm

   I was asked if it there is a way to autoscale Docker containers?

   Reply
8. 
   Andrew January 3, 2014 at 5:51 pm

   @nikola, here is my guess:

   If you have a docker instance on your local dev machine, and a docker instance on your dedicated server, you can develop locally & have more confidence it will Just Work ™ when you push to the cloud. Much easier to configure a local docker & a remote docker identically than to configure your dev machine & cloud identically. Can anyone confirm: is that the general idea?

   Reply
9. 
   camper67 January 3, 2014 at 7:39 pm

   This is very interesting. With a little glue, you can use this to scale linode(s) up/down the way you can scale aws, except that linode’s VMs don’t suck 🙂

   Maybe that’s for Q2, after the SSD hybrid rolls out in Q1? Pretty exciting stuff, thanks!

   Reply
10. 
    Aleksejs Ivanovs January 4, 2014 at 2:22 am

    Great stuff, keep it coming!

    Reply
11. 
    Jed January 4, 2014 at 6:48 am

    @NathanielInKS Check out the Flynn project (https://flynn.io) … Afaiu it aims to create an auto-scaling Heroku-like wrapper around Docker. Still in developer preview at the moment though I think.

    Reply
12. 
    Jeff Utter January 22, 2014 at 2:23 am

    If I have a linode setup “the old way” (Ubuntu 12.04). With pv-grub. Can I just switch to the new kernel, check xenify and reboot?

    Also does this new kernel have the kernel options for limiting memory on docker containers: “cgroup_enable=memory swapaccount=1” enabled?

    Reply
13. 
    methou January 23, 2014 at 2:10 am

    it seems that linux-image-extra-3.12.6-* is not included in the default repos.

    Reply
14. 
    Robert February 4, 2014 at 3:15 pm

    I just checked this with my ubuntu box. I switched off pv-grub then proceeded to load the latest daily build of lxc. Note that if you are on Ubuntu 12.04 you might have to do it like this.

    apt-get install –no-install-recommends lxc cgroup-lite lxc-templates

    This is due to a recommend entry for uidgen which is unavailable. I’m not sure why it was added though.

    Anyways I ran lxc-checkconfig and confirmed all necessary supports are enabled for lxc to run all by itself. 🙂

    shinji@icarus:~$ uname -a
    Linux icarus.robertpendell.com 3.12.6-x86-linode55 #2 SMP Tue Jan 14 08:41:36 EST 2014 i686 i686 i386 GNU/Linux
    shinji@icarus:~$ sudo lxc-checkconfig
    — Namespaces —
    Namespaces: enabled
    Utsname namespace: enabled
    Ipc namespace: enabled
    Pid namespace: enabled
    User namespace: enabled
    Network namespace: enabled
    Multiple /dev/pts instances: enabled

    — Control groups —
    Cgroup: enabled
    Cgroup clone_children flag: enabled
    Cgroup device: enabled
    Cgroup sched: enabled
    Cgroup cpu account: enabled
    Cgroup memory controller: missing
    Cgroup cpuset: enabled

    — Misc —
    Veth pair device: enabled
    Macvlan: enabled
    Vlan: enabled
    File capabilities: enabled

    Note : Before booting a new kernel, you can check its configuration
    usage : CONFIG=/path/to/config /usr/bin/lxc-checkconfig

    Note disregard the memory contoller being marked as missing. As far as I know it requires a kernel startup option to be set which we can’t do with the linode kernels. Also it isn’t enabled in the config anyways probably for the same reason. This only prevents you from setting a memory limit on the containers.

    Reply
15. 
    victor February 18, 2014 at 9:58 pm

    Error: container_delete: Cannot destroy container d5d4d7f442d7: Driver devicemapper failed to remove init filesystem d5d4d7f442d74b17824cbcf1216cb3730053f8cfaefe8a1ea12d328451fc36d7-init: Error running removeDevice
    2014/02/19 10:55:59 Error: failed to remove one or more containers
    how to fix it?

    Reply
16. 
    Adam March 20, 2014 at 3:40 pm

    Same problem…

    Driver devicemapper failed to remove init filesystem
    Driver devicemapper failed to remove root filesystem

    Not sure how to fix this yet…

    Reply
17. 
    Les Aker March 20, 2014 at 11:50 pm

    That’s an upstream Docker problem, not something Linode-specific. I’d recommend reaching out to them via their bug tracker.

    Reply