[lxc-devel] Probably more of a user list than a devel lis question but ...

Michael J Coss michael.coss at alcatel-lucent.com
Thu May 8 17:19:52 UTC 2014


On 5/8/2014 9:41 AM, Serge Hallyn wrote:
> Probably only means that you have a mtab file and should use
> mount -n.
Agreed.
> This is not mysterious.  You're mounting sysfs under /etc/lxc/xxx/sys
> from lxc.clone.mount hook.  This won't show up on the host bc you're
> already in a new ns.  Next lxc will pivot_root you under
> $LXC_ROOTFS_MOUNT and then umount the oldroot, so your
> /etc/lxc/xxx/sys gets umounted.  Now fuse can't parse it bc it's gone.
Unfortunately, I can't seem to get the right combination of namespaces 
or timing to do what I need.  Ultimately, I really need a containerized 
version of sysfs.  I was hoping that I could mimic that by having a FUSE 
provide a /sys to the container.  If sysfs were completely identical 
when mounted in a container and host, this wouldn't be an issue but that 
is not the case.  In addition, I need to allow for dynamic changes to 
the view that the container sees in it's /sys.  The plan was to have the 
FUSE run on the host, and provided a filtered view of /sys to the 
container.  But for that to work, I need the FUSE to run in the network 
namespace (not sure if that is the minimum needed) of the container, and 
need it launched before init starts running.  And I need to have access 
to an instance of sysfs for that network namespace for the FUSE to present.

-- 
---Michael J Coss



More information about the lxc-devel mailing list