apt / dpkg broken in tkl lxc container

I am using the lxc template version of the TKL LAMP

Somehow my apt/dpkg seems to be broken.

From inside the container I do an apt-get update and after some output it gives an error...

Hit http://security.debian.org wheezy/updates/main Translation-en
Hit http://cdn.debian.net wheezy/main Translation-en
E: dpkg was interrupted, you must manually run 'dpkg --configure -a' to correct the problem.

If I then run the command, it starts.... but..... It shutdowns the complete container.

root@tst01 ~# dpkg --configure -a
Setting up libc6:amd64 (2.13-38+deb7u1) ...
lxc-console: Input/output error - failed to read

So something is really broken here.... I can't imagine that the container breaks on this.

Any suggestions ?

Replies (3)

Hi Jeremy... (already

Reply 1 2014-03-28

Hi Jeremy... (already expected you would respond :)

what I did is using the standard lxc app and did the lxc-create with the lamp app

Then I tarred the whole container and brought it to my 'testing' server on RHEL 6 with lxc 0.75

Everything works ok there , expect when I started the apt-get upgrade.... when installing the updated libc6 it crashes on the post config stage...

I suspect that something is not handled correctly by lxc-init and it crashes the container, so i will try to build a newer lxc version and see how that goes.

On the TKL lxc app , everything still works... so that is good.

its complicated..... I have

Reply 2 2014-03-31

its complicated.....

I have build from src the latest lxc version 1.0.1...

When I now start the container I get

lxc-start: command get_cgroup failed to receive response

lxc-start: unknown capability mac_admin
lxc-start: failed to drop capabilities
lxc-start: failed to setup the container
lxc-start: invalid sequence number 1. expected 2
lxc-start: failed to spawn 'tst01'

I took a look in the config.... and I saw the mac_admin named for the lxc.drop capability.

only when I removed all the drop capabilities, the container started without problems.

However as soon as I did a shutdown from in the container.... the whole host went down

So bottom line (I think)... my RHEL host has a too old kernel/cgroups version

Somehow the 32 bit "old" TKL versions work within containers on RHEL without a problem :)

Hmmm That doesn't sound good Hans

Reply 3 2014-03-28

Without knowing how to recreate it I'm not sure... I've been meaning to take the LXC containers for a run so testing this is probably a good time! :)

If you have any ideas of what may have lead to it be great, otherwise I'll just do some road testing! I may not get to do much until next week though...