~alpine/users

Alpine issues in an unpriviledged LXC container

Details
Message ID
<7b0b6e2a9f0c94fac89d5ba8f19e4ee1@nebelschwaden.de>
DKIM signature
missing
Download raw message
 

Hello,

I am having trouble getting alpine to run properly in an
unpriviledeged lxc container attached to an ordinary bridge. Any service
I install fails because the networking service fails to start. 
The
network itself is working, I can ssh into the container and install
software, just the networking service is failing. 

Further, when I edit
the init files and remove the net dependency, they also do work (so far
at least). But this feels a bit hackish. 

I may suspect the
/etc/networking/interfaces file, but I am not sure what to put in here.


The network card inside the container is veth1@if9, where the index in
"@if" is not reliable, it may change with each boot. And I have no idea
how to get rid of the latter part. 

However, I'ver tried:

empty
interfaces file,
iface veth1 inet static
iface veth1@if9 inet
static

None did the job. Interestingly, /var/log/messages has no
information but the fact of the failure itself: 

daemon.err
/etc/init.d/networking[1905]: ERROR: networking failed to start

Sub
minutely I do see other error messages in the log, not sure, wether they
may point to the same cause:

Aug 2 14:23:17 halluzinogen daemon.info
init: process '/sbin/getty 38400 console' (pid 1875) exited. Scheduling
for restart.
Aug 2 14:23:17 halluzinogen daemon.info init: starting pid
1902, tty '': '/sbin/getty 38400 console'
Aug 2 14:23:17 halluzinogen
auth.err getty[1902]: can't open '/dev/console': Permission denied^Mvi

Aug 2 14:23:27 halluzinogen daemon.info init: process '/sbin/getty
38400 console' (pid 1902) exited. Scheduling for restart.
Aug 2 14:23:27
halluzinogen daemon.info init: starting pid 1904, tty '': '/sbin/getty
38400 console'
Aug 2 14:23:27 halluzinogen auth.err getty[1904]: can't
open '/dev/console': Permission denied^M

# ls -la
/dev/console
c--x--x--- 1 root nobody 136, 2 Aug 2 10:08
/dev/console

Changing the mode of /dev/console to 600 gets me rid of
those messages, however after a restart of the container it is back to
mode 0110. Maybe a bug, maybe I am missing something. 

Additionally,
despite being enabled @boot, syslog does not come up at boot. However,
starting it manually does work as well. 

Any idea of what may be wrong
or how to hunt the errors down? Especially those with the network? To
get the whole install a little polished?

Thanks

Ede
Reply to thread Export thread (mbox)