Permission denied when creating semaphore in ubuntu-cloud container

[Andreas Hasenack]

Deployed a precise host on canonistack, then a precise container inside that host:
andreas@nsn7:~$ juju status
environment: andreas-canonistack2
machines:
  "0":
    agent-state: started
    agent-version: 1.20.3
    dns-name: 10.55.32.177
    instance-id: b898eee8-9aef-4e6c-9749-dce73f490dc6
    instance-state: ACTIVE
    series: precise
    containers:
      0/lxc/0:
        agent-state: started
        agent-version: 1.20.3
        dns-name: 10.0.3.118
        instance-id: juju-machine-0-lxc-0
        series: precise
        hardware: arch=amd64
    hardware: arch=amd64 cpu-cores=1 mem=1024M root-disk=10240M
    state-server-member-status: has-vote
services:
  ubuntu-precise:
    charm: cs:precise/ubuntu-4
    exposed: false
    units:
      ubuntu-precise/0:
        agent-state: started
        agent-version: 1.20.3
        machine: 0/lxc/0
        public-address: 10.0.3.118

Test script:

andreas@nsn7:~$ juju ssh ubuntu-precise/0 'python -c "from multiprocessing import synchronize; synchronize.Lock()"'
Warning: Permanently added '10.55.32.177' (ECDSA) to the list of known hosts.
Warning: Permanently added '10.0.3.118' (ECDSA) to the list of known hosts.
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/lib/python2.7/multiprocessing/synchronize.py", line 147, in __init__
    SemLock.__init__(self, SEMAPHORE, 1, 1)
  File "/usr/lib/python2.7/multiprocessing/synchronize.py", line 75, in __init__
    sl = self._semlock = _multiprocessing.SemLock(kind, value, maxvalue)
OSError: [Errno 13] Permission denied
Connection to 10.0.3.118 closed.
ERROR subprocess encountered error code 1

On the precise host:
root@juju-andreas-canonistack2-machine-0:~# ls -la /dev/shm
lrwxrwxrwx 1 root root 8 Aug 6 21:51 /dev/shm -> /run/shm

root@juju-andreas-canonistack2-machine-0:~# grep shm /proc/mounts
none /run/shm tmpfs rw,nosuid,nodev,relatime 0 0

root@juju-andreas-canonistack2-machine-0:~# dpkg-query -W sysvinit-utils lxc
lxc 1.0.0~alpha1-0ubuntu14.1~ctools0
sysvinit-utils 2.88dsf-13.10ubuntu11.1

On the precise container:
root@juju-machine-0-lxc-0:~# ls -la /dev/shm
total 8
drwxr-xr-x 2 root root 4096 Jul 17 07:05 .
drwxr-xr-x 6 root root 4096 Aug 6 22:19 ..

root@juju-machine-0-lxc-0:~# grep shm /proc/mounts
none /run/shm tmpfs rw,nosuid,nodev,relatime 0 0
root@juju-machine-0-lxc-0:~#

Both are precise 12.04.5. The host was image b898eee8-9aef-4e6c-9749-dce73f490dc6 from canonistack's lcy02 region.

[Andreas Hasenack (ahasenack)]

I did the same deployment with the local provider, on my trusty host, and the precise container allowed that python snipped to be executed as a regular user without problems.

[Serge Hallyn (serge-hallyn)]

Thanks for submitting this bug. The problem is due to /run/shm not existing in the cloud images.

The ubuntu-cloud image fixes the /dev/shm by doing:

    if [ ! -L $rootfs/dev/shm ] && [ -d $rootfs/run/shm ] && [ -e $rootfs/dev/shm ]; then
        mv $rootfs/dev/shm $rootfs/dev/shm.bak
        ln -s /run/shm $rootfs/dev/shm
    fi

The second test fails because $rootfs/run/shm does not exist.

Since /run/shm will be created at runtime, I see no reason not to drop the second part
of the test. I'll suggest that to the list - there may be reasons I'm not considering.

[Launchpad Janitor (janitor)]

This bug was fixed in the package lxc - 1.1.0~alpha2-0ubuntu2

---------------
lxc (1.1.0~alpha2-0ubuntu2) utopic; urgency=medium

  * Cherry-pick usptream bugfix for lxc-usernic test.
 -- Stephane Graber <email address hidden> Thu, 02 Oct 2014 15:01:56 -0400