Comment 0 for bug 1620614

neutron-netns-cleanup --force fails to some namespaces with a following traceback: http://paste.openstack.org/show/567145/. Apparently, those namespaces that couldn't be deleted got blank permissions:

root@node-525:~# ls -lv /var/run/netns
total 0
-r--r--r-- 1 root root 0 Sep 1 22:25 haproxy
-r--r--r-- 1 root root 0 Sep 5 11:16 qdhcp-70ac5f9d-ff11-4d5c-b23f-6d917d1696ae
---------- 1 root root 0 Sep 5 11:29 qdhcp-72ce9f3c-2eb6-431e-912c-1f0e9aee9ce9
-r--r--r-- 1 root root 0 Sep 5 11:17 qdhcp-40711dc5-2385-4b88-9499-c6587da967b0
-r--r--r-- 1 root root 0 Sep 5 11:16 qdhcp-bacedf9b-351a-497b-8282-3b32dd695ab0
-r--r--r-- 1 root root 0 Sep 5 11:16 qdhcp-ca67bf18-f042-4bcb-93d9-70a9cc170c87
-r--r--r-- 1 root root 0 Sep 5 11:16 qdhcp-dd5d3d69-50cc-4a07-bded-579e7a5a3404
---------- 1 root root 0 Sep 5 12:19 qrouter-72e21f1b-8f96-4ac5-ac05-4095a617a581
---------- 1 root root 0 Sep 5 13:52 qrouter-c5111cd2-9d88-4420-b1b1-6df1c0d2bc68
-r--r--r-- 1 root root 0 Sep 1 22:25 vrouter

This could happen due to attempt to delete an namespace when it still has running processes [1].

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1095015#c2