[Openstack] disk on controller node fills up, now all nova-compute processes show as down

JR botemout at gmail.com
Thu Jul 10 22:45:49 UTC 2014


Greetings all,

I have a 4 node havana cluster (set up using the rackspace openstack
cookbooks) with 1 controller/compute node and 3 compute nodes.

Due some excessive logging, I filled up /.  Now all nova-compute nodes
when running nova-manage service list show state XXX (see
http://pastebin.com/1ZGvgvnR).

I've tried restarting all the nova processes on all the nodes,
rabbitmq-server, mysql as well as rebooting the controller node (after,
obviously, having made disk space available).  The web UI responds and
lists the running instances though I can't start new instances (of
course).  The dates on the machines are all the same (I've seen
something similar when time gets out of whack). The mysql db seems whole ...

The log files show essentially nothing.

Anyone ever seen this before?

Any place I can look (some mysql table)?

Thanks much,

JR


-- 
Your electronic communications are being monitored; strong encryption is
an answer. My public key
<http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x4F08C504BD634953>




More information about the Openstack mailing list