Unable to read from monitor: Connection reset by peer

Issue

• When launching a new instance in a specific compute node fails with following error(compute.logs):

[...]
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d]     rv = execute(f,*args,**kwargs)
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d]   File "/usr/lib/python2.6/site-packages/eventlet/tpool.py", line 77, in tworker
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d]     rv = meth(*args,**kwargs)
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d]   File "/usr/lib64/python2.6/site-packages/libvirt.py", line 716, in createWithFlags
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d]     if ret == -1: raise libvirtError ('virDomainCreateWithFlags() failed', dom=self)
2015-07-31 11:13:30.214 22036 TRACE nova.compute.manager [instance: a3b157ba-09b9-408d-9fdd-1ae677f4dc2d] libvirtError: Unable to read from monitor: Connection reset by peer
[...]