We are commissioning a new OpenNebula 4.8 head node and so
far we are seeing very good scalability. But
in tests where we are launching a lot of virtual machines at once
we sometimes see the error:
Tue Oct 7 08:46:43 2014 [Z0][VMM][I]: error: Failed to create domain from
/var/lib/one//datastores/100/240/deployment.0
Tue Oct 7 08:46:43 2014 [Z0][VMM][I]: error: monitor socket did not show
up.: No such file or directory
If we go to the system data store /var/lib/one/datastores/100/240/
and do a manual "virsh create deployment.0"
the VM is then created correctly.
So I presume this is some kind of race condition inside the VM host.
The question is what might that be? some strange condition where
two virsh creates are going after the same VNC port and one of them
is failing?
From /var/log/messages on the vm host I get the following
Oct 7 08:46:38 cloudworker1036 libvirtd: 31490: info :
virNetDevProbeVnetHdr:94
: Enabling IFF_VNET_HDR
Oct 7 08:46:38 cloudworker1036 kernel: device vnet8 entered promiscuous
mode
Oct 7 08:46:39 cloudworker1036 kernel: br1: port 10(vnet8) entering
forwarding
state
Oct 7 08:46:42 cloudworker1036 libvirtd: 31490: error :
qemuMonitorOpenUnix:300
: monitor socket did not show up.: No such file or directory
Oct 7 08:46:42 cloudworker1036 libvirtd: 31490: info :
qemuConnectMonitor:1426
: Failed to connect monitor for one-240
Steve Timm
------------------------------------------------------------------
Steven C. Timm, Ph.D (630) 840-8525
t...@fnal.gov http://home.fnal.gov/~timm/
Fermilab Scientific Computing Division, Scientific Computing Services Quad.
Grid and Cloud Services Dept., Associate Dept. Head for Cloud Computing
_______________________________________________
Users mailing list
Users@lists.opennebula.org
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
