Open Stack

Or rather, the compute cannot talk to the conductor. Try setting use_local=true in the compute nova.conf and restart nova-compute.

> -----Original Message-----
> From: Dimitri Maziuk [mailto:dmaziuk at bmrb.wisc.edu]
> Sent: 12 May 2014 19:10
> To: openstack at lists.openstack.org
> Subject: [Openstack] instances stuck in "powering on"
> 
> Hi everyone,
> 
> I moved my bootable volumes to better storage (nfs backend) by copying
> files and editing provider_location in cinder db. The instances were
> shut off for that, of course.
> 
> Now when I power them on via dashboard, some (not all) of them get
> stuck in "powering on" for as long as I care to wait. On the compute
> host compute.log is full of
> 
> > TRACE nova.openstack.common.rpc.amqp
> > task run WARNING nova.openstack.common.loopingcall [-] outlasted
> > interval by 50.018454 sec ERROR nova.openstack.common.periodic_task
> [-] Error during ComputeManager._sync_power_states: Timeout while
> waiting on RPC response - topic: "conductor", RPC method:
> "object_class_action" info: "<unknown>"
> 
> -- completely unhelpful since I'm getting tons of them during normal
> operation, too.
> 
> "virsh start instance-XXXXXX" works, "nova reset-state --active" clears
> up the "powering-on" but doesn't mark it "Running" regardless of
> instance's power state.
> 
> So it looks like dashboard can't talk to the compute. Any hints as to
> why & where to look?
> 
> (centos 6.current, rdo, havana, simple stupid setup w/ noop nova
> networking and bootable volumes.)
> 
> TIA
> --
> Dimitri Maziuk
> Programmer/sysadmin
> BioMagResBank, UW-Madison -- http://www.bmrb.wisc.edu