[Openstack-operators] Unexpected Removal of Shared Storage

Tzach Shefi tshefi at redhat.com
Wed Feb 26 15:51:56 UTC 2014

Hi Samuel, 

Not sure from your description, this shared storage does it hold ONLY images, or anything else as well (in else I mean that's related to openstack)? 

If ONLY images are saved on this storage, from my short experience with Openstack, i learned that images are copied from Glance (image server) to local computes. 
And so I'd expect your running instances wouldn't be effected. 
But you would have problems booting up new instances, that is until you restore image shared storage. 

Be sure to test this in a lab before going production, i'll try to verify it in my lab. 

If on the other hand this shared storage also holds Cinder volumes (instance persistent disks) or computes are connected to it, DON'T mess around with it, shut everything off and boot up again later.

As per second question not sure about answer, any way why your asking about this for what purpose?  
When you say "disk" do you mean instance's disk or the compute's disk? Do you use Cinder volume (persistent disks) ? 


----- Original Message -----
From: "Samuel Winchenbach" <swinchen at gmail.com>
To: openstack-operators at lists.openstack.org
Sent: Friday, February 21, 2014 9:39:31 PM
Subject: [Openstack-operators] Unexpected Removal of Shared Storage

Hi everyone, 

I was wondering what happens if the shared storage holding the disk images (NFS for example) is unexpectedly disconnected from the compute host? 

Also does pausing an instance prevent qemu/kvm from writing anything to disk? I would assume it would but I thought I should ask. 

We need to do some maintenance to our NFS server and I would like to do it without stopping (more disruptive to the user) or suspending (no as feasible due to the number of instances). 


OpenStack-operators mailing list
OpenStack-operators at lists.openstack.org

More information about the OpenStack-operators mailing list