Hello,
I am currently setting up a new Incus virtualization environment and ran into
some unexpected QEMU behavior in regards to RBD block devices.
I have two VMs that get their block dev inserted with the same QMP command[0],
yet one acquires a RBD lock (Checked with "rbd lock ls") and the other does
not.
Do you have any hints on why this may happen?
Or do you think I should ask the Ceph people about this.
We are running QEMU version 7.2.17 (Debian 1:7.2+dfsg-7+deb12u13).
I also could not find a way to set the lock to exclusive(See [1] for Ceph
documentation on that feature), did I miss it or is it not supported?
Best regards,
Julian
[0] {"execute":"blockdev-add","arguments":{"cache":{"direct":false,"no-
flush":false},"conf":"/etc/ceph/
ceph.conf","discard":"unmap","driver":"rbd","image":"virtual-
machine_test1.block","key-secret":"pool_incus_incus","node-
name":"incus_root","pool":"incus","read-only":false,"user":"incus"}}
