I'm still not convinced by this use case. You're using a shared instance path (/opt/stack/data/nova/instances/) with individual compute image caches hosted within it that are duplicating cached images?!
/opt/stack/data/nova/instances/_base_192.168.0.1/1da480fd0bd6d365cce06421659d922b1bbed441 /opt/stack/data/nova/instances/_base_192.168.0.2/1da480fd0bd6d365cce06421659d922b1bbed441
How is this better than hosting the cached image once for all computes?
I'm still not convinced by this use case. You're using a shared instance path (/opt/stack/ data/nova/ instances/ ) with individual compute image caches hosted within it that are duplicating cached images?!
/opt/stack/ data/nova/ instances/ _base_192. 168.0.1/ 1da480fd0bd6d36 5cce06421659d92 2b1bbed441 data/nova/ instances/ _base_192. 168.0.2/ 1da480fd0bd6d36 5cce06421659d92 2b1bbed441
/opt/stack/
How is this better than hosting the cached image once for all computes?