generic disk polling for block devices
|Assignee:||Javi Fontan||% Done:|
|Category:||Drivers - VM|
|Target version:||Release 5.0|
I am opening an issue to track the progress of https://forum.opennebula.org/t/feature-3782-return-disk-and-snapshot-information-for-ceph-disks/990
#2 Updated by Anton Todorov over 5 years ago
What do you think about reworking the patch in this way? (please read the commit description for details) https://github.com/atodorov-storpool/one/commit/ac02b629d882c908a83f5b5751ab978c1196c04c
#5 Updated by Anton Todorov about 5 years ago
According https://forum.opennebula.org/t/updating-the-context-data/2137/7 is it possible to share what are the plans for this issue?
Just to remind that the current KVM.get_disk_usage() is still working for RBD and qcow2 only. IMO with little rework on the patch the interface can be opened and then it will work for block devices the (iSCSI, LVM, StorPool). It is trivial to report even the context ISO image size too.
The latest patch I am using for production installations can be found here https://github.com/OpenNebula/addon-storpool/blob/master/patches/vmm/4.14/01-kvm_poll.patch and the poll_disk_info script in https://github.com/OpenNebula/addon-storpool/blob/master/vmm/kvm/poll_disk_info
A general approach should be like: throw the VM XML (base64 encoded blob) to external script. I can't see the benefits of all this code to be in the ruby. Isn't it easier to just add in the VM XML's metadata info for each disk which TM_MAD is in use and call the corresponding TM_MAD tool. Even without the metadata info it will be much easier to hook if there is an open interface for the disks polling...