I'm sure this has been answered somewhere on the forums, but after much digging I've yet to find the exact "how" to answer this question.
All of my developmental / sandboxing of ProxMox up to this point has been using the native RAW images on my ProxMox (using ZFS RAIDz2) server. I do know that RAW is only available because that's a block level image that runs on top of ZFS natively. However, in my reading, it appears the consensus is that QCOW2 is better than RAW feature wise, and some also throw ZVOLS into the mix as well.
How do I successfully create QCOW2 or ZVOLS on ZFS RaidZ2 servers? My best understanding seems to create a dataset on the ZFS drives which is block level, then within the dataset, create QCOW2 images there for the VM's. Is that correct? Performance hits seem to be minimal at best.
What about ZVOLS? I've yet to figure out how to create and use those on my server.
FYI, this is still a sandbox server and NOT a live deployment. I'm using this to learn before doing a live deployment.
All of my developmental / sandboxing of ProxMox up to this point has been using the native RAW images on my ProxMox (using ZFS RAIDz2) server. I do know that RAW is only available because that's a block level image that runs on top of ZFS natively. However, in my reading, it appears the consensus is that QCOW2 is better than RAW feature wise, and some also throw ZVOLS into the mix as well.
How do I successfully create QCOW2 or ZVOLS on ZFS RaidZ2 servers? My best understanding seems to create a dataset on the ZFS drives which is block level, then within the dataset, create QCOW2 images there for the VM's. Is that correct? Performance hits seem to be minimal at best.
What about ZVOLS? I've yet to figure out how to create and use those on my server.
FYI, this is still a sandbox server and NOT a live deployment. I'm using this to learn before doing a live deployment.