Why I cannot assign CephFS mountpoint to the LXC, from GUI?

[twinsen]

I'm doing scale-out clustered Samba with CTDB on top and AD integerated, backed by CEPHFS storage.
This works pretty well, but to achieve what I needed was to use cmdline, like this:

pct set 100 -mp0 /mnt/pve/vnas,mp=/ceph,shared=1

For all Samba cluster member in LXC containers.
Can we have this also in the GUI? The mountpoint appears after that, but is not editable, however it is deletable.
Is there any particular reason this option is not in GUI?

 

[fabian]

well, it allows giving the container access to any host dir, so it's limited to root@pam at the moment. once the mappings used for hardware pass through are extended to also support pre-defining host dirs that can be bind-mounted into containers (or made available to VMs using virtiofsd), it's also possible to give access to those to regular users and it will likely be available on the UI.

 

[twinsen]

Of course, I could mount CEPHFS inside the container itself, by creating CEPH keyring for access, and mount it then, but as to my tests, this is much slower, as for some reason FUSE inside LXC is slower than this kind of mapping I did. Also, using VFS CEPH backend inside Samba registry is slower. That's why i decided to go that way, to explain you why Even on this LAB I'm doing this has only 1 Gbit networking, I'm able to saturate it fully with the traffic