It sounds like you are using SMB to connect to your SAN, while I am not saying it will not work, but I might also expect worse/functionality/stability with that over something like NFS or iSCSI. I am no SAN expert by any means, so I can't tell you whether or not it would be a good idea or not to use some other protocol, but instinctively it sounds like a good idea to me.
I ran Proxmox clusters before mapping back VM disk image data stores to Netapp filers via NFSv3 - it’s not like you can’t run VMs on it - I also ran ONTAP 7-mode where I had stuff shared on CIFS (just not VM disk images) on the same filer automapped as data directories on VM images shared out via a different mountpoint on NFSv3 (It’s a trading firm where our back office stuff is run on a Windows laptop and reading it via a shared directory, which needs to pull analysis run on Linux VMs). A well managed NAS/SAN should work equally well attached on either Linux (Proxmox), Windows Server (HyperV) and/or VMWare VCenter.
however, there are certain gotchas you need to be aware of -
a) If it’s CIFS, make sure your CIFS/SMB server is sane, and that you won’t get tripped up on permissions. If it’s NFS, make sure that the exports are sane, and for iSCSI, the target/initiatior is correct. The smoking gun log issue is usually not on the VM host side, it’s on the SAN/SAN side.
b) if your server has multiple VLANs and NICs, make sure that they have sane network configurations - I remember a failover for a SAN fail 25% of the VMs in a cluster because someone put a /23 in a network interface net mask instead of a /22. As a corollary, never assume that the network connectivity to the SAN/NAS is fault free. If it glitches out or drops packets you will see intermittent mount/read/write issues.
c) on the VM datastores, make sure you don’t have stale lock files, or something happened and the NAS/SAN went into read-only for that datastore mountpoint. I remember a QSFP module flipped out in my gig, the SAN freaked out and all of a sudden 16 VMs locked up. It has happened before. When in doubt, ssh into the hypervisor, get into the data store location and see if you can touch a test file. If it just sit there, it’s probably the storage acting up.
).
