Hey all! I use a ZFS pool of mirrored vdevs that backs my ESXi datastore, and I recently moved it from being on a separate TrueNAS server to being on my Unraid system. Everything seems to work basically the same as it did before - with one exception.

No matter how much activity the pool has I never see any indication of any activity to the log device - it always just shows 0.0B/s. While I'll be the first to admit I'm not an expert, I do know enough to know that the zlog device isn't always used. My understanding though was that in my use case with an NFS share and quite a few running VM's it should definitely be being used - and I'm 95+% sure this was happening in TrueNAS.

Anyone have any pointers on what on earth could I be missing here? Thank you!!

I

Update on this for anyone else who comes along down the road. I'm not sure where the issue here lies with sync=standard on the dataset backing the ESXi shared storage...I suspect something with Unraid given the sheer size/age/development of something like vSphere, but I cannot prove that. In any case, I ended up running a "zfs set sync=always [pool_name]/[dataset_name]" and that did the trick for me here. I now see lots of writes to the log device and am seeing close-enough speeds to what I had before local disk performance wise, while hopefully being a little more protected VM-integrity wise.