Adding metadata vdev/fusion pool to an existing pool

I see from the documentation (Fusion Pools | TrueNAS Documentation Hub) that we can add fusion pools/metadata vdevs when creating new pools. I already have a truenas scale setup which includes one already created pool that has 1 data vdev and no metadata vdev. I’m looking to add another data vdev to increase the storage capacity for the pool. With the addition of a new vdev, I’d also be looking to add a metadata vdev as well.

So, some questions about that.

  1. Does each data vdev requires its own metadata vdev or each pool requires one metadata pool only?
  2. In case each data vdev requires its metadata vdev, can I create a metadata vdev for a data vdev that has been created already?
  3. In case one metadata vdev can serve a pool, what would be the purpose of multiple metadata vdevs within a single pool?


Each pool. It’s a pool-wide vdev.

That’s not really the right question (see above). However, keep in mind that any data/metadata that has already been written to disk will not use the special vdev. It only applies to future writes.

More capacity for the special blocks? Balanced reads/writes?

Please have an understanding of what you’re getting into. A special vdev is as important as a data vdev. If it fails or you lose it? You lose your entire pool.

Please use caution, and don’t think you need something just because it sounds useful.

EDIT: This is not hypberbole. You can find examples (in the old forums) of users losing their entire pool because they played around with a Dedup vdev or special metadata vdev.


Hi @winnielinnie,

Thank you for the response and especially for the word of caution. We run into slowdowns during times of higher network traffic so I’ve been looking into getting rid of any potential bottlenecks, hence the curiosity around it. I’d certainly tread cautiously given the risk associated and would exhaust safer options first before going for this one.

Cheers, thanks a lot.