That’d require basically forking incus. As I said, non-trivial. For reference (snippet from backup.yaml file of a container instance dozer/.ix-virt/containers/test in my case):
In my case, before exporting the pool, I went through every app and instance that referenced it and made sure to remove it. I used the TrueNAS UI to remove all. The UI showed that I had removed the disk from the powered-off Instance.
Then I exported the pool.
Then it hung forever, and now Incus configs fail to validate/start.
When it’s back, I’ll add a random mount, remove it through UI, export the pool, and see if it happens again.
Can’t edit current Incus configs with incus config edit etc since it won’t start.
I don’t have that data as easily, but the latest chart shows a pretty clear trend.
TrueNAS 25.04.0 has been well adopted. We plan to make 25.04.1 available this week (week of May 26)… so this version has almost peaked. The following week, the IP addressing of Apps will be turned on (June 2).
13.0 is very strong… especially for standard storage services (SMB, NFS, iSCSI). The majority of our larger Enterprise customers are using it. Its very solid.
For NVMe and newer capabilities, they will be upgrading to 25.x. We expect many of the systems will stay on 13.0 for the remainder of their lives (typically 5 years) and hence we do expect to do any security updates needed.
Newer Enterprise customers are mostly on 24.10. They will begin switching to 25.04 next month.