I updated from 11.3 U5 to 12.0 U1 today and had no issues migrating my jails. Namely Plex and Nextcloud.
I do however have an issue with iscsi on my vmware esxi 7.0 hosts. ISCSI access is intermittent and will not allow powering up of any vm's. No issues with windows ISCSI connections as far as I can tell. Before I start digging I was wondering if there is something obvious or known that I could have missed?
Hi! Well, I don't know where you got 12.0-U1, but I installed 12.0-RELEASE because I am not so advanced.
However, exact same thing here, only with XCP-ng. It turned out that somehow the password for CHAP changed to something far longer. I really don't remember if and when I changed it, but it seems unlikely, otherwise I would notice the iscsi sr missing. The damn thing did it

!!! I am joking, it was really easy to fix. Actually I peeked /etc/ctl.conf and took the new (?) password from there.
The only problem here is that it feels like I am being hacked. But the iscsi target was really without any importance: virtual machines i barely ever use.
As for the transition, it was very very very very very smooth. Did I say "very"?
Ok, I immediately bumped into a bug where you stop a virtual machine and it reports as active whereas it is definitely down. It's already reported and it seems it's going to be fixed in the next release. No big deal for me, as I hardly ever need to turn the virtual machines off. But you have to reboot, which might be bad for someone. I didn't try to restart the middleware, because it only happened once.
So, what I use:
- All the settings are: Ok.
- Tunables: OK. It obeys to arc_min and arc_max now. WOW, not a big deal, but nice.
- Virtual machines: OK.
Huge performance boost in Windows VMs, by the way. NOW we are talking, bhyve.
- 11.3 Jails: plex, syncthing, nextcloud, mariadb, xen-orchestra from sources, openvpn. Everything Ok and working as expected.
- SMB shares: OK
- Active directory join: OK.
- I immediately noticed that L2ARC is hitting harder, for whatever reason?
- It seems to me the log device is working better: at this point I think I am hallucinating...
What the heck is that ".bhyve_containers" dataset that appeared in the system pool? I don't know. Somebody?
Still I haven't upgraded the pools. I plan to do that 2 releases from now to finally get ZSTD compression and persistent l2cache. No need to rush really.
Best upgrade in years so far, possibly EVER.