We've also still had a need to go back to the manual method since moving to 7.3.1.
Viewdbchk is still the first port of call, but there have been times where it has sent things absolutely crazy. For example, I deleted a pool which got stuck on the last VM, so ran the script as normal. Instead of clearing the pool, all the previously recreated desktops were suddenly recreated and went through the reprovisioining process. Re-running the script once the provisioning had finished had no impact. I couldn't manage the pool as it was still stuck in deleting. I had to manually delete around 100 desktops from vsphere whilst composer was disabled, then tidy up the db, and then run the viewdbchk script again.
Before 7.3 it was the only thing we needed to tidy up errors. Now, we're a little more cautious for reasons such as the above.
