We faced an issue (bug description in MoS note 2407835.1) with having a bigfile tablespace at maximum size (8k blocksize = 32TB).
When backing up such a file rman seems to forget the related backupsets. When you list backupsets from within rman cli there’s no backupset anymore.
The good message is that the backupsets are still existing in the backup destination (ZFS share mounted via nfs in our case).
When you try to recatalog these files as described in the workaround in the MoS note mentioned above using catalog start with ‚<pathname>‘ you will see that they are still not existing. Even stated in Oracle’s documentation ‚catalog start with …‘ will not execute ‚catalog backuppiece‘ in the backupground.
Using ‚catalog start with …‘ will not create the missing metadata again, while using ‚catalog backuppiece …‘ will do … but for all the missing files we had (~200) it takes several hours.
To sum it up: it’s a quite strange bug but if you know about that it’s quite easy to fix.