This problem in my environment is very specific. As I have several clusters each with 2 nodes running netbackup.
And sometimes it can failover, not cleanly or say the running node with active backups crashes. There may be reservations pending and the only way to clear the reservations is actually bring netbackup back up on that 'failed server'.
The crawlreleasebyname normally works but in this instance it will not be cleared.
This is however, can be based on how you have your shared drives configured for SCSI reservation.
Another thing to check is in case there is a stale .lock flag file for the device(s) in PEND-TLD status.
They are normally somewhere under the subdirectories ../volmgr/devices/..
But I would not touch these unless the whole netbackup environment is shutdown/stopped.