I will bet that you are running HPO 5. There is some sort of strange spool
problem that will cause files to be left in error status with the active bit
turned on.  This will cause DMSRDR (the RDR command) to fail.  I do not have a
solution for the problem but I do have a circumvention on my LISTSERV of a
modified RDR module that will skip files with the inuse bit turned on.
 
If this is indeed your environment, drop me a note directly and will be send
you the module or you can do it yourself by changing the CMS code for DMSRDR
where the check is made for files in hold or syshold by also adding the
equated name for the inuse bit (sorry I don't remember off hand what the
equated name is as it has been a while).  I also have an open incident with
the IBM Support Center about this and any information that people in this
environment could provide me about the occurance of this problem would be
appreciated.