1. Locate and check the size of the current ERRLOG.SYS file:

$ SET DEFAULT SYS$ERRORLOG
$ DIRECTORY /SIZE /DATE /PROTECTION ERRLOG*.*

Directory SYS$SYSROOT:[SYSERR]

ERRLOG.OLD;15           4374  22-DEC-2017 11:26:05.97  (RWED,RWED,RE,)
ERRLOG.OLD;14             97   8-DEC-2017 11:26:04.89  (RWED,RWED,RE,)
ERRLOG.OLD;13            219  17-MAY-2010 16:42:34.30  (RWED,RWED,RE,)
ERRLOG.SYS;1             282   9-NOV-2018 09:50:42.14  (RWED,RWED,RE,)

Total of 4 files, 4972 blocks.

2. Create a new version of the ERRLOG.SYS logfile simply by RENAME-ing the ERRLOG.SYS file – Don't worry, VMS will notice and create a new version of ERRLOG.SYS the very next time it needs to write-out a buffer of errlog information:

$ RENAME /LOG ERRLOG.SYS *.OLD
%RENAME-I-RENAMED, SYS$SYSROOT:[SYSERR]ERRLOG.SYS;1 renamed to SYS$SYSROOT:[SYSERR]ERRLOG.OLD;16

3. Purge old file versions to free up space – keep as many recent versions as you like, but consider whether you'll ever actually need to investigate events in those older files (it hardly ever happens, but again, YMMV).

$ PURGE /KEEP=2 ERRLOG.OLD

The practical, forensic or diagnostic utility of any version of ERRLOG.OLD drops rapidly beyond the top-version of that file, so don't keep more than a couple of these file-versions.

Done.

Different systems and business environments have different operational requirements, but here are a few rules of thumb, and your own specific requirement is likely a combination of two or more of these:

• Whenever the System Error Logfile size gets larger than “a few hundred-KB (kilobytes).”
• Occasionally when trying to diagnose an intermittent or creeping hardware failure.
• Monthly.
• Quarterly.
• Maybe even annually.
• With every system reboot (not suitable for systems which run for months or years without rebooting); often, the ERRLOG.SYS rename/rollover command is included in the system's SYS$MANAGER:SYSHUTDWN.COM script, so this file gets rolled-over before each reboot.