How To Rollover VMS Logfiles
There are four major VMS logfiles that should be rolled-over, or refreshed, on a periodic basis, from both a disk space management perspective and from an ease of use (for the system administrator) perspective. These files are (shown in their “standard” or expected directory, although each could be relocated for specific purposes, usually in a VMScluster configuration):
- SYS$MANAGER:ACCOUNTNG.DAT (yes, really spelled thus, without the “I” in “…TNG”)
On contemporary multi-GB mass-storage media (disks, SSD, etc.), the argument for “saving disk space” – common in the early days of VMS with relatively small disks – is pretty much obsolete.
However, each of these VMS logfiles is sequentially appended with more data as the system generates it, and thus each of these files can, over time, become quite large, and with large size comes appreciable delays (time and computing effort) in reporting relevant data from these.
For each of these files, a file size of over just a few hundred megabytes can be considered “getting too large” – your mileage may vary (YMMV), of course, but this is a good rule of thumb. Much of the time, any reporting that you (the sys-admin) may want to do will be for “recent events,” which is data which has been appended to the end or tail of the file.
The VMS reporting commands (
TYPE /PAGE for
ANALYZE /ACCOUNTING for
ANALYZE /AUDIT for
ANALYZE /ERROR /EVL for
ERRLOG.SYS) read through the logfiles sequentially, start to end, and a long logfile simply takes longer to read through.
This, more than anything else, is the reason for rolling-over each of these logfiles. When a particular logfile is rolled-over, a new version of the logfile is created, so it is worthwhile to purge older versions of that logfile to free up disk space.
How To Do Each Rollover
The DCL command for rolling-over each of these major VMS logfiles is linked below, together with a brief discussion of good rollover periods and strategies: