>>>since 2-3 Years sometimes some Users get c00005 Errors in the Line SET SYSFORMAT ON.
>>>
>>>Any suggestions what this could be?
>>
>>I think the first thing I'd look is whether the users with problems have something in common in their system settings that SET SYSFORMAT ON is setting things to.
>
>Ok, i check this.
>
>Bud if there will be Settings which cause the Problem: why the c5 errors will happen only once in a month and not always?
>
>c5errors are mysterious

Reminds me of a "phase of the moon" problem that cropped up on some software on a few Unix systems, which would mysteriously crash on the evening of a full moon. It turned out there was a bug somewhere, which coincidentally happened to have periodicity that matched the moon and conditions for crash would coincidentally occur on a full moon.
BTW there are some programs that *do* have code that intentionally behaves differently on the night of a full moon -- the game rogue (a "dungeon crawler" type game) and other games like it have code that computes what the current phase of the moon should be, and boosts your luck when you play the game the evening of a full moon.

Speaking of coincidences... Back in the 1980s the Computer Science department at UCLA operated a network named "OlympusNet" with the machines (DEC VAX) having themed names -- being named after characters from Greek and Roman mythology. Occasionally there were some "quirks" that matched their naming -- such as Hera sometimes not wanting to cooperate (and occasionally obstructing) Hercules. At some point some IBM 4381 were added to the network -- and given their different nature, were given names based on Norse mythology (in keeping with gods/mythology theme, but being distinctly different from Greek/Roman names given to the DEC VAX computers). Unsurprisingly there were some problems with integration but that got largely worked out. There was one machine that was particularly troublesome -- the one named Loki.