>>>>Hi Tamar,
>>>>
>>>>

>>>>		insert into ERRORLOG ;
>>>>              (DATE, TIME, USER, TERMINAL, CUSTOMER, ALIAS,;
>>>>               RECNUM, KEYFLD, ERRMSG1, ERRMSG2, ERRMSG3, ACTION, ;
>>>>				MEMVARS, DETAILS, CALLSTACK) ;
>>>>		   values ;
>>>>		      (date(), time(), gcUSER_NAME, SysZero(), space(len(customer)), lcALIAS, ;
>>>>               lnREC, lcKEY, lcLINE1, lcLINE2, lcLINE3, 'REBOOT',;
>>>>               '', m.lcDetails, m.lcStackInfo)
>>>>

>>>>
>>>>- Could SysZero() change alias() or insert a log row itself or otherwise interfere with the insert midstream?
>>>>
>>>
>>>Nope, SysZero reads one environmental variable (GETENV()) and cleans up and returns the result.
>>>
>>>>- and -
>>>>
>>>>Does space(len(customer)) rely on selected alias? If it's a variable, try with mdot or what happens if you replace with "" ?
>>>
>>>Nope, Customer here is a field in ErrorLog, so just inserting an empty value there. Guess I could try just the empty string. (Inherited code, and policy through the app is to use space(len(field)) for empty values, but certainly makes no difference here.
>>>
>>>The weird thing, of course, is that all this is working for everyone else, so it has to be something local.
>>>
>>>Tamar
>>
>>Does it fail every time, or is there something that a bit feels random? If the later I would look for the hardware. Some corrupted memory that will only be reached if you log. While use of memory looks like random, in many cases it is not. Anyway, scanning the memory might be not a bad idea.
>
>It's crashing whenever this error handler code runs. The only thing that's at all random is whether I get an error message first.
>
>I can certainly run some hardware checks, though.
>
>Thanks,
>Tamar

Sounds like the problem isn't the error handler itself, but the error that triggers it. Can you reproduce by calling the error handler directly?