We have the omnimark script that takes 2gb sgml file size as input and output the file which is around 2.2 gb.The script is called from unix shell script and we are facing issues that sometimes script runs successfully and sometime it just aborted with no error....any idea or suggestions how to debug this?
I have seen this type of issue before in running OmniMark v5.3 when the script bombs due to lack of server resources/memory.
If you've specified writing to a log file, e.g. using -log logfilename.txt, then you would see something like an error code #3000 "insufficient memory error".
http://developers.omnimark.com/docs/html/error/3000.htm
If no log file, then initial step would be to run the script in a console session so that any such abort message is visible.
Stilo have a page listing fixes in various versions of OmniMark
http://developers.omnimark.com/docs/html/concept/806.htm
This mentions a variety of memory-related issues in various versions of the software (e.g. use of certain translate rules) which may help some investigation.
Alternatively, you could add to the script writing to a debug log file (with a global switch to activate debug on or off (so you don't waste further I/O resources when you don't need to)). Debug log file should be unbuffered. At certain breakpoints in the script add a message. The more verbose the better at narrowing down where/when the error is, but with the size of file I suggest it's a I/O or memory error.
Also depends what version of OmniMark you're using.