TIBCO ActiveMatrix Adapter for Files (Unix/Win) fails to parse large files due to memory limitations and throws the following error: “AEADFILES_980001 MBinaryBuffer::init”.
book
Article ID: KB0094620
calendar_today
Updated On:
Products
Versions
TIBCO ActiveMatrix BusinessWorks Plug-in for Files for Unix and Windows
-
Not Applicable
-
Description
Description: TIBCO ActiveMatrix Adapter for Files (Unix/Win) (FA), fails to parse large files due to memory limitations and throws the error, “AEADFILES_980001 MBinaryBuffer::init”. This error is
normally seen due to insufficient memory allocation. Note that Out-of-memory is more of an OS limitation than an FA error.
Symptoms: With 'Record Transfer', on AIX, TIBCO ActiveMatrix® Adapter for Files (Unix/Win) (FA) could process a record up
to 2.8 MB in size and up to 3 records in the input file. When the record
size is more than 3 MB or there are more than 3 records in the file,
then the 'out-of-memory' errors occur. Solaris seems to do better than AIX when it comes to memory
management. On Solaris, FA could process up to 10
records in the file before the 'Out-of-memory' errors are reported. Cause: This error is normally encountered due to insufficient memory allocation. With 'Record Transfer', on AIX.
Issue/Introduction
TIBCO ActiveMatrix Adapter for Files (Unix/Win) fails to parse large files due to memory limitations and throws the following error: “AEADFILES_980001 MBinaryBuffer::init”.
Resolution
This error is normally encountered due to insufficient memory allocation. Check system resources limits on AIX using the ‘ulimit’ command. This command allows setting or getting limitations on systems resources available to the current shell and its descendents. Try to tune the following parameters for the shell which launches the TIBCO ActiveMatrix® Adapter for Files (Unix/Win). instance:
- Data segment size - Virtual memory size
With 'Record Transfer', if a record is 3 MB in size, FA needs to read the entire 3 MB into memory before it could process it. With 'Simple File Transfer', an FA publisher reads the input file in much smaller chunks (not records) and publishes it while the FA subscriber reassembles the chunks back to file. There is virtually no limit on the file size.