Constraints on the use of the
Record Distribution to WebSphere MQ: one-way pattern
A number of constraints apply when you use this pattern.
You can use the
Record Distribution to WebSphere MQ: one-way pattern only if you meet all the following
constraints:
It must be possible to read the input file as records.
You can process files that contain fixed length records, variable length
records separated by a well-known delimiter, or variable length records that
can be extracted by a parser, a technique referred to as
Parsed Record Sequence.
This technique can be used to read a file
that contains multiple XML documents, returning each document as a record.
You cannot use this pattern to extract embedded records,
for example, records within a single XML document.
If a remote file is processed by the FTP service,
the broker must be permitted to delete the source file
because source files are moved to the broker and are not read in place.
The business function for which the pattern instance is used
must be compatible with the redistribution of records after a system failure.
This pattern distributes the records reliably,
but does not guard against duplicates after a system failure.
The business function, for which the pattern instance is used,
must be compatible with the redistribution of records after failure
to read a record from the file or to write a record that has been read.
In this case, this pattern moves the whole file to the backout directory and an error is written to the
error queue. You must arrange for the file to be resubmitted after the error has been corrected.
When a file is resubmitted, all the records in the file are reprocessed, which may lead to duplicate
messages.
If logging is required, the format that is used by this pattern is assumed
to meet the requirements of your own environment.