|
We are using MQ Series V5.3 on our Solaris system. The application requires persistence of messages exchanged between Solaris system and the backend Mainframe application.
Initially the Solaris program was designed to commit every message received. There was no problem, except for some minor performance issues.. To improve the performance and cater the need to process higher volumes - we modified the designed issue COMMIT with SYNCPOINT on all messages, for every 20 messages or within 5 sec. interval (which ever is earlier) – with the BEGIN SYNCPOINT ( to define a unit of work).
We have been facing the problem with errors AMQ7469 periodically.
The average size of application data the message is between 500 – 750 bytes. Recently we modified the qm.ini configuration to the following –
LogPrimaryFiles = 7 LogSecondaryFiles = 4 LogFilePages= 1024 Logtype = LINEAR LogBufferPages = 17
The default BATCHSZ of 50 is used for the sender/receiver channels.
The size of each Log file is about 4MB. We have enough storage spaces on our file system where /var/mqm resides. The average volume of data exchanged between solaris and mainframe – when the error occurs could be about 3000 messages.
Could some one explain, how to overcome this problem
Thanks lakshmi
|
