General Page
IBM MQ uses the queue files in the <qmgrName>/queue/<queueName> directory to save the messages. The size of these files could be very large if there are many and/or large messages on the queue. The default queue size is 2 TB or 2,088,960 MB.
If APAR IT29354 is not applied, the queue files cannot be larger than 2TB and the queue is damaged when the queue file reaches 2TB.
IBM MQ 9.1.5 CD Release has introduced a new queue attribute, MAXFSIZE, to control the size of queue files, and two new queue status attributes, CURFSIZE, to the current size of the queue file in megabytes, and CURMAXFS, the current maximum size the queue file can grow to.
When you increase the MAXFSIZE, the granularity needs changing. Therefore, warning message is written to the AMQERR logs:
- AMQ7493W : A request to change the maximum queue file size for queue <insert_3> has been accepted. This requires that the queue file be reconfigured by the queue manager, which will occur when the queue reaches a queue depth of zero and there are no outstanding transactions referencing this queue.
This gives you an indication that you need to plan for the queue to be emptied, in order for IBM MQ to adopt the new granularity.
This new attributes are useful if you have small filesystem and do not want to allow the queue files grow beyond certain size.
When a queue has messages removed from it, the CURFSIZE attribute might not decrease immediately. Typically, unused space in a queue file is only released in the following circumstances:
- When no applications have the queue open
- After 1000 writes to the queue manager log, or
- At queue manager shutdown
Attention:
You should be cautious of increasing the size of queue files without considering the way that your applications are written and the possible effect on performance. Accessing messages at random in a very large queue file can be very slow.
If you are considering increasing the maximum size of a queue file beyond the default you should be cautious of using message selectors such as correlation IDs and IBM® MQ classes for JMS selector strings. Larger queue files are better suited to first-in first-out access to the queue.
Having very large amounts of data in individual queue files should only be done on queue managers that are configured for circular logging, or where media imaging has not been enabled for the individual queue.
You should not limit the size of SYSTEM queues as this could affect the operation of the queue.
Note:
Some file systems and operating systems have limits on the size of the entire file system, and the size of individual files. You should check the limits on the systems that your enterprise uses.
[{"Business Unit":{"code":"BU053","label":"Cloud & Data Platform"},"Product":{"code":"SSYHRD","label":"IBM MQ"},"ARM Category":[{"code":"a8m0z00000008ObAAI","label":"Components and Features->Administration"}],"ARM Case Number":"","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"9.1.0","Line of Business":{"code":"LOB45","label":"Automation"}}]
Was this topic helpful?
Document Information
Modified date:
29 April 2020
UID
ibm16202568