I'm trying to wrap my head around this.
I've currently got several queue managers supporting an application that is using MQ as a buffer for logging application events.   They log at a high rate, but we have a lot of disk set up for these.
But when the app has a problem and isn't consuming messages they can build up fairly quickly to 10 or 20 or 50gb of data in the queues.   /var/mqm/log is 2x as big as /var/mqm, so we have room for a full media image, probably close to two of them.

But now that automatic log management is there it seems great, I don't need the script that does the record media image and deletes the old log files.  

But how to use it?   update the qm.ini for LogManagement=Automatic and then set the qmgr properties like this?

IMGINTVL(10) +
IMGLOGLN(OFF) +
IMGRCOVO(YES) +
IMGRCOVQ(YES) +
IMGSCHED(AUTO) +

Then I don't need to delete logs anymore?

What about changing log primary and log secondary?   Should those counts and sizes add up to more or less than the amount of disk space I have for logs?  

Has anyone gone through this process and come up with a plan?  Any thoughts?

Thanks

Tom

------------------------------
Tom Kane
Senior Technical Team Lead
AT&T
Kansas City
------------------------------

Hi Tom,

There are a couple of docs pages/blogs that might be worth having a read of. Firstly Mark's blog on the automatic log management feature is a great overview of how to set up automatic log management:

https://www.ibm.com/support/pages/logger-enhancements-mq-v902-and-v91

Then regarding sizing the primaries and secondaries vs the file system size it's worth taking a look at this topic in the IBM Docs:

https://www.ibm.com/docs/en/ibm-mq/9.2?topic=lost-calculating-size-log

It might be worth setting IMGLOGLN as well as IMGINTVL so that queue manager will take media images once the amount of written log data exceeds a given amount. This may allow you to set a larger IMGINTVL than  10 minutes since if you write more than IMGLOGLN in that time a media image will be taken anyway.

------------------------------
Matthew Whitehead
------------------------------

Original Message:
Sent: Sun January 30, 2022 07:55 AM
From: Tom Kane
Subject: converting from linear logging to linear logging with automatic log management

I'm trying to wrap my head around this.
I've currently got several queue managers supporting an application that is using MQ as a buffer for logging application events.   They log at a high rate, but we have a lot of disk set up for these.
But when the app has a problem and isn't consuming messages they can build up fairly quickly to 10 or 20 or 50gb of data in the queues.   /var/mqm/log is 2x as big as /var/mqm, so we have room for a full media image, probably close to two of them.

But now that automatic log management is there it seems great, I don't need the script that does the record media image and deletes the old log files.  

But how to use it?   update the qm.ini for LogManagement=Automatic and then set the qmgr properties like this?

IMGINTVL(10) +
IMGLOGLN(OFF) +
IMGRCOVO(YES) +
IMGRCOVQ(YES) +
IMGSCHED(AUTO) +

Then I don't need to delete logs anymore?

What about changing log primary and log secondary?   Should those counts and sizes add up to more or less than the amount of disk space I have for logs?  

Has anyone gone through this process and come up with a plan?  Any thoughts?

Thanks

Tom

------------------------------
Tom Kane
Senior Technical Team Lead
AT&T
Kansas City
------------------------------