Logs about an incident on a Pacemaker Cluster are lost because of the log file turn-over policy

The following options specify how many pe* files should be kept:

        pe-error-series-max=
        pe-warn-series-max=
        pe-input-series-max=

They are to be added to the "property cib-bootstrap-options:" section of the Cluster Information Base (cib) using the "crm configure edit" command.

Allowed are integer values, the value "-1" will store files unlimited and probably create an out of disk space condition at some point. When looking for values, please monitor how many files are written per day and calculate how many files should be kept for how long. This is an example from a test system:

hana01:/var/lib/pacemaker/pengine # ls -l * | grep "Sep 30"| wc -l
3511
 

This is the total value of files created on a given day, including input, warn and error files. 
The majority of the files created will be state changes (pe-input*bz2), if the goal is to keep files for e.g. two weeks, the following values might be considered:

        pe-error-series-max="-1"
        pe-warn-series-max="5000"
        pe-input-series-max="50000"

Once the maximum number has been reached, pacemaker will start to overwrite existing log files. If /var/lib/pacemaker/pengine is backed up every day, the numbers might be adjusted in a way, that the backup contains the most recent changes only. 

For more information see: 
SLE 12 SP5 based pacemaker deployment:
https://clusterlabs.org/pacemaker/doc/deprecated/en-US/Pacemaker/1.1/html/Pacemaker_Explained/_available_cluster_options.html

SLE 15 based pacemaker deployment:
https://clusterlabs.org/pacemaker/doc/2.1/Pacemaker_Explained/html/options.html#cluster-options
man 7 pacemaker-schedulerd