Syslog overrun hack

Posted on 2013-02-05 by Ed Braaten

One of our prototype Linux systems was writing warnings about PCIe hardware correctable errors non-stop to /var/log/messages. The otherwise well-functioning system would run out of disk space before completing longer test runs due to the flood of warning messages in /var/log/messages. The root-cause of the messages (bad hardware/BIOS) will get addressed eventually but in the meantime the following quick hack worked to suppress the bogus warnings, saving disk space.

Change the default “*.info” to “*.err” in the /etc/rsyslog.conf file. This sends all messages with priority “err” or higher to the /var/log/messages file. The actually priority list in increasing priority is: debug, info, notice, warning, err, crit, alert, and emerg.
Restart the rsyslog daemon with “service rsyslog restart”.

This hack worked fine on Red Hat Enterprise (RHEL) 6.3.