OES11 SP2 server spawns a "slurpfile() read() buffer overflow on file /proc/stat" message every 20 seconds.

  • 7015651
  • 11-Sep-2014
  • 11-Sep-2014

Environment

Novell Open Enterprise Server 11 (OES 11) Linux Support Pack 2

Situation

OES11 SP2 servers running Ganglia to collect server metrics, at one point in time start flooding the /var/log/messages file with the following message (one such message in every 20 seconds).

Sep 10 20:51:46 lxsrv01 /opt/novell/ganglia/monitor/sbin/gmond[5901]: slurpfile() read() buffer overflow on file /proc/stat
Sep 10 20:52:06 lxsrv01 /opt/novell/ganglia/monitor/sbin/gmond[5901]: slurpfile() read() buffer overflow on file /proc/stat
Sep 10 20:52:26 lxsrv01 /opt/novell/ganglia/monitor/sbin/gmond[5901]: slurpfile() read() buffer overflow on file /proc/stat
Sep 10 20:52:46 lxsrv01 /opt/novell/ganglia/monitor/sbin/gmond[5901]: slurpfile() read() buffer overflow on file /proc/stat


No dis-functional impact on the server was observed, other than the continuous flood of these apparently harmless messages that are being logged.

As a work-around, restarting the /opt/novell/ganglia/monitor/sbin/gmond deamon temporarily resolved the problem.

Resolution

This was resolved with the public release of the August 2014 OES11 SP2 Scheduled Maintenance update.

Cause

This was a buffer size problem in the 3.2 Ganglia code which was forked for OES.
See Ganglia bug #298