Red Hat Bugzilla – Bug 3219
High network traffic causes snmpd to seg fault
Last modified: 2008-05-01 11:37:50 EDT
When this is read by snmpd to report stats for mrtg, snmpd will die. This is install of RedHat 6.0, with addition of mrtg, MySQL, and some perl libraries all in RPM format. This makes it hard to use SNMP to monitor the machines. =====----- BEGIN cat /proc/net/dev -----===== [jrothste@www jrothste]$ cat /proc/net/dev Inter-| Receive | Transmit face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed lo: 5272882 39868 0 0 0 0 0 0 5272882 39868 0 0 0 0 0 0 eth0:2235709667 11573352 1 0 0 0 0 0 3739544558 11742385 0 0 0 210896 0 0 eth1:1624327485 4789640 0 0 0 0 0 0 1072228659 3884288 0 0 0 147391 83 0 [jrothste@www jrothste]$ =====----- END cat /proc/net/dev -----=====
I'm changing this component to ucd-snmp because that's where the fix will be. The easiest workaround is to rebooot the machine in order to reset the counters that are causing the snmpd segfault. There may also be a more precise way (than renbooting) to reset the ethernet counters. ------- Email Received From Jason Rothstein <fdragon@io.com> 06/07/99 12:24 -------
Fixed in the ucd-snmp 3.6.2 release. The snmp daemon now rolls over rather than core dumps when the size of the traffic counter exceeds 32 bits. MRTG has no problem with that happening, though its counts may be off for the 5 minutes where the rollover happened. I have 3.6.2 RPM's at ftp://ftp.estinc.com/pub/unsupported and will upload them to a Red Hat incoming site as soon as someone tells me how. -- Eric Lee Green, Enhanced Software Technologies Inc. eric@estinc.com
Fixed (by upgrading) in ucd-snmp-3.6.2-1. Thanks for the report.