Hello,
I recently encountered an issue with snmp queries intermittently failing when polling the reth interfaces of an SRX210 cluster. We're using firmware version: [11.4R1.6]
It's worked for years without any problems and just started behaving this way. We use cacti to monitor and build graphs around traffic usage. Sometimes these graphs will populate for 12 hours normally, then they will be blank for several hours. The logs in our cacti system show that the the SNMP target is timing out, but I'm not seeing any alarms on our SRX210 cluster.
I've tried restarting the snmp service and it doesn't seem to make a difference. We've also rebooted each SRX individually using "request system restart node 0". We can't reboot them both at the same time due to this site being in live production.
I found this information relating to a similar issue with the MX series routers, but I'm not sure if this is a similar bug:
PR#
NUMBER 566681
SEVERITY major
CATEGORY sw
STATE closed
SYNOPSIS show interface statistics stuck, show pfe statistics traffic returns an error
COMMITTED-IN RELEASE -
ARRIVAL DATE 2010-11-17 13:47:29
LAST MODIFIED 2011-01-27 13:51:26
CLOSE DATE 2010-12-21 10:31:39
RELEASE NOTE On MX routers running Junos 10.2 and above, under certain conditions it is possible for command 'show interface statistics' to get stuck (i.e. not return anything and hung); 'show pfe statistics traffic' to return "error: the mib-process subsystem is not responding to management requests" and SNMP queries failing. If this issue is encountered, a restart of the pfed process recovers from the error. Restart PFED via root shell ps -A | grep pfed | egrep -v grep | awk '{print $1}' | xargs kill Only if running 10.2S4, where in pfed tracing got enabled by mistake, execute the below to disable pfed tracing. ps -A | grep pfed | egrep -v grep | awk '{print $1}' | xargs kill -INFO Note the kill -INFO results in a toggle action - if pfed tracing is OFF, it is truned ON and if it is enabled it is turned OFF.