All,
I've started to see this message come up constantly on our core 3850 stack.
%SNMP-3-INPUT_QFULL_ERR: Packet dropped due to input queue full
Not sure how long its been going on for, but its a constant bombardment of messages, every second.
A show snmp shows the following:
Chassis: xxxx
Location: London
49966935 SNMP packets input
0 Bad SNMP version errors
8874 Unknown community name
0 Illegal operation for community name supplied
0 Encoding errors
242796799 Number of requested variables
6 Number of altered variables
24977944 Get-request PDUs
260668 Get-next PDUs
6 Set-request PDUs
10700543 Input queue packet drops (Maximum queue size 1000)
39256517 SNMP packets output
0 Too big errors (Maximum packet size 1500)
187407 No such name errors
0 Bad values errors
0 General errors
39256517 Response PDUs
0 Trap PDUs
SNMP global trap: disabled
SNMP agent enabled
We have a single poller for Solarwinds, and SNMP is setup on an RO string with a small access-list applied.
I've run a snmp debug as well, which shows the following:
Feb 16 08:49:03.473: SNMP: Packet received via UDP from 172.xx.xx.xx on Vlan10
Feb 16 08:49:03: %SNMP-3-INPUT_QFULL_ERR: Packet dropped due to input queue full
It shows a bunch of other stuff, but this is the only bit that seems to coincide exactly with the queue full errors. Weirdly the packet received messages are coming from our poller, which is allowed via the ACL. There are other debug messages that seem to show communication working, but interspersed with those messages are the above.
I've checked Solarwinds, and its set to poll at 120 second intervals, collect stats every 10 mins and topology every 30 mins.
I dont know why the poller is trying to send snmp packets to it every second?
Any ideas?
Cheers
Greg