Add SNMPv3 Traps on a Publisher

This article explains how to configure snmpd to send v3 traps to your Network Monitoring System.

Edit /etc/snmp/snmpd.conf on the Publisher and add this block (use the same creds you created with net-snmp-config):

######################################################################## 
# SNMPv3 traps (send to your NMS) 
######################################################################## 
# Enable trap generation from snmpd 
authtrapenable 1 
# Where to send traps (replace <NMS_IP>) 
# Use the same v3 user, auth/priv protocols & passphrases 
trapsess -v 3 -l authPriv -u snmpv3user \
   -a SHA -A <AuthPassword> -x AES -X <PrivPassword> \
   <NMS_IP>:162

If you have multiple NMS receivers, add multiple trapsess lines. Define what should trigger traps.

You already have:

# (from your config) 
disk / 2000 
extend-sh conntrackCount cat /proc/sys/net/netfilter/nf_conntrack_count

Add the monitor/threshold directives below. These poll every 60 seconds and emit a trap as soon as the condition is true.

######################################################################## 
# Threshold-based monitors that raise traps ########################################################################
 
# 2.1 Disk on "/" lower than the 2GB minimum (kB) set by `disk / 2000` 
#      -> trap when the flag flips non-zero 
defaultMonitors yes 
monitor -r 60 -o UCD-SNMP-MIB::dskErrorFlag \ 
  "Low disk space on root filesystem" != 0 

# 2.2 High CPU: trigger when idle% < 20 (i.e., used > 80) 
#     UCD-SNMP-MIB::ssCpuIdle == .1.3.6.1.4.1.2021.11.11.0 
monitor -r 60 -o .1.3.6.1.4.1.2021.11.11.0 \ 
  "High CPU usage (>80%)" < 20 

# 2.3 High Memory: compute % used with a tiny `extend` and alert > 80% 
#     Create an extend that prints an integer 0..100 to stdout 
extend-sh memUsedPct awk '/MemTotal|MemAvailable/ {v[$1]=$2} \ 
  END{printf("%.0f\n", 100-(v["MemAvailable:"]/v["MemTotal:"]*100))}' /proc/meminfo 

#     Monitor that extends value (> 80) 
monitor -r 60 \ 
  -o NET-SNMP-EXTEND-MIB::nsExtendOutput1Line."memUsedPct" \ 
  "High memory usage (>80%)" > 80

Notes

• defaultMonitors yes enables a few built-ins; the explicit monitor on dskErrorFlag makes the disk alert unambiguous.
• CPU rule watches idle% (simple and robust).
• The memory rule uses an extend-sh one-liner to compute Used% from /proc/meminfo, then monitors that value.

Restart the agent to apply all changes:

sudo systemctl restart snmpd 
sudo systemctl status snmpd --no-pager