Zabbix Proxy only polls SNMP item once, then randomly every few hours

[DmitryMigunov]

Hi everyone,

We’re experiencing a strange issue with Zabbix Proxy and SNMP items. For some reason, the proxy only polls the SNMP device once, right after the proxy is restarted. After that, it stops polling regularly and instead retrieves data at random intervals, sometimes only once every 3–4 hours.

The item interval is correctly set to 1m, and we’ve double-checked all relevant configuration settings. We’ve also tested this behavior with both our own Zabbix server and Zabbix Cloud, using both passive and active proxies — the result is the same.
We’ve verified that the SNMP device itself is responsive — using the snmpget command manually returns results instantly, with no delay.
As soon as we restart the proxy, it immediately polls the SNMP item once, and then it becomes idle again for hours.

Proxy configuration:

Server=XXX.zabbix.cloud:10051
Hostname=proxmox-1
LogFile=/var/log/zabbix/zabbix_proxy.log
LogFileSize=0
DebugLevel=4
PidFile=/run/zabbix/zabbix_proxy.pid
SocketDir=/run/zabbix
DBName=zabbix_proxy
DBUser=zabbix
DBPassword=password
ProxyBufferMode=hybrid
ProxyMemoryBufferSize=16M
StartSNMPPollers=100
Timeout=4
FpingLocation=/usr/bin/fping
Fping6Location=/usr/bin/fping6
LogSlowQueries=3000
TLSConnect=psk
TLSAccept=psk
TLSPSKIdentity=proxmox
TLSPSKFile=/psk

Has anyone encountered a similar issue?

[tomse]

Yes.. I've got that problem as well..
Running 7LTS ubuntu/postgresql and in docker, proxy is 7lts in alpine docker with mysql. using the official docker compose configs.
Some OID's are polled regularly from the same host, but others take hours.
and as you said.. random intervals even though poll is set to 50sec (default for one of my triggers).

would love to hear a solution to this

[markosa]

So, are you using preprocessing and it's feature "discard unchanged / discard unchanged with heartbeat" with those items? That might be in use at template you are using... With tcpdump you may check that if proxy/server is actually doing requests or not, command like : sudo tcpdump -i eth0 port 161 should show all traffic going to port 161(snmp listening port), replace eth0 with your interface name.