Long list of items in Queue Details

[ptryks]

Hey, guys


I'm implementing a monitoring environment with Zabbix.

But strange behavior is happening. All items listed in the queue.details are UP. However, some items stay in the queue for 5 hours.

How could I debug this problem?
What SQL queries and commands could help identify the root cause of this issue?

[Markku]

The most common reason for the queue to increase is: the Zabbix agent is not running on the host, or the host (agent) is not reachable by the server/proxy (or vice versa).

If the items are passive items: see zabbix_proxy.log (I assume the host is monitored by proxy)

If the items are active items: see zabbix_agent.log

Markku

[ptryks]

The most common reason for the queue to increase is: the Zabbix agent is not running on the host, or the host (agent) is not reachable by the server/proxy (or vice versa).

If the items are passive items: see zabbix_proxy.log (I assume the host is monitored by proxy)

If the items are active items: see zabbix_agent.log

Markku

Firstable thanks for the answer @Markku

On zabbix_proxy.log, there are some entries, like these bellow

resuming SNMP agent checks on host "xxxxxx-xxxxxx": connection restored
SNMP agent item "net.if.status[ifOperStatus.9999]" on host "xxxxxx-xxxxxx" failed:


I find it strange because SNMP is stateless. It doesn't need to keep the connection stateful to perform the queries.

What could be causing this type of error? Because pinging the host has no packet loss.

Are there some tips to debug this issue?

[Markku]

If Zabbix sees SNMP not working (like in timeouts), it logs the state change. Likewise when SNMP starts working again.

Run tcpdump for long enough to capture the packets on the proxy (like sudo tcpdump host ip-of-host -w capture.pcap -v) and use Wireshark to analyze what happens.

Markku