Supposed you need to monitor several status values by SNMP on a device (traffic in 5 interfaces). If the device is operating on, this will take some time, let´s say 100ms for each value, multiplied by 10 (5 traffic traffic up + 5 down) would take 1 second.
Now suppose that the device is turned off and the timeout is 1 second. In this case will take as long as the timeout takes to return, this multiplied by the 10 samples would be like 10 seconds. This means that Zabbix is going to be busy for 10 seconds with a device that will not respond because it can´t. This I believe can significantly increase the number of inquiries into "Queue".
The question is, can I do a simple check, like a ping from a device and, if this is successful, continue with the rest of the SNMP query?
Now suppose that the device is turned off and the timeout is 1 second. In this case will take as long as the timeout takes to return, this multiplied by the 10 samples would be like 10 seconds. This means that Zabbix is going to be busy for 10 seconds with a device that will not respond because it can´t. This I believe can significantly increase the number of inquiries into "Queue".
The question is, can I do a simple check, like a ping from a device and, if this is successful, continue with the rest of the SNMP query?
Comment