icmppingloss only with zabbix?

[Daniel4711]

Hi everyone.

i have a question about the icmploss item from the hp enterprise switch template.
Currently, we're implementing a Network Segmentation Firewall.
One Management-Interface from an HP Switch has been switched to the new Firewall, so the icmp traffic to this switch has to go over the firewall.
After that, the average icmp response time has raised from 0.44ms to 0.63ms, what is still a good value I think.

But after that, I often got the icmp loss warning (308 times in 8 days after the change) from the hp enterprise switch template.
The average loss value is 21.7%.

I have noted this to the project manager, and we have tested it with ping (windows) and pinginfoview (windows).
Ping and Pinginfoview have reported 0 loss in a 12h test in the standard settings (Timeout - 1second).
From the commandline of the zabbix-server with the command "fping 10.21.1.11 -c 100 -t 1000", I got loss after a few seconds.
xmt/rcv/%loss = 100/96/4%, min/avg/max = 0.425/1.05/45.1

The zabbix documentation says, that the default values are:
3 packets
1 second delay between the packets
1 second timeout

Windows Ping, Pinginfoview an fping (version 5.1 on Debian 12) has identical settings, but only fping throws ping loss.

Does someone have an idea about this?

[Markku]

Hi, yes, 21% measured loss is huge. The only changed thing in the connection was adding the firewall, right? What would then be the natural point of investigation?

Markku

[Daniel4711]

Hi Markku,
thanks for your reply.
I also noted this immediately.
But since the Windows Ping and Pinginfoview (perhaps just a GUI add-on for Ping) do not report any losses with the same settings, I am in need of an explanation here.

The "error" can be narrowed down to Fping / the Zabbix server.

I will continue to monitor the issue and talk to the technician.
He should be able to see the corresponding packets on the firewall