Why can a (icmp, snmp) host with freebsd be reachable and not reachable with linux?

Question

linux

The situation is this: there is a Chinese piece of iron (optical amplifier) ​​with an ethernet port and the ability to assign an ip address to it, as well as receive status information via snmp.
There are 2 PCs, one of which is on freebsd, the other on linux.
All 3 have ip addresses from the same network.
The problem is that it responds to requests from freebsd (ie ping and snmpget) but not to requests from linux.
At the same time, on the interface looking at the hardware (tcpdump), no answers are visible, except for arp, i.e. poppy address, I can see it.
During various tests, I connected a PC directly to linux, bypassing any equipment, and the situation was identical.
Maybe the matter is in some parameter of the linux kernel of the machine?

Answer 1

I'll duplicate it here:
iptables is empty.
There are two Linux machines and both see freebsd. Calculate and gentoo. But thus stably do not see a piece of iron. what does the poppy address in arp tables have for both linux and freebsd and the same.
Linux machines, like frya, see other hosts in the same vlan and in the same subnet.
-------------------------------
I found the cause of the problems. By default, Lunix sets the DF flag on the packet if there is no special route. The piece of iron cannot (apparently problems with religion) process and answer packets with such flag. FreeBSD doesn't set any flags.
------
How to remove the DF flag from packets I found here
In short: you need to add the "mtu lock" option to the route to the network

Answer 2

perhaps the reason is iptables? lay out the table

Answer 3

and if fribsd is accessed from linux, is the situation the same? what distribution?