Customer feedback when initiate Ping, MX responds with "icmp reply" and "frag needed and DF set".
TRAF_alb-FEE_fe01_a1 root@PL-3:~# ping -M do -s 1472 -c 100 192.168.134.254 -I 192.168.134.1
PING 192.168.134.254 (192.168.134.254) from 192.168.134.1 : 1472(1500) bytes of data.
From 192.168.134.254 icmp_seq=1 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=1 ttl=64 time=0.540 ms
From 192.168.134.254 icmp_seq=2 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=2 ttl=64 time=4.87 ms
From 192.168.134.254 icmp_seq=3 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=3 ttl=64 time=0.762 ms
From 192.168.134.254 icmp_seq=4 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=4 ttl=64 time=0.652 ms
From 192.168.134.254 icmp_seq=5 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=5 ttl=64 time=0.606 ms
From 192.168.134.254 icmp_seq=6 Frag needed and DF set (mtu = 1514)
1480 bytes from 192.168.134.254: icmp_seq=6 ttl=64 time=1.13 ms
1480 bytes from 192.168.134.254: icmp_seq=7 ttl=64 time=1.43 ms
From 192.168.134.254 icmp_seq=7 Frag needed and DF set (mtu = 1514)
test1：
mx104-A ping mx104-B
xe-2/0/0 & xe-0/0/0 mtu=4000
xe-0/2/0 & xe-2/0/1 mtu=3000
labroot@belmont-mx104# run ping 192.168.0.3 logical-system A size 2500 do-not-fragment
PING 192.168.0.3 (192.168.0.3): 2500 data bytes
2508 bytes from 192.168.0.3: icmp_seq=0 ttl=63 time=1.420 ms
2508 bytes from 192.168.0.3: icmp_seq=1 ttl=63 time=1.318 ms
2508 bytes from 192.168.0.3: icmp_seq=2 ttl=63 time=4.178 ms
2508 bytes from 192.168.0.3: icmp_seq=3 ttl=63 time=1.365 ms
2508 bytes from 192.168.0.3: icmp_seq=4 ttl=63 time=1.373 ms
This is the expected behavior, MTU greater than 2500
test2：
labroot@belmont-mx104# run ping 192.168.0.3 logical-system A size 3500 do-not-fragment
PING 192.168.0.3 (192.168.0.3): 3500 data bytes
36 bytes from 12.1.1.2: frag needed and DF set (MTU 2986)
Vr HL TOS Len ID Flg off TTL Pro cks Src Dst
4 5 00 0dc8 1684 2 0000 40 01 4904 12.1.1.1 192.168.0.3
4 5 00 0dc8 16a3 2 0000 40 01 48e5 12.1.1.1 192.168.0.3
Due to 3500 is greater than MTU, so it will be discarded by MX.
Test3
Configure port mirror for xe-1/0/0.0 (mtu=1500), issue reproduced.
set forwarding-options analyzer test input ingress interface xe-0/0/0.0
set forwarding-options analyzer test output interface xe-1/0/0.0
xe-1/0/0 mtu=1500
36 bytes from 12.1.1.2: frag needed and DF set (MTU 1500)
4 5 00 09e0 92ba 2 0000 40 01 d0b5 12.1.1.1 192.168.0.3
2508 bytes from 192.168.0.3: icmp_seq=0 ttl=63 time=6.757 ms
4 5 00 09e0 92e5 2 0000 40 01 d08a 12.1.1.1 192.168.0.3
By the way, even without configuring ”do-not-fragment“, the situation remains the same for test3
labroot@belmont-mx104# run ping 192.168.0.3 logical-system A size 2500
4 5 00 09e0 b79f 0 0000 40 01 ebd0 12.1.1.1 192.168.0.3
2508 bytes from 192.168.0.3: icmp_seq=0 ttl=63 time=6.320 ms
4 5 00 09e0 b7c8 0 0000 40 01 eba7 12.1.1.1 192.168.0.3
If the MTU of the mirror interface is too small, the device does not fragment packets and sends the fragment packets to the collector.
The MTU is adjusted according to the size of packets sent and received by the device.