Hi All,
I'm testing ping command on two servers.
[say A-->B and C--->D]
A,B,C,D are servers
From A-->B:
ping -c 3 -p ff -s30720 <B IP-Addres>
This works fine.
From C-->D:
ping -c 3 -p ff -s30720 <D IP-Addres>
This is NOT working and 100% packe loss 
But if I use 8700 instead of 30720 for packet size '-s', then it works.
Now, Kindly let me know which variable is restricting this packet size '-s' and how to increase it in AIX server??
Thanks in Advance.
How do you know it is not a network restriction instead of an AIX setting?
What does the ethernet adapter and interface attribute settings on all four systems show?
It looks to be a packet size issue so check on each adapter and interface on each system:
lsattr -El enx
lsattr -El entx
Where enx/entx it the adapter/interface, like en0, en1, en2 or ent0, ent1, ent2, Etc.
Look at the mtu packet size, jumbo frames, large send, Etc.
HTH
In all the four servers, the following values are same
tx_que_sz 8192
txdesc_que_sz 512
remmtu 576
mtu 1500
jumbo_frames no
large_send yes
Only variable change in busintr
In A:
ent0
busintr 8736
ent2
busintr 581
In B:
ent 0
busintr 65657
In C:
ent0
busintr 279
In D:
ent0
busintr 131193
Is it something to do with busintr???
Do the packets traverse the same physical network ? Perhaps the second one has a router on the network that is throwing away the large icmps?
No in different physical network:
A & B in different segment
say for ex
A--> 10.172.10.X ; B--> 10.172.30.X
But C & D are in same segment:
C--> 10.173.10.X; D--> 10.173.10.X
Sounds like a network problem then.
The busintr shouldn't make any difference.
I would suspect the problem network is not enabled for large send.
Might be worth using "traceroute" to get a list of routers between the two destinations and then trying the ping of the large packet. TBH, from the articles Ive read it seems that the large ICMP packets (over 1492) are handled in vastly different ways by differing network equipment, so I wouldnt expect consistancy.