Setting the following ndd parameter may help:
# ndd -h ip_ire_redirect_interval
ip_ire_redirect_interval:
All routing table entries resulting from ICMP "Redirect"
messages are deleted after this much time has elapsed,
whether or not the entry has been recently used. Setting
ip_ire_redirect_interval to 0 causes
- all existing routing table entries specifying
a redirect to be deleted after the previous value of
ip_ire_redirect_interval has elapsed and
- no further routing entries are created for the future
ICMP REDIRECT messages.
Processing of ICMP REDIRECT messages can be enabled again
by setting ip_ire_redirect_interval to a non-zero, positive
value.
[0, -] Default: 300000 (5 minutes)
Basically, such entries will be created when the system receives ICMP redirect message from the default gateway. If the default gateway detects better path to the remote address, it will send ICMP redirect message to tell the system to use alternative GW. In this case, the default gateway thought
an alternative GW at 192.1.4.240 would be better and sent ICMP redirect message back to the HP-UX
box. If such an ICMP redirect message is received, such a route is created automatically and the alternative gateway is used.
Therefore, if you set the ndd parameter above to zero, you may see some network problem when you attempt to access a remote host over the default gateway. This is because the fact that you saw such a route created indicates that the default gateway has some difficulty to transfer the packet to the remote system and suggests us to use alternative path.
Thus, if you set the parameter to zero, you need to be well-aware of such risk. If you want to minimize the lifespan of such *tentative* route created by ICMP redirect message from the default GW, you can set the parameter value to smaller number than default.
Hope this helps.
kab
I work for HPE
