lsi processing with ICMPv4

I guess, this is still valid. Can someone else than RWTH contribute to fix this issue?

Thanks.

Markus reported to me that he is having problems with this: ICMP messages have long delays, most of the are dropped and do not arrive to the application layer. Markus was running the firewall on both ends.

I tried this and I could ping successfully from one end but not from the other end. I was running firewall at both ends.

The automatic /proc forward rule was disabled earlier. One idea could be to allow all forward and (although dummy0 could be just enough) and icmp-related rules in e.g. /etc/sysctl.conf. Then reboot the machines and try if it changes the situation:

http://www.ducea.com/2006/08/01/how-to-enable-ip-forwarding-in-linux/

Did not have a chance to try this out myself yet, so just scribbling notes here...

Can you please explain why forwarding is causing trouble here. LSI processing takes place only on the end-hosts. Hence, there should be no need for forwarding.

[If forwarding in /proc is the problem, we can re-add this functionality in hipfw. However, forwarding should not be enabled by default: I might want to run hipfw connection tracking on an end-host and not enable forwarding in this scenario. Instead, there should probably be yet another command-line option.]

The proc setting were not the issue (tested). Something else is causing this problem.