difference between "ping -I INTERFACE" and "ping -I IPADDRESS" ?

[Frantisek Hanzlik]

JB wrote:
> Frantisek Hanzlik <franta <at> hanzlici.cz> writes:
> 
>> ...
> 
> Suggestions:
> 
> 1. use -R option to see routing (like ping plus traceroute in one)
>    Perhaps it will give you some clue about how it travels.
>    # ping -R -n -c 2 -I source_ip target_ip 
>    # ping -R -n -c 2 -I source_if target_ip
> 
>    You may try with -B option as well, just in case.
>    # ping -B -R -n -c 2 -I source_ip target_ip
>    # ping -B -R -n -c 2 -I source_if target_ip
> 
> 2. install an older F13 or a newer F15 client on the same network, or perhaps 
>    you would be able to remove on F14 the iputils package and replace it by 
>    e.g. F13's iputils.
>    Then run the tests again to see if it is a case of ping regression.
>  
> 3. play with -Q tos option (man ping)
> 
> 4. get in touch with your ISP and ask them for co-testing with you (capturing
>    your ping's on the way out and on the way back).
>    I guess 'tcpdump' or similar net tool would allow them to see both ICMPs
>    (request and echo).
>    They could test the target_ip from their net too.
> 
> JB

Hello JB,
I vote for ending this case - "ping -I source_ip" works as I expected,
traceroute works as I expected. "ping -I source_if" for some reason
behaves differently (it seems as it tries find target at source_if
segment, according to ARP requests), but I have not time for further
discovery. Linux box works fine, and it is crucial for me.
IMO right way would be ping sources overview and eventually some its
debugging.

Thanks, Franta