Hi Alan,
Phil Pierotti wrote:As opposed to looking at the debug logs from the server?
> The main reason my initial checking was with tcpdump was to identify
> what the packets were doing.
You can look at the high level "packet in / packet out" view. Or, you
can look at the detailed log messages from the server saying what's
going wrong, and why.
You chose the high-level view. I'm trying to convince you to look at
the informative view. You don't seem to see this as useful.
There is a LOT more that can go on than just packet in / packet out.
> So the problem is not due to the downstream RADIUS failing to respond at
> all. (ie genuinely/obviously dead)
Maybe you might want to look for more information.
Since you've ALREADY looked at the tcpdump logs and they didn't
help... maybe looking for MORE information would be a good idea.
If you had bothered to run the server in debugging mode, OR to use
> Freeradius is deciding that my accounting proxy destination 'looks like
> it is dead'.
>
> Tue Feb 16 07:45:32 2010 : Proxy: Marking home server {{radius-ip}}
> port 1813 as zombie (it looks like it is dead).
"raddebug" as I suggested, you might find out WHY this is happening.
Can you explain why you're not doing that?
You've said that. It's simply one step out of many. For some reason,
> I'm not entirely sure why, though. TCPDUMP shows a stream of accounting
> requests being proxied to that server, and being received from that
> server, while at the same time freeradius continues to log 'looks like
> it is dead' messages.
you can't get past "BUT PACKETS ARE GOING BACK AND FORTH". You have
refused to follow the instructions which can help you solve the problem.
There is no "naturally" here.
> Also, status check (via request) succeeds, naturally, given that its
> not the auth proxy that freeradius is complaining about.
> (single downstream RADIUS configured as auth+acct)
Run in debugging mode. Have I said that enough? Is there anything
else I need to say to convince you to run in debugging mode?