>> That tells you what port amcheck started out on (853 in this case).
>
>on each client, or on the server?
Everywhere. I got tired of trying to guess what was going on so all
the code now logs what it's doing (but see below).
>i see no amcheck debug files on the clients. ...
The file name is related to what program runs it. You run amcheck on
the server. The client side is amandad and selfcheck.
>on the server i see two such
>files from my most recent amcheck run, containing:
>
># cat amcheck.20010422100006.debug
>...
>amcheck: dgram_bind: socket bound to 0.0.0.0.856
OK, that one makes sense.
># cat amcheck.20010422100006000.debug
>...
>amcheck: dgram_bind: socket bound to 0.0.0.0.684
But that one doesn't. Hmmmm. Nor does it make any sense that you have
two amcheck*debug files with the same timestamp (the extra 000 says the
second one knew better than overwrite the first). Did you run it twice
very quickly (with the same second)? This doesn't happen on my system --
I just have one file per run.
Note that the portrange stuff is a hint to Amanda, not an absolute.
The sequence of events is:
1. Try for the portrange, if it was requested during ./configure.
2. Try for a privileged port.
3. Try for any port.
So the implication is that the second time you ran amcheck, or the
second time it tried to get a port (since it looks like the two "runs"
have the same base timestamp) could not get a portrange port and fell
back to a "plain" privileged one.
I guess I need to add yet another debug message to bind_portrange() to
log details when it could not do what was asked for. In fact, would
you give the following a try? I don't think it will shed much light
on your actual problem except to show that Amanda tried to do what you
asked but for some reason all those ports are in use.
>Todd Pfaff
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
bind_portrange.diff
