rsync 2.6.8 timeouts

[Robin Getz]

Hi.

I have an anonymous rsync server set up on cvs.blackfin.uclinux.org, which 
allows people to take our entire cvs/snv respositories (which are _huge_) 
and do local checkouts. It works really well for most people.

I have a client set up on a static IP numbers, off the backbone, which work 
flawlessly. It never errors, and never has a problem. Thanks to everyone 
for their hard work - for an application that just works as advertised.

But - I have a client at my house (via comcast, dynamic IP number) which 
has nothing but problems...

I am using 2.6.8 (release) on all setups.


rsync -azvvv cvs.blackfin.uclinux.orguclinux533/CVSROOT/history 
/cvsroot/uclinux533/CVSROOT/history

opening tcp connection to cvs.blackfin.uclinux.org port 873
opening connection using --server --sender -vvvlogDtprz . 
uclinux533/CVSROOT/history
receiving file list ...
recv_file_name(history)
received 1 names
done
recv_file_list done
get_local_name count=1 /cvsroot/uclinux533/CVSROOT/history
recv_files(1) starting
generator starting pid=27391 count=1
delta-transmission enabled
recv_generator(history,0)
generating and sending sums for 0
count=33367 rem=2109 blength=33352 s2length=3 flength=1112824941
generate_files phase=1

and then nothing.

If I strace the process, I get


home/home/rgetz/testcvs/gcc1 # strace -p 27391
Process 27391 attached - interrupt to quit
select(5, [4], [], NULL, {22, 983000})  = 0 (Timeout)
select(5, [4], [], NULL, {60, 0})       = 0 (Timeout)
select(5, [4], [], NULL, {60, 0})       = 0 (Timeout)
select(5, [4], [], NULL, {60, 0})       = 0 (Timeout)

and it repeats, every 60 seconds.

Any idea? I an update the the pre-release of 2.6.9 if someone can tell me 
they think it might help (and why).

Thanks
-Robin