I also am getting timeouts now when transferring a large amount of data
from an rsync daemon running on Linux kernel 2.2.16-3smp to a solaris 2.7
client using rsync 2.4.6. Traces show that one of the two processes on the
client side is not getting any activity and then it times out. I tried
lots of things and haven't been able to find a solution yet for this very
reproducible problem. I tried Neil Schellenberger's timeout fix from
http://lists.samba.org/pipermail/rsync/2000-June/002466.html and that did
help but then I still got another timeout later on. My guess is that the
fix only masks the problem, and that really both client processes are
supposed to keep getting activity and the one is getting hung before the
other. I tried turning off the "set_nonblocking()" calls at the beginning
of main.c:client_run() because those weren't in rsync 2.3.2 but that didn't
help. I even tried undoing the change to io.c from just after rsync 2.3.2
(CVS revision 1.66 on io.c) which removed the ssh "Uggh" extra buffering
and that didn't help either. I haven't yet tried going back to rsync
2.3.2, I was hoping to find the problem in 2.4.6.
More tomorrow, I'm done for the day.
- Dave Dykstra
