On Wed, Oct 25, 2000 at 10:37:21AM +0200, Eckebrecht von Pappenheim wrote:
> > I and a colleague have now done some sniffing and it does indeed look like
> > a Linux bug, but not the one you describe. In fact, we once saw just
>
> We've the same Problem between two solaris-machines. Switching back to
> 2.3.1 solved the problem...
Then it must have been a different problem with the same symptom. I
tested putting the 2.3.x "Uggh" buffering hack back in but it didn't help
this case; I saw in gdb that the rsync daemon server/sender was stuck
during an io_flush() which writes from within a reading function, and
in that case the hack does not try to read more because it is already
wanting to read.
I also have been experiencing problems with solaris-solaris with the same
set of test data, going over rsh, but I'm quite convinced it is a different
problem. I haven't yet been able to reliably reproduce it solaris-solaris
so I'm not sure if the buffering hack will work there but I'm guessing that
it will. Were using rsh, ssh, or daemon mode?
- Dave Dykstra
