On Fri, May 04, 2007 at 08:39:30PM -0400, shivers_at_ccs.neu.edu wrote:
> Fifteen minutes later, I cd down into the coda volume on client A and create a
> 700kb file (by copying one of my students' dissertation to a new file).
> Then I say "cfs strong."

If you want to force a reintegration of already logged operations it is
better to use 'cfs forcereintegrate' (or 'cfs fr').

cfs strong changes the reintegration parameters of a volume so that any
further operations are synchronously reintegrated, but it doesn't start
reintegration for already logged operations.

> Ten minutes later, it's still not visible on client B.
> During this ten minutes, venus on client A spins, maxing out my load average.

It may be that the new reintegration parameters set by cfs strong are
causing some sort of a problem. I'll have to see if I can reproduce the
problem.

> Then I try it again, with a new copy, and it works instantly.

This operation starts off using the parameters set by cfs strong, and as
such I would have expected this to work. cfs strong sets the
reintegration parameters so that there is an implicit forcereintegrate
on before we send a response back to the kernel, so that as far as the
application is concerned the operation has been committed on the server
before the system call returns.

> Client A is running Fedora Core with the coda-client-6.9.1-1 rpm.
> However, it is running an older kernel: 2.6.17.14 with the coda kernel
> that came *with* that kernel. (Do I need to be compiling from the
> latest coda kernel modules?)

I don't think it is a kernel module issue.

Jan