We have a mix of SL3 (304, 305, 307) systems
for desktops and compute servers, and a mix
of NetApp filers and RAID-based PCs filers
running various releases of SL and older RH.

We have a problem where files created on NFS
mounts from one node don't appear on other
nodes for some time.  Most of our mounts use
TCP (this was done to get around other problems).
All these mounts exhibit the above problem.  we
also have one older PC that uses UDP only, and
so far nobody has seen the problems there.

NetApp has suggested we set the "noac" option
on the NFS mounts.  We're concerned this will
have a pretty noticeable performance hit.  So
we were wondering:

1) Is this commmon practice in clusetered environments?
2) Would something like actime=1 be better?
3) Would we have the same problem with SL4?

Any and all feedback appreciated.

Thanks,
Miles