No attempt is made to verify the information in
/etc/mtab
itself.
Since swap file mounts are not recorded in
/etc/mtab ,
a heuristic specific to SunOS is used to determine whether such a mount
is actual (replacing the string "swap" with "root" and verifying the resulting
path).
Symbolic links on the server will cause the path in the remote entry to differ
from the one in
/etc/mtab .
To catch those cases, a filesystem is also deemed mounted if its
local
mount point is identical to the remote entry.
I.e., on a SunOS diskless client,
server:/export/share/sunos.4.1.1
is actually
/usr/share .
Since the local mount point is
/usr/share
as well this will be handled correctly.
There is no way to clear a stale entry in a remote mountd after the
local hostname (or whatever reverse name resolution returns for it)
has been changed. To take care of these cases,
the remote /etc/rmtab file has to be edited and mountd restarted.
The RPC timeouts for mountd calls can only be changed by recompiling.
The defaults are 2 seconds for client handle creation and 5 seconds for
RPC calls.