I have troubles in using nfs-file-locking
on a linux based cluster. Here the relevant
system parameters:
kernel 2.2.2
knfs on
SuSE 6.0 (glibc)
I need process synchronization via the flock
mechanism in a heterogenious linux/solaris
cluster. But even within the linux cluster
file locking does not work properly. Having
linux nodes lnxA and lnxB and solaris (2.6)
nodes solA and solB I get the following
results, depending on which node is the
nfs-server and which two nodes try a con-
current flock (ok means, the second node
does not get an exclusive file lock).
nfs-server nfs-lock-1 nfs-lock-2 result
lnxA lnxA lnxA ok
lnxA lnxA lnxB not ok
lnxA lnxA solA not ok
lnxA solA solB ok
solA solA solB ok
solA solA lnxA not ok
solA lnxA lnxB not ok
solA lnxA lnxA ok
To me these results appear quite strange.
Do experts expect this behaviour and is there
a way of getting four "ok"'s for either
an linux or solaris nfs server?
Any more input required? Any tests I can
perform?
Thanks,
Ruben
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/