On Mon, Aug 16, 1999 at 09:39:41AM -0700, john daley wrote:
>
> I had a similar situation (don't ask ;) and worked around it by increasing the value
>of
> NR_SUPER (default 64) in .../include/linux/fs.h and recompiling the kernel. Note
> that this was on a 2.0.35 kernel. NR_SUPER is set to 256 on the 2.25 kernel (which
> comes with redhat 6.0), so maybe just upgrading your kernel would be easier.
>
> -john
I know of what you speak. We ran into thsi problem early on in running out of
the total number of superblocks that we coudl track, but we increased it to
an absurdly large number, and hte problem has gone away. This is an EFIL.
That is "mount table full" where we run out of pseudo-minors under
NODEV_MAJOR. I think I have a general-case solution in mind, but it will
take a decidedly non-trivial time to implement.
It seems odd to me that no one has needed to mount more than 250 or so
nfs filesystems simultaneously before.
Gene
