3c59x: recurring transmit errors

Ravi Swamy ravi@rtp.mea.com
Tue Jul 28 12:07:40 1998

Please don't start this sync vs. async file system crap.  Nobody really
cares and while it starts out as a real debate it never ends in any kind
of agreement and just a bunch of silly name calling.

Your friend is somewhat correct on Linux NFS writes.  Linux NFS writes
to another Linux box are quite fast.  I've gotten over 2.5MB/s which I'm
quite happy with.  Linux NFS writes to a Solaris server are horrendous.
I can't even get 300kB/s.  No I'm not assigning blame, for all I know it
could be Solaris but I've seen at least 3 people verify the same behavior
and none knew how to "fix" it.  No, changing the rsize and wsize did not
help at all.  Maybe 5% but that's it.  I've tried 1k, 2k, 4k, 8k, 16k, 32k.

I WAS pleasantly surprised that Solaris was able to quickly write to an
Linux NFS server.  I would love to have this fixed and would gladly test
out suggestions or patches anyone has.

On Tue, 28 Jul 1998, Kevin Cameron wrote:

>         have you tried nfs writes under linux ?
>         in the best case, it is 10% of the performance
>         of nfs writes under solaris or freebsd.
>         (i use freebsd because it handles large jobs,
>         much, much better than linux, and handles nfs
>         a million times better than linux.  however,
>         linux is more popular, and may win in the long
>         run.  i tell people:  if you use linux instead
>         of windoz-nt, consider yourself very, very fortunate.
>         if you use freebsd instead, consider yourself
>         enlightened).  watch out for disk crashes under
>         linux - the extfs file system is like the ufs
>         filesystem running with the -async switch - ie,
>         fsck equivalent under linux is not guarnteed to be
>         able to recover from all power crashes (bob
>         ran into this, and has now switched
>         to freebsd).
> You might be blaming the wrong thing.
> Kev.