Weird Linux - FreeBSD/ZFS NFSv4 interoperability problem

[Rick Macklem]

> 
> root at jester1d / ->mount -t nfs -o wsize=65536,rsize=65536
> servant.ixsystems.com:/a/isos /mnt
> root at jester1d / ->cd /mnt
> root at jester1d /mnt ->dd if=PCBSD8-STABLE-20100420-x64-DVD.iso
> of=/dev/null
> bs=1m
> 3344+1 records in
> 3344+1 records out
> 3507386368 bytes transferred in 34.562502 secs (101479528 bytes/sec)
> 
> root at jester1d /mnt ->cd ..
> root at jester1d / ->umount /mnt
> root at jester1d / ->mount -t newnfs -o nfsv3,rsize=65536,wsize=65536
> servant.ixsystems.com:/a/isos /mnt
> root at jester1d / ->cd /mnt
> root at jester1d /mnt ->dd if=PCBSD8-STABLE-20100420-x64-DVD.iso
> of=/dev/null
> bs=1m
> 345+0 records in
> 345+0 records out
> 361758720 bytes transferred in 46.191718 secs (7831679 bytes/sec)
> 
> The first run hits network limits.
> 
Hmm, the newnfs case seems to have terminated prematurely. That's a
different problem than the others seemed to report, but you definitely
have a slow read rate.

Could you by any chance run the newnfs test again and capture a "ps axHl"
on the client (I'm hoping that will hint at where the threads are sleeping).

Thanks for doing the test, rick