[ubuntu-us-mi] NFS thread stats showing as zero

[Robert Citek]

Anyone know why lines 2-12 are showing up as zero?

# grep ^th /proc/net/rpc/nfsd | tr -s ' ' '\n' | tail -n +2 | cat -n
     1	64
     2	0
     3	0.000
     4	0.000
     5	0.000
     6	0.000
     7	0.000
     8	0.000
     9	0.000
    10	0.000
    11	0.000
    12	0.000

This is on an Ubuntu 10.04 LTS server running an NFS service.  I'm
guessing that lines 2-12 should not be zero, especially line 3.  But
since they are, what does that mean?  Has anyone else seen this
behaviour on their NFS servers?

The "understanding-linux-nfsd-statistics"[1] reference does a pretty
good job of explaining the th line.  Briefly, line 1 is the number of
threads.  Line 2 is the number of times all threads were needed.
Lines 3-12 are a histogram of thread usage in 10% groups, starting
with 0-10%.

References:

[1] http://blog.kamilkisiel.net/2007/11/understanding-linux-nfsd-statistics.html
[2] http://billharlan.com/pub/papers/NFS_for_clusters.html
[3] http://nfs.sourceforge.net/#section_b
[4] http://nfs.sourceforge.net/nfs-howto/ar01s05.html (section 5.6)

Regards,
- Robert