Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 4 Jan 2006 23:43:57 -0500
From:      David Gilbert <dgilbert@dclg.ca>
To:        Francisco Reyes <lists@stringsutils.com>
Cc:        FreeBSD ISP <freebsd-isp@freebsd.org>, David Gilbert <dgilbert@dclg.ca>
Subject:   Re: FreeBSD +NFS + mail services
Message-ID:  <17340.41997.290075.251660@canoe.dclg.ca>
In-Reply-To: <cone.1136401744.131083.65830.1000@zoraida.natserv.net>
References:  <cone.1136336702.793300.48688.1000@zoraida.natserv.net> <17339.61532.649959.1059@canoe.dclg.ca> <cone.1136401744.131083.65830.1000@zoraida.natserv.net>

next in thread | previous in thread | raw e-mail | index | archive | help
>>>>> "Francisco" == Francisco Reyes <lists@stringsutils.com> writes:

Francisco> David Gilbert writes:
>> remaining processes have little or none.  As more of your NFS
>> processes are charged for time, you may want to increase the number
>> you run.  FreeBSD seems to run more client daemons dynamically, so
>> you only need to worry about the number of server daemons you run.

Francisco> I have the following on a machine: ps aux|head -n 1|ps auxw
Francisco> |grep nfs root 317 0.0 0.0 0 8 ??  DL 12:47PM 0:00.41
Francisco> [nfsiod 0] root 318 0.0 0.0 0 8 ??  DL 12:47PM 0:00.31
Francisco> [nfsiod 1] root 319 0.0 0.0 0 8 ??  DL 12:47PM 0:00.19
Francisco> [nfsiod 2] root 320 0.0 0.0 0 8 ??  DL 12:47PM 0:00.12
Francisco> [nfsiod 3] root 2316 0.0 0.0 0 8 ??  DL 1:17PM 0:00.06
Francisco> [nfsiod 4] root 2317 0.0 0.0 0 8 ??  DL 1:17PM 0:00.04
Francisco> [nfsiod 5] root 5486 0.0 0.0 0 8 ??  SL 1:39PM 0:00.02
Francisco> [nfsiod 6] root 5637 0.0 0.0 0 8 ??  SL 1:39PM 0:00.01
Francisco> [nfsiod 7] root 5638 0.0 0.0 0 8 ??  SL 1:39PM 0:00.01
Francisco> [nfsiod 8] root 6563 0.0 0.0 0 8 ??  SL 1:46PM 0:00.01
Francisco> [nfsiod 9] root 6564 0.0 0.0 0 8 ??  SL 1:46PM 0:00.00
Francisco> [nfsiod 10] root 6565 0.0 0.0 0 8 ??  IL 1:46PM 0:00.00
Francisco> [nfsiod 11] root 6566 0.0 0.0 0 8 ??  IL 1:46PM 0:00.00
Francisco> [nfsiod 12] root 6567 0.0 0.0 0 8 ??  IL 1:46PM 0:00.00
Francisco> [nfsiod 13] root 7089 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 14] root 7090 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 15] root 7091 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 16] root 7092 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 17] root 7093 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 18] root 7095 0.0 0.0 0 8 ??  IL 1:51PM 0:00.00
Francisco> [nfsiod 19]

Francisco> It seems that for the most part most of those nfs clients
Francisco> are not used. I then wonder why they were started.

Francisco> What would be the impact on setting the upper limit of
Francisco> clients to 10?  Specially on a case like the above it
Francisco> doesn't seem like 20 were needed to begin with..

It does seem like most of them arn't used.  I'd say your number is
somewhere between 4 and 8, but considering that these are just kernel
threads, having 20 around may not be costing you enough to worry.
Certainly, you don't require more.

Francisco> Also, on the server how does one change the number of
Francisco> running server daemons? Only way I found was to kill all
Francisco> nfsd daemons and start it with the new number of
Francisco> clients.. there must be a better way (I sure hope so).

Now... this could be memory from other versions of UNIX showing up,
but I seem to remember running nfsd with a larger -n argument starting
mroe daemons ... but that might not be FreeBSD.  The server side is
certainly less dynamic.

Dave.

-- 
============================================================================
|David Gilbert, Independent Contractor.       | Two things can be          |
|Mail:       dave@daveg.ca                    |  equal if and only if they |
|http://daveg.ca                              |   are precisely opposite.  |
=========================================================GLO================



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?17340.41997.290075.251660>