Date: Thu, 6 Sep 2001 14:23:39 +0200 From: Francesco Casadei <fcasadei@inwind.it> To: Jon Loeliger <jdl@jdl.com> Cc: Francesco Casadei <fcasadei@inwind.it>, questions@freebsd.org Subject: Re: No memory for rx list? Message-ID: <20010906142339.A2504@goku.kasby> In-Reply-To: <E15ehmV-0001oj-00@jdl.com>; from jdl@jdl.com on Wed, Sep 05, 2001 at 01:52:09PM -0500 References: <20010905203010.A1829@goku.kasby> <E15ehmV-0001oj-00@jdl.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--zYM0uCDKw75PZbzx
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Wed, Sep 05, 2001 at 01:52:09PM -0500, Jon Loeliger wrote:
> So, like Francesco Casadei was saying to me just the other day:
> >=20
> > from xl(4) manpage:
> >=20
> > xl%d: no memory for rx list The driver failed to allocate an mbuf=
for
> > the receiver ring.
> >=20
> > xl%d: no memory for tx list The driver failed to allocate an mbuf=
for
> > the transmitter ring when allocating a pad buffer or collapsing an=
mbuf
> > chain into a cluster.
> >=20
> > You're running low on mbufs. What does netstat -m show?
> >=20
> > Francesco Casadei
>=20
> Yep, low mbufs. Good question. It shows:
>=20
> www 1429 # netstat -m
> 137/1248/4096 mbufs in use (current/peak/max):
> 131 mbufs allocated to data
> 6 mbufs allocated to packet headers
> 129/1024/1024 mbuf clusters in use (current/peak/max)
^^^^ ^^^^
> 2360 Kbytes allocated to network (76% of mb_map in use)
> 180 requests for memory denied
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> 0 requests for memory delayed
> 0 calls to protocol drain routines
>=20
> This is currently a semi-quiet network. Is most of that 76%
> now dead space?
>=20
> At least we're not freeing free mbufs!
>=20
> jdl
>=20
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-questions" in the body of the message
>=20
> end of the original message
=46rom tuning(7) manpage:
NMBCLUSTERS may be adjusted to increase the number of network mbufs the
system is willing to allocate. Each cluster represents approximately =
2K
of memory, so a value of 1024 represents 2M of kernel memory reserved =
for
network buffers. You can do a simple calculation to figure out how ma=
ny
you need. If you have a web server which maxes out at 1000 simultaneo=
us
connections, and each connection eats a 16K receive and 16K send buffe=
r,
you need approximate 32MB worth of network buffers to deal with it. A
good rule of thumb is to multiply by 2, so 32MBx2 =3D 64MB/2K =3D 3276=
8. So
for this case you would want to se NMBCLUSTERS to 32768. We recommend
values between 1024 and 4096 for machines with moderates amount of mem=
=AD
ory, and between 4096 and 32768 for machines with greater amounts of m=
em=AD
ory. Under no circumstances should you specify an arbitrarily high va=
lue
for this parameter, it could lead to a boot-time crash. The -m option=
to
netstat(1) may be used to observe network cluster use.
So do your own calculations to get a new value for network mbufs, let it be=
n.
Then put the following line in your kernel config file and recompile:
options NMBCLUSTERS=3Dn
Francesco Casadei
--=20
You can download my public key from http://digilander.iol.it/fcasadei/
or retrieve it from a keyserver (pgpkeys.mit.edu, wwwkeys.pgp.net, ...)
Key fingerprint is: 1671 9A23 ACB4 520A E7EE 00B0 7EC3 375F 164E B17B
--zYM0uCDKw75PZbzx
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (FreeBSD)
Comment: For info see http://www.gnupg.org
iD8DBQE7l2rLfsM3XxZOsXsRAiFOAKDNLUouS+jWUDga9ZLo/3U//XNgxgCdHlUr
rCqA/vD+Brn96aewLPOyEe8=
=w5oU
-----END PGP SIGNATURE-----
--zYM0uCDKw75PZbzx--
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010906142339.A2504>