Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 6 Sep 2002 15:47:49 +0300
From:      Giorgos Keramidas <keramida@ceid.upatras.gr>
To:        Wahyu Hidayat <wahyuhid@yahoo.com>
Cc:        freebsd-questions@FreeBSD.org
Subject:   Re: running out of mbuf
Message-ID:  <20020906124749.GD2168@hades.hell.gr>
In-Reply-To: <20020906022036.90263.qmail@web40013.mail.yahoo.com>
References:  <20020905184835.P1069@seven.alameda.net> <20020906022036.90263.qmail@web40013.mail.yahoo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
[-- Removed freebsd-stable from the Cc: header. ------]
[-- Please do not cross post to more than one list. --]

On 2002-09-05 19:20 +0000, Wahyu Hidayat wrote:
> Thanks for all your immediate reply.
> 
> Sure I have raised NMBCLUSTERS (in my netstat shows to
> 20480).
> 
> What is the explanation for my previous post about max
> mbufs has not reached, but it already shows about mbuf
> allocation failed ? What if I increase nmbclusters but
> this problems still occurs ?

Mbufs are internal kernel structures used for buffering network data.
When an mbuf needs to hold a lot of data, an `mbuf cluster' is
attached to it as extra space.  You have a lot of free mbufs to use,
but your clusters are running out.  This way when someone requests
that an mbuf is allocated to hold a large amount of data the
allocation that has two stages (first allocate an mbuf, then a cluster
to attach to it) fails during the second stage--the mbuf cluster
allocation.  Looking at the netstat -m output you have posted:

> 22070/22544/81920 mbufs in use (current/peak/max):
>	   20490 mbufs allocated to data
>	   1580 mbufs allocated to packet headers
> 20462/20480/20480 mbuf clusters in use (current/peak/max)
> 46596 Kbytes allocated to network (7% of mb_map in use)
> 87932 requests for memory denied
> 274 requests for memory delayed
> 0 calls to protocol drain routines

you can easily see that the peak use of mbuf clusters is 100%.  When
all of your mbuf clusters have been used, an allocation that will
request more clusters will fail.  This is what happens, when you are
getting those messages.

- Giorgos

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?20020906124749.GD2168>