Date: Sat, 13 Jan 1996 13:30:09 -0600 (CST) From: "John A. Booth" <john@ulantris.infinop.com> To: lenzi@cwbone.bsi.com.br (Sergio Lenzi) Cc: questions@freebsd.org Subject: Re: mb_map full Message-ID: <199601131930.NAA18719@ulantris.infinop.com> In-Reply-To: <Pine.BSF.3.91.960113143017.4051A-100000@cwbone> from "Sergio Lenzi" at Jan 13, 96 02:33:45 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > > > what does this error mean? > > You're out of phisical memory in the kernel. (I think). > > what caused this...remedies...etc > > Remedies: recompile the kernel with the option NMBCLUSTERS=xxx > try xxx=2048,3072,4096 depending on the ammount of memory your > machine has... > Well...your out of Network Memory Buffers....usually you'll only encounter the mb_map full message when your on a busy network...try 1024 then go up by 512...1024 will take up 2 meg of the physical machine memory...so you don't really want to put a very large number for NMBCLUSTERS. I had problems with this on a local net that had ~ 200 pc's running netware clients, 2 NetWare servers, and about 20-30 unix boxes. You might also look @ netstat -m, it'll show how many mbuf clusters you have total (this is equivalent to NMBCLUSTERS in your configuration file.) I'm on a much less busy net now, but here's what a netstat -m on my machine looks like. I could probably drop my NMBCLUSTERS back down to 1024. 654 mbufs in use: 589 mbufs allocated to data 49 mbufs allocated to packet headers 15 mbufs allocated to protocol control blocks 1 mbufs allocated to socket names and addresses 227/2048 mbuf clusters in use 4177 Kbytes allocated to network (12% in use) 0 requests for memory denied 1 requests for memory delayed 1 calls to protocol drain routines
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199601131930.NAA18719>