Date: Fri, 02 May 1997 09:19:32 +0000 From: Kachun Lee <kachun@zippo.com> To: dg@root.com Cc: freebsd-stable@freebsd.org Subject: Re: kmem_map full with 2.2beta to 2.2-releng-970422 Message-ID: <1.5.4.32.19970502091932.006d6bc4@dopey.pathlink.com>
next in thread | raw e-mail | index | archive | help
At 02:05 PM 5/1/97 -0700, you wrote: >>clients, and our NFS servers have quite a lot of these: >> >> /kernel: nfsd send error 55 >> >>They does not seem cause any ill effect. I assume 55=ENOBUFS No buffer space >>available. But I thought with the unified VM/Buffer, there are no MAXBUFFER >>setting any more. Do you know what parameter I should adjust? > > ...that one indicates that network buffers were depleted. This usually >happens when the interface queue reaches it's limit of 50 queued packets, >although it can also happen if the global pool of network buffers is deleted. >The most common cause of this error is a stuck network device driver - some >NICs, for example will stop sending if the cable is unplugged. It can also >happen on PPP/SLIP lines if hardware flow control (CTS) stays de-asserted too >long. I kind of know it is probably related to the 'de' drivers and the 50+% collicion rate. I am almost there giving up waiting for the 'new' de driver to make it to stable. I should just throw away all the SMC cards, get some Intel ones and use your driver. Thank again for the info.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1.5.4.32.19970502091932.006d6bc4>