Subject: Re: kern/11923: new NFS client hangs with "ne0: warning - receiver ring buffer overrun"
To: None <gnats-bugs@gnats.netbsd.org, netbsd-bugs@netbsd.org>
From: ITOH Yasufumi <itohy@netbsd.org>
List: netbsd-bugs
Date: 01/12/2001 18:59:29
In article <20010110024126.A22121@vaasje.org>
fvdl@wasabisystems.com writes:

> You should probably try reducing the read and write size for the NFS mount.
> The ne2000 card probably can't handle the new, bigger defaults of 32k.

I tried some values of -r / -w.

    acha# mount_nfs -r 8192 -w 8192 pino:/usr /mnt2
    acha# (cd /mnt2; pax -rwpe src /usr)
	OK.  No warnings

    acha# mount_nfs -r 16384 -w 16384 pino:/usr /mnt2
    acha# (cd /mnt2; pax -rwpe src /usr)
	Works, but frequent "ne0: warning - receiver ring buffer overrun"

    acha# mount_nfs -r 32768 -w 32768 pino:/usr /mnt2     # default
    acha# (cd /mnt2; pax -rwpe src /usr)
	"ne0: warning - receiver ring buffer overrun"
	 and hangs-up

    acha# mount_nfs -T pino:/usr /mnt2
    acha# (cd /mnt2; pax -rwpe src /usr)
	Works.  Some "ne0: warning - receiver ring buffer overrun" messages


In article <20010110134820.A22832@vaasje.org>
fvdl@vaasje.org writes:

> Quality I suppose.. some ne2000 cards work ok, others choke quickly
> (and perhaps have less shared ram). It's the same with other cheap
> cards, like the rtl8139 based ones. Some work fine, others barely work.

Hmm, isn't it possible to handle such situation more user-friendly?
"If you forgot the options, system hangs" doesn't seem so cool....


In article <20010110194240.A211@antioche.eu.org>
bouyer@antioche.lip6.fr writes:

> Did you check a, say, 1.5E client with the 1.5Q server ?

I'll try this later (yesterday's build failed :( ).