Greetings,
The answer to my query about "le0: out of mbufs - packet dropped"
messages preceding a system failure seems to be to install patch
100804-03 (TCP socket and reset problems). I have done this, and have
not seen these messages since, but they are infrequent enough so that I
don't know yet if this fixes the problem.
There are also fixes to mbuf leakage in the 100173 NFS jumbo patch, and
I have replaced my -10 version with -12.
Thanks go to:
"John A. Murphy" <jam@philabs.philips.com>
martin.williams@research.natpower.co.uk
Marc Santucci <marc@protocol.zycad.com>
Peg Heffron <heffron@mailbox.syr.edu>
Dan Lorenzini Greenwich Capital Markets
dal@gcm.com 600 Steamboat Road
203-625-6088 Greenwich, CT 06830
------------------------------------------------------------------------
Original message:
One of our NFS servers has recently been having a problem with mbufs.
The symptom is that we will get a few "le0: out of mbufs - packet
dropped" messages over a few minutes, and then the machine will
disappear from the net. From the console, we can usually reboot, but
sometimes it needs to be power cycled.
The machine is a sparc 10 with 256 MB of memory, running SunOS 4.1.3
with the following patches: 100173-10, 100075-09, 100359-06, 100726-13,
and with prestoserve installed. Eight 2GB Seagate Barracudas are
attached to two FSBE/S cards. It has connections to several busy
ethernets, and sometimes the messages refer to one of the other
interfaces. Also, we will sometimes see "nfs_server: bad sendreply"
messages happening around the same time.
Does anyone have any suggestions as to what to do to prevent this from
happening? We have already doubled the value of MBPOOLBYTES in param.h
but it has not fixed the problem. Thanks for any ideas you might have.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:31 CDT