oberst@phoenix.Princeton.EDU (Daniel J. Oberst) (01/10/89)
answered. Several times when I have left A/UX running unattended overnight (I usually run my MacII with MacOS during the day) I have come in to see the following error messages: ae0:overflow NIC reset failed ae6_intr:Receive overflow warning ku_fastsend: if_output failed: error=70, sendpck=0, am=0 Has anyone else seen this? Any idea what is causing it? I am still awaiting an upgrade to my (older) Ethertalk board. Could it be the source of the problem?
kucharsk@uts.amdahl.com (William Kucharski) (01/11/89)
In article <5262@phoenix.Princeton.EDU>, oberst@phoenix.Princeton.EDU (Daniel J. Oberst) writes: > answered. Several times when I have left A/UX running unattended > overnight (I usually run my MacII with MacOS during the day) I have > come in to see the following error messages: > > ae0:overflow NIC reset failed > ae6_intr:Receive overflow warning > ku_fastsend: if_output failed: error=70, sendpck=0, am=0 We've had a similar problem at our site. Basically, it seems A/UX can't deal with much Ethernet traffic. If your Ethernet is at all heavily used, the system will print the first two messages and bring the Ethernet card off-line. Perhaps that is the source of your third message. I don't know how to fix it, I'm just waiting for a fix from Apple. (From what I hear A/UX 1.1 fixes this, but when that will be out is another question...) -- William Kucharski ARPA: kucharsk@uts.amdahl.com UUCP: ...!{ames,decwrl,sun,uunet}!amdahl!kucharsk Disclaimer: The opinions expressed above are my own, and may not agree with those of any other sentient being, not to mention those of my employer. So there.
king@orion.arc.nasa.gov (Frank S. King) (01/11/89)
Daniel J. Oberst writes:
answered. Several times when I have left A/UX running unattended
overnight (I usually run my MacII with MacOS during the day) I have
come in to see the following error messages:
ae0:overflow NIC reset failed
ae6_intr:Receive overflow warning
ku_fastsend: if_output failed: error=70, sendpck=0, am=0
Has anyone else seen this? Any idea what is causing it? I am still
awaiting an upgrade to my (older) Ethertalk board. Could it be the
source of the problem?
=====================================================================
I used to have the same problem with my machine. Some time ago some
one posted a good suggestion that seemed to have cured the overflow
problem. Try
kconfig -n /unix
NMBUFS=800
^d
sync
sync
reboot
This will allocate 300 more buffers for networking, the default is 500.
you can read the manpages on kconfig to get the details on NMBUFS.
Best of luck
ANTHONY INTRAVAIA
NASA AMES RESEARCH CENTER
Space Life Sciences Payloads Office
liam@cs.qmc.ac.uk (William Roberts) (01/11/89)
In article <ecXu9c3TV410102tQOg@amdahl.uts.amdahl.com> kucharsk@uts.amdahl.com (William Kucharski) writes: >> ku_fastsend: if_output failed: error=70, sendpck=0, am=0 > >We've had a similar problem at our site. Basically, it seems A/UX can't >deal with much Ethernet traffic. If your Ethernet is at all heavily used, >the system will print the first two messages and bring the Ethernet card >off-line. We never see this problem, and our Ethernet is fairly well loaded (130+ machines inc Macs/Suns, lots of EtherTalk, lots of TCP/IP, a few diskless Suns, 45 machines mounting virtually everything from file servers via NFS, etc., etc.). However, we are using the Kinetics EtherPort II cards, so this may be a difference at that level (hardware/driver). For what it's worth, ku_fastsend is probably associated with NFS and the error=70 is a standard errno message, namely ENETDOWN, which is what you'd expect if the driver has just turned its toes to the sky... -- William Roberts ARPA: liam@cs.qmc.ac.uk (gw: cs.ucl.edu) Queen Mary College UUCP: liam@qmc-cs.UUCP LONDON, UK Tel: 01-975 5250
mmeyer@mcdurb.Urbana.Gould.COM (01/12/89)
I don't have a MAC-II, but the ku_fastsend you see is a procedure inside of NFS that takes NFS data/requests/etc adds a UDP and IP header, and calls (*ifp->if_output)(...) . Heavy NFS traffic could possibly be the cause of the Ethernet hang. --morris Morris Meyer mmeyer@urbana.mcd.mot.com uunet!uiucdcs!mcdurb!mmeyer Motorola Microcomputer Division, Champaign-Urbana Design Center 1101 E. University, Urbana, Illinois 61801, USA. My opinions are my own, and are not the opinions of my employer, or any other organisation.