robert@computing-maths.cardiff.ac.uk (Robert Evans) (04/29/91)
A week after upgrading to Ultrix 4.1, we started getting the follow error:
----- EVENT INFORMATION -----
EVENT CLASS OPERATIONAL EVENT
OS EVENT TYPE 250. ASCII MSG
SEQUENCE NUMBER 143.
OPERATING SYSTEM ULTRIX 32
OCCURRED/LOGGED ON Sat Apr 27 11:48:36 1991 WET DST
OCCURRED ON SYSTEM v1
SYSTEM ID x05903B14
PROCESSOR TYPE KA825
PROCESSOR COUNT 2.
PROCESSOR WHO LOGGED x00000002
MESSAGE ku_fastsend: transmit failed on de0,
_errno 55
The message repeats every time something tries to access the network so all
network access is blocked.
The system is an 8350 with a DEUNA on the UNIBUS.
Field Service have (i) suggested that we are running out of mbufs and that this
can be increased by upping MAXUSERS. So we put it up to 256 with no effect.
(ii) changed the DEUNA (still no good). (iii) had me patch vmunix to change
ifmaxqlen to 250 (ifmaxqlen is defined in if.o).
But the problem still occurs. Anybody come across this before. Is there a
known fix that works?
Our configuration is as follows:
ident "V1"
machine vax
cpu "VAX8350"
maxusers 256
processors 2
maxuprc 50
physmem 26
timezone 0 dst 3
smmax 256
smseg 12
smbrk 1024
options QUOTA
options INET
options NFS
options RPC
options UFS
options SMP
options SYS_TRACE
options PACKETFILTER
... then the device info ...
scs_sysid 1
pseudo-device pty 64
pseudo-device loop
pseudo-device ether
pseudo-device inet
pseudo-device nfs
pseudo-device rpc
pseudo-device ufs
pseudo-device sys_trace
pseudo-device packetfilter
(The one other difference between this Ultrix and the last version 4.0 on I ran
is the packetfilter stuff which I've added for nfswatch. I haven't run
nfswatch though).
--
Robert Evans, Dept of Computing Maths, University of Wales College of Cardiff,
PO Box 916, Cardiff, Wales, UK, CF2 4YN. Tel: +44 (0)222 874000 x 5518
Internet: R.Evans%computing-maths.cardiff.ac.uk@nsfnet-relay.ac.uk
UUCP: R.Evans@cf-cm.UUCP or ...!uunet!mcsun!ukc!cf-cm!R.Evans
Janet: R.Evans@uk.ac.cardiff.computing-mathsgjap@ukc.ac.uk (Godfrey Paul) (05/01/91)
In article <1991Apr29.101150.8213@cm.cf.ac.uk> robert@computing-maths.cardiff.ac.uk (Robert Evans) writes: >A week after upgrading to Ultrix 4.1, we started getting the follow error: > ku_fastsend: transmit failed on de0, _errno 55 We get this error message on the console of DECstation 3100s if the ether is unplugged for more than a few seconds. All machines on the disconnected segment display this message repeatedly on their consoles, and have to be rebooted. Errno 55 is ENOBUFS, "No buffer space available"; reasonable if the network is jammed. But they ought to be able to recover. Suns have no problem with a temporary break in the ether and neither should they! Anybody know a fix for this? It's more than a bit irritating when moving kit about. We're running a version of MIPS Ultrix older than 4.0. I thought it was 3.something, but "strings /vmunix | grep Ultrix" gives "Ultrix Worksystem V2.1 (Rev. 14)" Godfrey