linux-mips-fnet
[Top] [All Lists]

Re: 2.1.131 strikes back!

To: linux-mips@fnet.fr
Subject: Re: 2.1.131 strikes back!
From: Richard van den Berg <R.vandenBerg@inter.NL.net>
Date: Wed, 20 Jan 1999 21:58:44 +0100 (MET)
In-reply-to: <199901201754.SAA07693@sparta.research.kpn.com>
On Wed, 20 Jan 1999, Karel van Houten wrote:

> > To all DECstatoin hackers: shows `ifconfig` such amount of errors?

> eth0      Link encap:Ethernet  HWaddr 08:00:2B:37:63:76
>           inet addr:139.63.1.12  Bcast:139.63.1.255  Mask:255.255.255.0
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:635500 errors:0 dropped:0 overruns:0
>           TX packets:0 errors:445581665 dropped:741538 overruns:0
>           Interrupt:2 Memory:a0020000-a0040000 
> 
> Running NFS-root without TX-ing packets :-), I think the driver mixes up 
> some statistics.

I think the statistics are right, in case of the LANCE error doesn't mean
there is no packet transmitted, it means that there was a transmitter
timeout. Look at the number of dropped packets. I'm trying to figure out
what's causing this. First thing that made me suspect LANCE's functioning
was ping'ing from a DS:

[root@mips /root]# ping 192.168.1.1
PING 192.168.1.1 (192.168.1.1): 56 data bytes
64 bytes from 192.168.1.1: icmp_seq=0 ttl=64 time=10.0 ms
ping: recvfrom: Unknown error 512
64 bytes from 192.168.1.1: icmp_seq=1 ttl=64 time=0.0 ms
ping: recvfrom: Unknown error 512
64 bytes from 192.168.1.1: icmp_seq=2 ttl=64 time=0.0 ms
ping: recvfrom: Unknown error 512

Like Harald said: 
As I've said, I wouldn't call 0.007 rock-solid, but, hey, we're hacking
aren't we?

:-)

> > > P.P.S: I am going to do some work on the annoying "Warning: File '...' 
> > > has modification time in the future" feature :-).
> > 
> > rdate <name of your nfs-server> in the beginning of rc.sysconfig solves
> > that too. :-)

> We get about 30% to much clock ticks, so my solution is:

128 Hz instead of 100 Hz (Harald told me), dec_time_init() in
arch/mips/dec/setup.c, if I know what to change the DS's are running
more accurate.

> #!/bin/sh
> while true
> do
>     ntpdate turgon gandalf olorin frodo
>     sleep 5
> done &
> 
> (This will cause ca. 1 second correction each 5 seconds)

The other problem is that the DS starts with an incorrect date, I don't
know if `clock` would take care of that, but I found rdate a working
workaround (platform independent, works with NetBSD too).

Regards,
Richard

<Prev in Thread] Current Thread [Next in Thread>