Re: slow ethernet
On Wed 24 Jan 2001, Luke Shulenburger wrote:
> I just bought a 100mb switch and have noticed that my alpha is getting
> really poor
> performance over the network. I have tried other computers and they get
> transfer
> rates that are acceptable (20mb/s or so). However, whenever I transfer
> to or from
> my alpha (an LX with a tulip ethernet card with woody installed) I get
> about 100kb/s
> download and a maximum of 50kb/s upload speed.
This is fairly typical of a wrong halfduplex / fullduplex setting.
There's a utility available from Donald Becker's site that can tweak
that setting (I forget the site name at the moment, and the utility's
name, although that was something starting with mii).
> I don't exactly know where to start attacking the problem. Here are a
> few things
> that look suspicious. Whenever I ping anything I get messages like so:
>
> 64 bytes from 18.239.0.1: icmp_seq=6 ttl=255 time=4.0 ms
> wrong data byte #8 should be 0x8 but was 0x5c
> c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22
> 23 24 25 26 27 28 29 2a 2b
> 2c 2d 2e 2f 0 0 0 0 0 0 0 0 0 0 0 0
This is a stupid bug in ping, where sizeof(struct timeval) was used
everywhere except for 3 places where there was a hardcoded 8. I've
found the bug and reported it, but I haven't seen any action on it
yet. I may have to do an NMU...
> and this shows up in /var/log/syslog
> Jan 24 22:59:53 bartender kernel: ping(19841): unaligned trap at
> 0000000120003110: 000000012002671c 29 2
Hmm, hadn't noticed that.
However, I recommend installing iputils-ping instead of netkit-ping
(at least for the time being), that ping works better. I do get this,
though:
$ ping -c 3 gw
PING gw.local (192.168.1.254) from 192.168.1.2 : 56(84) bytes of data.
64 bytes from gw.local (192.168.1.254): icmp_seq=0 ttl=255 time=442 usec
64 bytes from gw.local (192.168.1.254): icmp_seq=1 ttl=255 time=66 usec
Warning: time of day goes back, taking countermeasures.
64 bytes from gw.local (192.168.1.254): icmp_seq=2 ttl=255 time=1.209 msec
Until you see the warning message, don't rely on the times. I have no
idea what this means, though.
Anyway, that will have nothing to do with your slow ethernet problem.
> I'm also getting messages in the /var/log/syslog like this whenever I
> try to transfer files
> with samba:
> Jan 24 14:18:08 bartender nmbd[19120]: warning: /etc/hosts.deny, line 1:
> missing ":" separator
> Jan 24 14:18:08 bartender nmbd[19120]: warning: /etc/hosts.deny, line 2:
> missing ":" separator
So fix lines 1 and 2...
> Jan 24 14:18:08 bartender nmbd[19120]: params.c:OpenConfFile() -
> Unable to open configuration file "/etc/samba/smb.conf":
> Jan 24 14:18:08 bartender nmbd[19120]: ^INo such file or directory
It definitely looks like you never configured samba? There *must*
be a /etc/samba/smb.conf file.
> Jan 24 15:38:01 bartender modprobe: modprobe: Can't locate module
> net-pf-10
Edit /etc/modutils.conf, uncomment the "alias net-pf-10 off" line.
It's for IPv6 which exim tries to use (and then it falls back to IPv4,
so this is no problem either).
> I have outputs from tulip-diag and from mii-diag as well. Basically
Ah, that's the one!
> I have lots of information but no idea where to start using it. Any
Figure out how to turn on / turn off fullduplex with mii-diag,
and see what difference it makes (you can do it on the fly).
Paul Slootman
--
home: paul@wurtel.demon.nl http://www.wurtel.demon.nl/
work: paul@murphy.nl http://www.murphy.nl/
debian: paul@debian.org http://www.debian.org/
isdn4linux: paul@isdn4linux.org http://www.isdn4linux.org/
Reply to: