UPDATE: 100 mb/s problem- Slow One Way

From: Aaron Flin <af_at_dogpile.com>
Date: Wed, 29 Jul 1998 16:51:11 -0700

Here is the original post:

At 07:54 AM 6/25/98 , Aaron Flin wrote:
>Dear Managers,
>
>I have a 164lx motherboard 533 mhz chip and DU4.0b with 2 Genuine DEC tulip
>ethernet cards.
>
>The tu0 is currently the only one being used. It is hooked into a 100 mb/s
>hub which is in turn hooked into a switch. It is on a network of
>Intel/Linux boxes which all have clone tulip cards (Netgear FA310tx).
>Communication between the linux boxes occur as expected as does data going
>out of the alpha. However data coming into the Alpha is extremely slow
>(whether nfs or ftp).
>
>Here is an example of speeds with ftp client run on the Alpha:
>
>ftp> get testfile
>200 PORT command successful.
>150 Opening BINARY mode data connection for testfile (19908479 bytes).
>226 Transfer complete.
>19908479 bytes received in 1.1e+02 seconds (1.8e+02 Kbytes/s)
>ftp> put testfile
>200 PORT command successful.
>150 Opening BINARY mode data connection for testfile.
>226 Transfer complete.
>19908479 bytes sent in 3.7 seconds (5.3e+03 Kbytes/s)
>
>Here's some relevant info:
>
> arfie 07:45:26 /tmp# netstat -r
> Routing tables
> Destination Gateway Flags Refs Use Interface
> Netmasks:
> Inet 255.0.0.0
> Inet 255.255.255.0
>
> Route Tree for Protocol Family 2:
> default gate-101b.sjc.abov UG 5 3108076 tu0
> 10 alpha U 1 74 tu1
> localhost localhost UH 1 0 lo0
> 207.126.101 arfie U 9 24931425 tu0
> arfie 07:45:30 /tmp# ifconfig -a
> tu0: flags=c63<UP,BROADCAST,NOTRAILERS,RUNNING,MULTICAST,SIMPLEX>
> inet 207.126.101.200 netmask ffffff00 broadcast 207.126.101.255
>ipmtu 1500
>
> tu1: flags=c63<UP,BROADCAST,NOTRAILERS,RUNNING,MULTICAST,SIMPLEX>
> inet 10.0.0.5 netmask ff000000 broadcast 10.255.255.255 ipmtu 1500
>
> sl0: flags=10<POINTOPOINT>
>
> lo0: flags=c89<UP,LOOPBACK,NOARP,MULTICAST,SIMPLEX>
> inet 127.0.0.1 netmask ff000000 ipmtu 1536
>
> ppp0: flags=10<POINTOPOINT>
>
>tu0 is also aliased with two additional ip addresses.
>
>Any ideas?

Responses centered around the idea that either the settings for the
ethernet cards were wrong (i.e. half duplex when they should be full or
vice versa) or that the fact that the hub into which the the alpha was
connected was half while the switch was believed to be full.

Turns out that neither playing with the full/half settings using "ifconfig
tu0 xxx.xxx.xxx.xxx speed x00" nor playing around with the hubs and
switches worked. Changing the speed to 200 cut me off completely (though
not immediately, which I thought was strange). Further, it turns out the
switch runs in half or full duplex or mixed just fine.

The only way I was able to get the proper speed out of the network cards
was to hook both cards up to the same hub and ifconfiged within the same
class C ip address range and ftp to myself. Ftping to any of the linux
machines or to the outside world resulted in the same problem.

Additional information:

All linux machines report something similar to this:

eth0 Link encap:10Mbps Ethernet HWaddr 00:40:05:36:CA:93
          inet addr:207.126.101.246 Bcast:207.126.101.255 Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
          RX packets:181864259 errors:0 dropped:0 overruns:0
          TX packets:153803946 errors:1 dropped:0 overruns:1
          Interrupt:10 Base address:0xe880

Transfering files via NFS and FTP between the linux boxes in both
directions works just as expected.

All traffic between the machines stay on the switch and none goes out to
the router.

The machines in total are on 3 class Cs, 2 public and 1 private. I tried
all three and this doesn't seem to play a factor.

The two ethernet cards in the Alpha are DE500-AA.

I am completely out of ideas.

Aaron


Aaron Flin
C.E.O.
Pile, Inc.
_______________________________________
| Stop Searching and Start Finding |
|______ http://www.dogpile.com/ _______|
Received on Wed Jul 29 1998 - 23:55:59 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:38 NZDT