SUMMARY: SPARCstation LX and 10baseT

From: Mike Rembis 66520 (ebumfr@ebu.ericsson.se)
Date: Tue Mar 02 1993 - 18:48:10 CST


Last week I posted the following:

| From ebumfr Wed Feb 24 15:54:40 1993
| To: sun-managers%eecs.nwu.edu@ericom.ericsson.se
| Subject: SPARCstation LX and 10baseT
|
| We have taken delivery of 15 LX's. These will be our first Solaris 2.1
| machines on our network. Our entire network is twisted-pair. This is good,
| since the new LX's come with TP-ethernet port by default. Unfortunately
| I cannot make any of them communicate to any UTP box that we have.
|
| We have Luxcom and Alantec UTP hubs. All are running various Suns and PCs
| using Milan AUI-UTP converters. All are working properly.
|
| The little hardware guide that comes with LX's makes reference to
| the EEPROM variable "tpe-link-test?". It makes note of the fact that
| "according to UTP standards", machines "should" do Link Status test.
| All of the Milan transceivers have this enabled. The Luxcom and PowerHUB
| have LED's that show Link Status.
|
| When I connect an LX...I get a lit LED indicating link status. If I try to
| ping (or any other net accessing command) I get:
|
| le0: no carrier or disable link tests?
|
| If I set "tpe-link-test?" to false and try the net commands again, such as,
| ping...it just returns...
|
| no answer from .......
|
| Has anyone seen this behaviour? Is there something that I have missed, possibly
| due to "volimnous" documentation that comes with the LXs? I have tried many
| different cables.... all tested on our UTP tester. I have plugged into
| known working ports. Help!
|

The responses ranged from assuring that I had correctly followed the 10baseT
spec for cable wiring to bad ports on the Sun.

It turned out to be a couple of problems. LX number 1 does indeed have a bad
10baseT port on it. Back to the factory for repair. LX number 2 has a bad disk
and would not boot properly. LX numbers 3 and 4 work once the machine is
completely up, loaded and sitting at login prompt, however all through the
JumpStart process, that same error message was frequently displayed. I have
been advised by Sun that this is a known bug and will be repaired in the next
release of Solaris... (2.2).

Responses follow:
                                %%% %%%

|
| Did you use the cables that came with the LX?
| I have heard about people having problems that disappeared when using Sun's
| cables.
|

                                %%% %%%

|
| You no doubt checked this, so apologies if I am off track,
| if you run etherfind on the LX, do you see packets? Are
| the LXs set to the correct broadcast address for your
| network? If you check ifconfig -a as root, is the
| interface configured correctly?
|
                                %%% %%%

|
| The TP-ethernet on the LX is 10BaseT. If your network does not
| meet true 10BaseT standards, you may have some problems. The may
| be a way to disable the Link status (via a CPU board jumper or maybe
| EEPROM setting) that may do the trick. But you may have to get Sun's
| input on this.
|

                                %%% %%%

| How did you name your systems? I first named mine "name.tamu.edu" and it could
| not even ping itself. I had to reload and name the system "name". I run a
| twisted pair cabletron box and i get the error message you get on bootup but
| it pings ok.
|
| I have my "tpe-link-test" set to true.

                                %%% %%%

|
| This may be a little off target, but make sure that you are only
| connecting two pairs to the LX. We've noticed that SS-10s and
| possibly classics put some kind of signals on the extra two pairs
| of the RJ-45. Due to an accidental miswiring, this brought one of
| our hubs to a crawl. So if your hub also uses the extra two pairs
| for something, the machines could be fighting each other.
|
                                %%% %%%

|
|
| Well, Mike, we just received 2 SPARCclassics and had a similar problem.
| I found that I did not have to change the tpe-link-test? value in the EEPROM,
| but when the network person wired them up, they did not work. It turned
| out the problem was when we had to make a cable to convert the 8 pin RJ45
| connecter to a 4 pin RJ11 (I think). I asked my local FE for a pinout of
| the RJ45 port and this is what he sent:
|
| PIN SIGNAL PIN SIGNAL
| |-----|--------|-----|--------|
| | 1 | TPO+ | 5 | n/c | O = RD
| | 2 | TPO- | 6 | TPI- |
| | 3 | TPI+ | 7 | n/c | I = TD
| | 4 | n/c | 8 | n/c |
| |-----|--------|-----|--------|
|
| Check this with the pinout on your UTP hubs and see if there is a difference.
|

                                %%% %%%

| I have encountered cases in the past, using various vandors' equipment,
| where I had to build a 10baseT cable that had pairs 2&3 swapped. So, if
| you havent tried this, I suggest that you swap pairs 2&3 in your patch cable.
| Assuming a standard RJ45, it would be something like this: (note that for
| an RJ45 plug, of you hold the connector with the tab down, and cable
| opening facing you, that pin 1 is on the left). Also note that the pairs
| are assigned like this:
|
| Pair RJ45 pins
| -------------------------
| 1 4 & 5
| 2 1 & 2
| 3 3 & 6
| 4 7 & 8
|
| A "straight through" cable will be wired such that each pin is connected
| to the same pin (number) on the opposite connector, and each pair is
| actually a twisted pair. Sometimes, depending on the vendor, you will
| need to connect your machines using a straight through cable. Other
| times, you may need to build (or buy) a crossed cable. This is what
| it looks like:
|
| RJ45 pin RJ45 pin
| --------------------------
| 1 3
| 2 6
| 3 1
| 6 2
|
| Note that since 10baseT only uses paire 2 & 3, you only need connect
| those pairs. MAKE SURE THEY ARE PAIRS, AND NOT SIMPLY FOUR DISCRETE WIRES!
| Convince yourself that we have crossed pairs 2 and three...
|

                                %%% %%%

|
|I had the same problem with the first LX I setup on the network with my Twisted
| Pair Hub. I found that the SUN TP connector was at fault. If I pushed the RJ45
| plug in to the SUN LX twisted pair port to were it snapped in place I lost
| carrier, but if I unsnapped it and pulled it out alittle it worked. I have a
| call into Sun about the problem. The RJ45 connectors we have are made by
| AMP. You may be having the same problem.... Good Luck !!
|
|

                                %%% %%%

Thanks to:
-------------------------------------------
birger@vest.sdata.no (Birger A. Wathne)
kuhn@math.harvard.edu (Robert M. Kuhn)
Bill Maggio <bmaggio@dcs.bellcore.com>
fred@johnd.tamu.edu (Fred Seals 5-2298)
Richard Elling <Richard.Elling@eng.auburn.edu>
scheller@c3ot.saic.com (Mark J. Scheller)
Rick Kruse <rlk@ncs.net.utulsa.edu>
todds@CFSMO.Honeywell.COM (Todd Sigurdson)
danielle@systems.caltech.edu (Danielle Sanine)
peffley@endurance.nrlssc.navy.mil (Monty Peffley)

---
Mike Rembis       	 	       | Internet: mike.rembis@ebu.ericsson.se
Ericsson Business Communications, Inc. | UUCP:   uunet!ericsson.com!mike.rembis
````````````````Reality is in the Eye of the Beholder'''''''''''''''



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:32 CDT