Duplex setting can cause system reboot?

[Aminur Rizal]

I have problems with my server which reboots frequently for not known reasons.
Reading from core dump at /var/adm/crash it says that some errors with the duplex setting in one of the machines connected with the server.

Here are the partial output of # dmesg command
btlan3: NOTE: Link Partner does not support Auto Negotiation at 8/0/20/0
btlan3: NOTE: 10Base-TX Half-Duplex started at 8/0/20/0....


This is my lanscan output.
Hardware Station Crd Hardware Net-Interface NM MAC HP DLPI Mjr
Path Address In# State NameUnit State ID Type Support Num
8/0/20/0 0x0060B0EA446C 0 UP lan0 UP 4 ETHER Yes 75



Can this cause the reboot? The card is running 100Mbps half duplex but I believe that the hub connected is running 10Mbps.
Hub details
Hub HP-J3128A Model AdvanceStack Hub 8e

[T G Manikandan]

Please check the latest patches installed on your system.

PHNE_22244 1.0 Built-in PCI 100BASE-T patch
PHNE_22245 1.0 PCI 100B(Tx/Fx)cumulative patch
PHNE_22728 1.0 4 port PCI 100BASE-T cumulative patch

Also check the MAC address entry in the /etc/rc.config.d/ directory for the lan card specific file .(in your case I think it is hpbase100conf
and the MAC address (ouput got from lanscan command).


If your hub is running at 10Mbps then please set the speed of the card to 10Mbps so that there is not mismatch of speed.
Also you can avoid the duplex mismatch for both ends.

Thanks

[Stefan Farrelly]

Your server is not supposed to crash from getting the duplex/speed setting wrong, but its certainly possible.

The solution is to ensure your lancard and the hub/switch it connects to are the same speed. If it doesnt support autonegotation then set your card AND switch/hub to the same setting (10HD or whatever) to stop conflicts. This is normal practice.

[William King_2 · Accepted solution]

As noted from the last reply. Unless your
switch does auto negotiation and the card on
your system does auto. You should configure
the card to run 10hd. And in your case the Advancestack hub only does 10hd.

Also note that the Procurve switches had issues
with auto negotiation, so update to the latest
Firmware revision C.09.09

[rick jones]

that message about the link partner not supporting auto-negotiation is "normal" and the chances of it being related to your system's crashes are so remote as to be virtually impossible.

The message means precisely what it says - that the remote does not do autonegotiation. That may be "normal" if the remote is just some old 10 Mbit/s kit, or it could mean you have a network config problem if the remote is a switch port hardcoded to full-duplex. However, duplex-mismatch does not lead to system panic.

The thing to look at in the crash dump is the stack trace. Better still, fire-up your support contract and submit the dumps to the RC and have them take a look.

[Emil Velez_2]

Is this system a MC/ServiceGuard system. If he lan card fails and there is no backup that could be a reason for panicing but I agree that just a lan card should not panic the system

[Cheryl Griffin]

Bad HW (meaning bad or not installed correctly) could make the system panic whereas a duplex setting wouldn't normally cause a system to panic.

Check the NIC and HW, the cable, etc.

Make sure that you have the right patches installed for btlan3.

PHNE_25580 (pci 100BT lan cumulative) and it's dependencies. This patch is for s800 11.00.

If they are already installed, make sure they're configured:
# swlist -l fileset -a state |grep -v \# | grep -v config

Cheryl

[Hai Nguyen_1]

Aminur,

I recommend that you first try to set the speed and duplex on your HP box the same as the hub's it is connected to. The speed, duplex mismatches within the same LAN generally cause a lot of network issues.

Hai

[Santosh Nair_1]

Rick (the network GOD :-) is right. The problem most likely is NOT the network card settings.

You're best bet would be to run the crash dumps through Q4 and send the results to HP. They should be able to pinpoint the problem.

-Santosh

[Aminur Rizal]

Thanks a lot you guys for the response.

What I did so far is isolating the HP Hub, and connect directly to the core hub.Its been like 5 days now, and no system reboot is happening again.

I doubt if there is a problem which need patches as I have another machine running the same hardware and software but getting no such problem. (Connected directly to the core hub).

rick jones,
Are you sure? Because this error was the first message that appear on the screen when the system rebooted.

I found this article very useful explaining the duplex issue.
http://searchnetworking.techtarget.com/tip/1,289483,sid7_gci802539,00.html

Thanks again to all.