UNIX for Dummies Questions & Answers

SunOS 5.6

I have a remote test system (test T1/T3, etc) that runs SunOS 5.6. I rebooted the system for no other reason than that it had been up for over 300 days. It failed to boot completely.

I have an ok prompt and when I type boot, I see the following messages..
Boot device: /iommu/sbus/espdma@5,8400000/esp@5,8800000/sd@0,0 File and args: -r
Short read. 0x2000 chars read
Short read. 0x2000 chars read
Short read. 0x2000 chars read
Short read. 0x2000 chars read
|
Watchdog Reset

There are some diagnostic tools that pointed me to the problem:
CS4231 ASIC SelfTest Passed.
Error: DVMA failure internal loopback
expected 2 received 0
Error: DVMA failure internal loopback
expected 4 received 0
Error: DVMA failure internal loopback
expected 8 received 0
Error: DVMA failure internal loopback
expected 10 received 0
Error: DVMA failure internal loopback
expected 20 received 0
Error: DVMA failure internal loopback
expected 40 received 0
Error: DVMA failure internal loopback
expected 80 received 0
Selftest failed. Return code = 7
ok

Based on this output, I used another diag tool to check the memory map and got the following results:
Virtual : 0000.0002
Context : @ 0.01ff.f000 001f.eec1 # 0
Region : @ 0.01fe.ec00 001f.ee71
Segment : @ 0.01fe.e700 001f.ee61
Page : @ 0.01fe.e600 0000.001c Invalid
Stack Underflow

So it appears to me that the system can't find the pages it's looking for. I'm in very short supply of CPUs so will probably attempt to reload the OS/software. This is a bit of a hassle because it's a remote site.

What I'm looking for is opinions on what the most likely cause of the problem is and if there is some way to possibly fix this without a CPU replacement or OS/software reload. I've already reseated the CPU and I get the same results.

Thanks for you time!

OS reload probably won't help you at all.

Thanks RTM. I appreciate the feedback and the education on the watchdog timer. I will attempt the reload first keeping in mind that I will likely need to dig up a CPU.

Why would you attempt the reload at all? Did you note the last line in my other post?

"This could be a watchdog reset on a system with its obprom "watchdog-reboot?"
flag set to true. "

I suppose I don't know enough to see how this would make a reload useless. I basically want to try everything possible before changing the CPU. I only have one left.

If you could expound on what I'm missing (why the reload may be pointless) I would appreciate it.

Watchdog resets are usually hardware related.

Does this server have one or more than one cpu? If more than one, remove one and see if the issue goes away. If it doesn't (still gets a watchdog reset), put that cpu back in, and pull the other one. If the issue goes away, you could always just run on one cpu (if they are so hard to come by) or replace the bad one. If the watchdog reset doesn't go away, then it could be your memory. But attempting to reload the OS will not make a hardware issue go away. You will still have the watchdog reset issue, possibly mess up what is on the hard drive already, and have wasted your time.

The second part of the information I posted states:
The system leaves no messages when it reboots.

This could be a watchdog reset on a system with its obprom "watchdog-reboot?" flag set to true.

This means that if at the ok prompt, if you do a printenv, you will see all the different parameters. If parameter watchdog-reset IS set to true, there may not be a message as to why it had a watchdog-reset. Set it to false for more info.

If you have more information about which CPU or possibly which memory board is bad, post the info.

Thanks for the explanation. The watchdog-reboot is set to false currently. There is only one CPU in this box. I realize from what you've posted that reloading the software is probably an exercise in futility, but I have a good backup of the config and I really do want to avoid replacing the CPU if at all possible......so I'm going to try the reload just in case.

If the reload fails, I'll replace the CPU. Either way, I'll let you know what ultimately fixes the problem. Thanks again for the input.