E220R Freezes and have to type 'go' at ok prompt to resume

zoonsf
zoonsf used Ask the Experts™
on
I've got a fresh install of Solaris 9 on an E220R that seemed to be running fine until I'd leave it for a while (~1 hour) and then come back to find it not responding.

After going in through the serial port I only need to type 'go' a couple times at the ok prompt and everything is running fine again...  for a while that is.

First I expected it had to do with the power management features which I think I've completely disabled.

1.  Editing the /etc/power.conf file and changing it to disable autoshutdown didn't do anything.

2.  Renaming rc2.d/S85power didn't seem to do anything either.

3.  Uninstalling the power packages didn't help.  (pkginfo | grep -i power returns nothing)

4.  Multiple hard reboots in between the processes seems to do nothing.

How can I get this machine running so I didn't wast a big chunk of money?  I'm new to Solaris but would expect there's some configuration setting that I'm not seeing.  I've looked at many logs in /var but haven't seen anything to help me out.

Help!
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®

Commented:
Thsi is strange, the fact that you can type "go" and get the system up indicates that its not a powerdown that has
kicked in.
After a "sys-suspend" you need to start it up with the power button.

Do you see a promt on your terminal when you connect to the serialport ?

Is there anything in the systemlog ? (/var/adm/messages*)

I think that you want see anything in the log which points to that the system somehow gets a "break" signal and drops to the prom-monitor

I'll check around

Author

Commented:
When I connect to it serial I'm at the 'ok' prompt.  Type 'go' and then it's up without a hitch.

There's nothing funny in /var/adm/messages*  Actually I didn't see much of anything since nobody can even get to the machine right now.

Commented:
The fact that the "ok" prompt is visible shows that the system has gone to the prom-level.

At this point Solaris has died and the ok prompt is displayed by the bootprom-program.
This will happen if you hit the STOP and A keys at a Sun keyboard or the BREAK key on a terminal keyboard.
( Note that key in locked position does not prevent BREAK
on the 220R)

The result is a dead stop of the operationgsystem but as long as the system isnt changed in any way, all registers, pointers, pc and stacks still has the right contents.
So thats why you can type "go" and the system is up and running again.

So, this is hardware reset, if it had been triggered by a command it would have been in the logs.

Is the terminal always attached to the serial port ?

If not leave it there and see what comes up when the systems "goes to sleep" (I guess you dont have a graphic console on this system, right ?)

Could it be powersupply related ?
I had a really wierd similar problem once, a system died almost every day at
4 am. Turned out that the power dropped when the ventilationfans started.

Check the terminal and come back.
CompTIA Security+

Learn the essential functions of CompTIA Security+, which establishes the core knowledge required of any cybersecurity role and leads professionals into intermediate-level cybersecurity jobs.

Author

Commented:
Hmm...  There's one thing that I'm going to have to try.  I've never actually seen it sleep when I was there.  Last night I had it run over an hour when I just let it run while watching the serial comm and nothing happened.  I do have a graphics card in the server and have hooked a monitor up to it but don't have a sun compatible keyboard.

Not thinking about it I went to work and every time I get into the office the darn thing isn't responding.  I come home and I have to do a 'go' to get it running again.

What I'll have to check this evening is could it possibly be that when I turn of my Windows machine I've been using to connect to the server with, it sends out a signal that's getting interpreted as a STOP signal?  Maybe the powering off of my machine while it's still connected with the null modem cable to the Sun box is stopping it.  I hope so.

Author

Commented:
Hmm...  There's one thing that I'm going to have to try.  I've never actually seen it sleep when I was there.  Last night I had it run over an hour when I just let it run while watching the serial comm and nothing happened.  I do have a graphics card in the server and have hooked a monitor up to it but don't have a sun compatible keyboard.

Not thinking about it I went to work and every time I get into the office the darn thing isn't responding.  I come home and I have to do a 'go' to get it running again.

What I'll have to check this evening is could it possibly be that when I turn of my Windows machine I've been using to connect to the server with, it sends out a signal that's getting interpreted as a STOP signal?  Maybe the powering off of my machine while it's still connected with the null modem cable to the Sun box is stopping it.  I hope so.

Author

Commented:
Hmm...  There's one thing that I'm going to have to try.  I've never actually seen it sleep when I was there.  Last night I had it run over an hour when I just let it run while watching the serial comm and nothing happened.  I do have a graphics card in the server and have hooked a monitor up to it but don't have a sun compatible keyboard.

Not thinking about it I went to work and every time I get into the office the darn thing isn't responding.  I come home and I have to do a 'go' to get it running again.

What I'll have to check this evening is could it possibly be that when I turn of my Windows machine I've been using to connect to the server with, it sends out a signal that's getting interpreted as a STOP signal?  Maybe the powering off of my machine while it's still connected with the null modem cable to the Sun box is stopping it.  I hope so.

Author

Commented:
Hmm...  Looks like if you hit refresh after a post, it adds your comment again.  Sorry!
Top Expert 2005
Commented:
Turning off the windows box that's running the terminal emulator for the console (or in some cases closing HyperTerm or whatever) is equivalent to sending a BREAK to the serial port. And a BREAK is the tty equivalent of Stop-A on a Sun keyboard, which halts the processor.

To prevent the system from halting you have two choices. One would be to never power off or disconnect the windows box that is running the terminal emulator (and don't ever close the terminal emulator application). The other would be to get any of the RS232 console switchers and interpose that device between the Sun and the terminal. A console switch knows about not placing the serial port in a BREAK condition, so even when not looking at that server it won't leave the serial line in a conditon that will cause the processor to halt.

Commented:
Welcome to the wonderful world of windows.

So pull the serialcable before shutting down the toy,
that should help.
Top Expert 2005

Commented:
No, don't pull the serial cable. That can also result in the port seeing a BREAK, which will shutdown the server.

Commented:
Oh, please.
Pulling the cable will not generate a break.
This problem existed on the old Sunsystems for years and the problem comes when you put the cable back again.
It is the same as the keyboards, pulling the cable is ok, putting it back takes the system down.

There is actually three solutions for this:
1. By a cheap termainal
or
2. modify your serialcable with a "pullup resistor" which will keep the signal high.
or
3. When you know that your systems goes down as you attach to serial a, just type "go" and downtime will be a couple of secs. ( not the best solution but it works)

I would recommend a terminal, just remeber to not switch it off.

Author

Commented:
Yep, this was the problem, but pulling the cable before turning off the attached terminal doesn't hurt and was the solution.

Commented:
pulling the cable CAN generate a break.

the solution is to set "KEYBOARD_ABORT=alternate" in your /etc/default/kbd file. (or set to disable) When you edit the file you'll see a comment on this at the bottom of the file.

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial