Service manual

3-76 AlphaServer GS80/160/320 Service Manual

3.9 Dealing with a Hung System

Troubleshooting a hung system depends upon what was running at the

time of the hang. In general, these systems are designed not to hang. If

a transaction times out or forward progress is not made for some

reason, such events are considered faults and a running system should

crash.

Table 3– 12 Hung System Suggestions

Try logging in remotely and investigate what the system is doing.

Check LEDs in QBBs and power supplies and if there is something

abnormal fix it.

Check for I/O activity. If the system is in fact hung, there won’t be any.

But if the disks are shared with another system, disk activity won’t

necessarily mean much.

4. Try to get to the SRM console (OpenVMS - Ctrl/p, Tru64 UNIX - SCM halt

in). If you can, issue commands that might help you identify where the

problem lies. Try a Ctrl/x to see if you can get a new console shell.

Try to get to the SCM using the escape sequence (default -

<esc><esc>scm) and issue commands like show csb and show fru to see

if you see something odd or see old errors. Check to see if there is an alert

pending.

If the system is partitioned, the SCM halt in command will halt the

partition. Pushing the Halt button will halt the entire system.

If you can, use the SRM crash command to force a system crash for

analysis later.

8. If all else fails, fault, reset, or power cycle the system.