Alpha ES40 rebooted

[dudharkar chandra]

Alpha console environment is set to auto_action
halt. I observed that it has rebooted during night while backups were running. I feel it should come down to p0>>> SRM prompt if any issue. oerator pannel-Halt button is check for operation-it is working proper.

On analysis of error log HP found problem-Brief Description: Uncorrectable System Error Detected Suspect Multiple Correctable ECC Burst Condition,

We have refixed memory on cleaning and run memexer no problem detected.

Can anyone tell why it has not come to SRM prompt if it has any panic issue.

[Volker Halle]

Hi,

the setting of the AUTO_ACTION console environment variable is only honoured on cold-start or if the running OpenVMS system HALTs unexpectedly (i.e. HALT in kernel mode, MCHECKPAL, KRNLSTAKNV etc.), then the SRM console looks at the setting of AUTO_ACTION and either leaves the system at the console prompt (if AUTO_ACTION = HALT) or restarts or reboots the system. In case of a restart attempt after a halt, OpenVMS will write a restart bugcheck and reboot, if the system parameter BUGREBOOT is set to 1 (default value).

After a normal system crash, the system will also reboot with the default setting of BUGREBOOT=1.

So in this case, the system most likely crashed and booted automatically. For information about the crash, look at the crash history or at the CLUE file for that crash:

$ TYPE CLUE$HISTORY
-> reports 1 line per crash (132-columns !)

$ TYPE CLUE$COLLECT:CLUE$node_ddmmyy_hhmm.LIS
-> crash summary information, you can add this file as an ASCII attachment and I'll have a look at that crash problem symptoms.

Volker.

[Hoff]

AUTO_ACTION set to RESTART is the correct setting for most OpenVMS production environments; that's how you get crashdumps for crashes written for the cases that don't otherwise directly trigger the crashdump.

AUTO_ACTION set to HALT is typically a less desirable setting, save for cases when the system environment is unable to withstand an automatic reboot, or when sufficient spare capacity exists in a cluster or grid-like configuration to take up the load from the lost server.

An AlphaServer ES40 is comparatively slow and wasteful (cooling, space, power) these days, too. Most any of the Integrity series will easily outrun it, which means you can downsize.

[Bob Blunt]

Its very likely that IF your system "just rebooted" you'll have a much better chance to get to the bottom of the problem (if any) if you have output from the console available for review. There are many things that can cause an Alpha to "restart" but unless you've got a crash to examine you might as well say your shoes are on the wrong feet.

SO if you are using a graphics interface on your ES40 as the console device you might be well served to redirect console activity to the serial port. Then connect the system's console port to either a hardcopy terminal or to some device or system capable of saving the console output. A crash dump or CLUE output is a great place to start provided the issue has been captured in a dump but there are still a good number of causes for a Alpha to restart that don't always cause a crash. In addition you can also access the RMC port on the ES40 which can give very limited information about failures that caused a restart (temperature or power issues for instance).

bob