Sun Fire E25K Server
 
Sun Fire 15K Server
 
Sun Fire 12K Server
 
Sun Fire E20K Server
 

PLA-Support>Sun Systems>SPARC>Enterprise>SN-SPARC: SF-Exxk
 
_Old GCS Categories>Sun Microsystems>Servers>High-End Servers
 

Applies to:

Symptoms

Problem Description:

I sometimes see the following errors when trying to boot my domain:

Sun Fire[TM] 15000, using IOSRAM based Console Copyright 1998-2002 Sun Microsystems, Inc. All rights reserved. OpenBoot 4.7.4, 8192 MB memory installed, Serial #44567796. Ethernet address 0:0:be:a8:c:f4, Host ID: 82a80cf4. Rebooting with command: forced-no-boot NOTICE: forced-no-boot: OBP has been requested by SC to not boot. NOTICE: forced-no-boot: See log files on SC for more information.

Cause

Why does this happen?

DSMD will initiate automatic system recovery (ASR) in order to bring a domain back to a running state. This may happen during normal operations, such as a reboot or init 6, or it may happen after an error, such as recovering a hung domain. DSMD will give up its recovery efforts if it encounters a dstop in the domain or if it fails more than 6 times to run hpost and start OBP. If it fails in this way, DSMD will set the next boot path for the domain to "forced-no-boot" in order to signal OBP that there have been repeated errors while trying to recover the domain. This is SMS's way of requiring administrator intervention to determine the problem with the system.

Solution

How do I investigate the cause of this problem?

Check the platform logs for errors during the timeframe of the attempted ASR. Messages like this indicate that hpost is failing:

Feb 25 17:43:41 2003 sc0 dsmd[5959]-C(): [5304 3719971132639644 ERR SysControl.cc 1769] Domain failed by hpost: ecode=39
Feb 25 17:43:41 2003 sc0 dsmd[5959]-C(): [2550 3719971236215386 NOTICE Domains.cc 421] Max limit of 6 recovery attempts has been reached.

The following are some error messages that may indicate failures in starting OBP.

Note that these errors do not include the normal SMS timestamp and logging information.

• Console bus open failed, can not continue
• Unable to read GDCD, can not continue: ecode=%d
• CPU list construction failed: ecode=%d
• Master CPU selection failed: ecode=%d
• Master CPU memory map failed: ecode=%d
• Master CPU OpenBoot PROM download failed: ecode=%d
• Slave CPU memory map failed: ecode=%d
• Slave CPU jump failed: ecode=%d
• Master CPU jump failed: ecode=%d 
• OBP has been requested by SC to not boot

In what way is this condition cleared?

If the underlying error is resolved, use setkeyswitch -d <domain> "off" then "on" and the forced-no-boot flag will be cleared once the domain boots successfully.

If the error can not be cleared please collect an explorer from the main system controller and open a service order with your service provider.

As a last resort run setdefaults(1M) on the domain.

The domain must not be active and setkeyswitch must be set to off. This command removes all platform configuration database (pcd) entries (including the domain's console and messages logs), except network information and, optionally, NVRAM and bootparameter data. When asked if you want to remove the NVRAM and bootparameter data, answer no.

When setdefaults completes, re-configure the domain using setupplatform and addboard commands.

Can I manually edit the nextbootpath file to remove the forced-no-boot path setting?

No, do not edit the SMS file by hand. It is changed automatically by SMS. Instead, work on diagnosing the problem that is keeping the domain from booting properly, and this problem will be cleared up automatically

forced-no-reboot, setdefaults, recovery attempts
Previously Published As 70200

This solution has no attachment