Back to the main page

Managing Solaris system crash

Why I want to do this

Because when a system crashes you want to have saved crash dump file in order to troubleshoot an issue.

What can happen - see below:

How to manage system crash dump

Use the command dumpadm to configure (make sure service svc:/system/dumpadm:default is enabled): Note: I've been seeing people use swap as dump device. In such setup, the swap is under control of SVM and this is recommended. But in this case, Sun documentation advises to keep dump device separate.

Syntax - see man page for details

dumpadm -c content -d dump_device -m min_free_disk_space -n -s savecore_dir

-c = kernel | all |curproc
-d = must be block device, like /dev/dsk/c1t0d0s4
-m = something like 100m
-n = don't run 'savecore' when system reboots
-s = specify alternate dir for storing crash dump file, the default is /var/crash/hostname

What now after having crash dump ?

I guess you'll send to support provider, or you can examine yourself using 'mdb' tool, something like:

mdb -k unix.0

> ::status
... see some output (note: 2 x : before status)
>::system
... see some output
exit with Ctrl+D
Back to the main page