NAME

crashconf − enable/disable/configure crash dumps

SYNOPSIS

/usr/sbin/crashconf [−L |−M | −S | −s n] [−a] special_de vice
/usr/sbin/crashconf −L | −M | −S | −s n

DESCRIPTION

The crashconf command with arguments configures the crash dump system to take a crash dump to the special_device if the operating system should panic.  The special_device must follow the conventions for device specific files as given in intro(7). If a disk slice, it must also be tagged with V_SWAP. 

If the −a option is given, the crash dump system will be in automatic mode and will not prompt the user before beginning the crash dump when the system panics; otherwise, the crash dump system will be in manual mode and prompts the user to prepare the crash dump device prior to beginning the crash dump.  Manual mode also allows the user to attempt to recover if a failure occurs if the device supports recovery. 

Administrators of systems with larger memories or limited disk space for storing crash dumps may wish to use compressed crash dumps.  The -L (large), -M (medium), and -S (small) options select one of 3 predefined dump sizes.  Dumps configured with the -M or -S flags are referred to as compressed dumps and must be uncompressed by ldsysdump(1M) before standard analysis tools may be used.  The size defaults to -L. 

The file /etc/master.d/dump defines the 3 dump sizes for each system and, by default, the large dump is defined to be the entire main memory of the machine, medium is the entire kernel virtual space, and small is the entire kernel virtual space less the segmap area.  Unless problem analysis requires looking into processes’ address spaces, the medium or small dumps usually suffice; however, the system administrator must determine the proper dump size for each machine based on local needs. 

The -s n selects a dump size locally defined in /etc/master.d/dump (see). 

The dump size may be set without specifying a special_device but in order to do so, dumps must already be enabled. 

The crashconf command with no arguments disables the crash dump system, reseting the default dump size. 

The buildsys(1M) manpage contains information regarding preservation of the bootable unix corresponding to a crash dump. 

EXAMPLES

This example shows configuration of a disk slice for meduim-sized crashdumps requiring no user intervention. 

$crashconf -a -M /dev/rdsk/m328_c0d0s1
$

FILES

/etc/init.d/CRASHDUMPscript to configure crash dumps during boot
/dev/rdsk/prefix_cndnsndevice used for disk slice
/dev/rmt/prefix_cndndevice used for tapes
/etc/master.d/dumpdefault dump configurations

SEE ALSO

buildsys(1M), crash(1M), fmthard(1M), ldsysdump(1M), sysm68k(2), sysm88k(2), "Disk and Tape Management" section of the System Administrator’s Guide. 

DIAGNOSTICS

If the crash dump system can not be configured to the device special_device for some reason, crashconf will print explanatory messages and the exit value will be 1.  The exit value is 0 upon success. 

NOTES

The special_device must be large enough to accommodate the selected dump size or data may be lost. 

If the special_device also serves as a swap slice, it may need to be significantly larger than the amount of physical memory to keep from corrupting the crash dump image after reboot.