Configuring Your System for Crash Dumps

Language:

This section describes the tasks for managing crash dump procedures for your system.

Keep the following points in mind when you are working with system crash information:

You must assume the root role to access and manage system crash information. For more information, see Using Your Assigned Administrative Rights in Securing Users and Processes in Oracle Solaris 11.4.
Starting with the Oracle Solaris 11.3 release, when a system crashes, a copy of its physical memory is held in the RAM until the system reboots, then is written to the file system. This process is called a deferred dump. If a deferred dump is not possible, the physical memory is written to a dump device during a crash. For more information, see Crash Files Deferred Until After Reboot in Oracle Solaris 11.3 in Troubleshooting System Administration Issues in Oracle Solaris 11.4.
Do not disable the option of saving system crash dump files on the system. System crash dump files provide an invaluable way to determine what is causing the system to crash.
Dedicated ZFS volumes are used for swap and dump areas. For instructions, see Managing ZFS Swap and Dump Devices in Managing ZFS File Systems in Oracle Solaris 11.4.

For more information about managing system crash dump, see dumpadm(8) man page.

Displaying the Current Crash Dump Configuration

To display the current crash dump configuration, assume the root role and issue the dumpadm command with no arguments.

$ dumpadm
 Dump content: kernel with ZFS metadata
       Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash
  Savecore enabled: yes
   Save compressed: on

This example output shows the following configuration:

The dump content is kernel memory pages with ZFS metadata
Kernel memory will either be held in memory until a reboot or dumped on a dedicated dump device, /dev/zvol/dsk/rpool/dump
System crash dump files will be saved in the /var/crash directory
Saving crash dump files is enabled
Crash dump files are saved in compressed format

Modifying the Configuration for Crash Dumps

To modify the crash dump configuration, assume the root role and use the dumpadm command. Depending on the kernel memory and other internal information, the dumpadm command establishes a dump device to accommodate a dump file. The size of the dump device should be big enough to store the dump file, otherwise dumpadm displays an error message and the operation fails. For more information about managing the configuration of crash dumps using dumpadm, see the dumpadm(8) man page.

Example 1 Modifying a Crash Dump Configuration

In this example, all memory is dumped to the dedicated dump device, /dev/zvol/dsk/rpool/dump. 10% of the file system space must be available as minimum free space after the crash dump files are saved.

$ dumpadm
      Dump content: kernel with ZFS metadata
       Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash
  Savecore enabled: yes
   Save compressed: on

$ dumpadm -c all -d /dev/zvol/dsk/rpool/dump -m 10%
       Dump content: all pages
       Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash (minfree = 5935131KB)
  Savecore enabled: yes
   Save compressed: on

$ dumpadm -n
       Dump content: all pages
       Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash (minfree = 5935131KB)
  Savecore enabled: no          
   Save compressed: on

$ dumpadm -y
       Dump content: all pages
       Dump device: /dev/zvol/dsk/rpool/dump(dedicated)             
Savecore directory: /var/crash (minfree = 5935131KB)
      Savecore enabled: yes         
        Save compressed: on

Example 2 Disabling the Saving of Crash Dumps

This example shows how to disable the saving of crash dumps on your system.

$ dumpadm -n
       Dump content: all pages
       Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash (minfree = 5697105KB)
  Savecore enabled: no
   Save compressed: on

Caution - Do not disable the saving of crash dumps. Crash dumps provide a way to determine what causes your system to crash.

Example 3 Enabling the Saving of Crash Dumps

This example shows how to enable the saving of crash dumps on your system.

$ dumpadm -y
      Dump content: all pages
      Dump device: /dev/zvol/dsk/rpool/dump (dedicated)
Savecore directory: /var/crash (minfree = 5697105KB)
  Savecore enabled: yes
   Save compressed: on

Comparison of Different Dump Devices

The following table describes the differences between various dump devices. Note these differences when you set up a dump device.

Table 1 Comparison of Different Dump Devices

Features	`swap`	`zvol`	`none`
Size	Size of the dump device is proportional to the size of physical memory. It also depends on the settings of the `dumpadm` command and the dump content. You can use the `–e` option of the `dumpadm` command to get an estimate of the space required.	Size of the dump device is proportional to the size of physical memory. It also depends on the settings of the `dumpadm` command and the dump content. You can use the `–e` option of the `dumpadm` command to get an estimate of the space required.	Not applicable.
Resizing capability	You cannot resize the dump device easily.	You can resize the dump device. For more information, see Adjusting the Sizes of ZFS Swap and Dump Devices in Managing ZFS File Systems in Oracle Solaris 11.4 in Managing ZFS File Systems in Oracle Solaris 11.4	Not applicable.
Minimum free space required	You need sufficient space in the crash directory and on the volume which will store the dump data.	You need sufficient space in the crash directory and on the volume which will store the dump data.	You need sufficient space in the crash directory.
Risk factors	The crash dump can be overwritten by swapping activity.	If the physical memory of the system is upgraded without resizing the `zvol`, then the dump device can be too small to hold the dump data in case of a system panic.	If the deferred dump is inactive, then crash dump will not be captured in case of a panic. Deferred dump can be inactive if your system has low installed physical memory or if the firmware does not support it.

Note - A ZFS based device cannot be used as a dump device.

The estimate provided by the dumpadm -e command depends on the currently running system and dump configuration. Therefore, you should run this command once your system is properly warmed up and the final dump configuration is in place. Otherwise the estimate might be wrong by a large margin.

For more information about the none dump device and the nodump attribute, see the ai_manifest(5) man page.