Understanding core files

About core files

When a hardware or software failure causes the storage system to panic, the system creates a core file that technical support can use to troubleshoot the problem. The storage system stores the core file in the /etc/crash directory on the root volume.

What the savecore command does

The savecore command, which is included in the default /etc/rc file on the root volume, performs the following tasks:

Produces a core.n.nz file. The n in the file name is a number. The string nz indicates that the file is compressed.
Displays a message on the system console.
Logs a message in /etc/messages on the root volume.

Core dump writing

A core dump file contains the contents of memory and NVRAM. Core dumps are written over reserved sections of any working disk owned by the local storage system. When a core dump is created, it is stored in uncompressed format if sufficient space is available; otherwise, it is stored in compressed format. If there is insufficient space to store a complete core dump in compressed format, the core dump is canceled.

Note
If the failed storage system belongs to a cluster and options cf.takeover.on_panic is enabled, a core dump file is written to a spare disk on that system.

Core dump files are not compatible between Data ONTAP releases because where the core starts on disks depends on the release. Because of this incompatibility, Data ONTAP might fail to find a core dump file dumped by another release.

Options that control core dump file creation

There are two options you can use to control core dump file creation, as listed in the following table:

Option	Description	Default value
`coredump.dump.attempts`	Controls how many attempts are made to create a core dump file.	`2`
`coredump.metadata_only`	Specifies whether the buffers containing user data are dumped.	`on` (user data buffers are not dumped)

Option

Description

Default value

coredump.dump.attempts

Controls how many attempts are made to create a core dump file.

2

coredump.metadata_only

Specifies whether the buffers containing user data are dumped.

on

(user data buffers are not dumped)

For more information about these options, see the na_options(1) man page.

Automatic technical support notification

Your storage system sends e-mail automatically to technical support upon each system reboot, if the AutoSupport feature is enabled and configured correctly. Technical support uses the AutoSupport message and the core file to troubleshoot the problem.

Note
If you have disabled AutoSupport e-mail, you should contact technical support when your system creates a core file.