Working With Core Dumps

9 Working With Core Dumps

Core dumps contain crash information for userspace applications and services running on Oracle Linux. They can be generated on demand by using a debugger, or the systemd-coredump service can be configured to generate them automatically in the event of a process stopping prematurely.

Core dumps contain a log summary of the crash event that typically includes the process ID, owner, termination signal, and a stack trace. For more information, see the systemd-coredump(8) manual pages.

The coredumpctl command can be used to review core dumps that have been written to the system journal or saved as a file. For more information, see the coredumpctl(1) manual pages.

To learn more about Systemd, the daemon that initializes the system and manages running services in Oracle Linux, see Oracle Linux 8: Managing Core System Configuration.

Enabling Core Dumps

Core dumps aren't enabled by default, so you must configure Systemd to generate them.

Create the /etc/systemd/system.conf.d/10-enable-coredumps.conf configuration file and add the following content:
```
DumpCore=yes
DefaultLimitCORE=infinity
```
Restart the systemd daemon to apply the change without restarting Oracle Linux:
```
sudo systemctl daemon-reload
```

Configuring Core Dumps

To adjust the scope of the data captured in Systemd core dumps and define where Systemd stores them, change the /etc/systemd/coredump.conf configuration file. For more information, see the coredump.conf(5) manual pages.

Before running the coredumpctl command, remove any core dump size limits that apply to the current shell session:

sudo ulimit -c unlimited

For more information about the ulimit command, see the ulimit(1) manual pages.

Analyzing Core Dumps

Use the coredumpctl command to list the core dumps that are available on the system:
```
coredumpctl list
```
To review more information about the core dumps stored for a particular application, specify the executable as an option:
```
coredumpctl list executable-path
```
To review all the core dumps that are stored for a failed process on the system, specify the process ID instead:
```
coredumpctl list process-id
```

Exporting Core Dumps

To export the core dump for bug reporting purposes, specify the process ID and output file when you run the coredumpctl dump command:
```
coredumpctl dump process-id -o output-file
```
Optionally, you can export an SOS report with extra information about the system. For more information, see Working With the sos Command.
On the same system or a different one, install the gdb package and then step through a core dump with the GNU Debugger by using the coredumpctl debug command:
```
sudo dnf install gdb
```
```
coredumpctl debug process-id
```

For more information about the coredumpctl command, see the coredumpctl(1) manual pages.