Skip to main content

Setting up coresnapd on FreeBSD

To install Backtrace's package mirror, run the following command:

# fetch https://packages.backtrace.io/LICENSE/stork/freebsd/RELEASE/backtrace.conf -o /etc/pkg/backtrace.conf

Replace <LICENSE> with your license number. If you haven't received your license number yet, contact us. Replace RELEASE with the major version number (for example, 11, 12, etc.) that matches your FreeBSD installation. Tthe minimum required version for coresnap is 11.0.

This command configures pkg to pull from Backtrace's mirror. To verify proper pkg configuration, run the following commands:

# pkg update
# pkg search -x '^backtrace-*'

Once verified, you can proceed to install the required packages:

# pkg install backtrace-coroner backtrace-coresnap backtrace-ptrace backtrace-libbt

Initial Configuration​

After installing the packages, coresnapd requires a few configuration changes to start. Run the following commands:

# sysrc coresnapd_enable=YES
# sysrc crashinfo_program="/opt/backtrace/sbin/kernel_coresnap"
# sysrc local_startup="$(sysrc -n local_startup) /opt/backtrace/etc/rc.d"

coresnapd uses coroner to perform submissions of crashes. To configure it, edit the /usr/local/etc/coroner.cf file. For more information, consult the coroner documentation.

Now you can start coresnapd:

# service coresnapd start

To check its status, use the following command:

$ service coresnapd status
coresnapd is running as pid 613.

Testing​

Coresnapd leverages devd on FreeBSD to get notified about userland crashes, which are automatically stored in /var/coresnap/archive/corefiles. To test, simply run a program that crashes, and the crash information will be processed.

Kernel crashes are processed by /opt/backtrace/sbin/kernel_coresnap instead of the standard /usr/sbin/crashinfo and may require configuring a dump device. To configure it, run the following command:

# sysrc dumpdev=AUTO

or more information on configuring the dump device, refer to the rc.conf documentation. To force a kernel crash for testing purposes, run the following command:

# sysctl debug.kdb.panic=1

When the system comes up again, it will process the kernel crash and deliver it to coresnapd to submit using coroner.