| commit | 11cd254b9bdd04788df12a062fdc873e9ec4e60a | [log] [tgz] |
|---|---|---|
| author | Andrew Jeffery <andrew@aj.id.au> | Mon May 03 11:03:30 2021 +0930 |
| committer | Andrew Jeffery <andrew@aj.id.au> | Mon Jun 07 11:28:55 2021 +0930 |
| tree | d2734bd0055153349834f62cae96e58b932b98d5 | |
| parent | 2c693c50f9128f84ab48f5efbbc5ede61467962f [diff] |
debug-trigger: Do what we can to debug unresponsive systems debug-trigger is a small daemon which translates an external signal that the system is unresponsive into a system crash for debug data collection. Signed-off-by: Andrew Jeffery <andrew@aj.id.au> Change-Id: Ic04f1a95aaf651d56b75f5261d8a6f76d34477a7
debug-trigger listens for an external signal that the BMC is in some way unresponsive. When the signal is received it triggers a crash to collect debug data and reboots the system in the hope that it will recover.
debug-trigger implements a simple protocol over an LPC KCS device as its trigger source.
debug-trigger implements a single action once the trigger event is received, which is to crash the kernel via /proc/sysrq-trigger. For systems with kdump configured this results in collection of system state as context for why the system was externally unresponsive.