How to debug the wal-redo process？

[softhuafei]

Hello, thank you for your work. I am a database enthusiast and would like to learn about the implementation details of Neon. However, when I debug the wal-redo process using lldb on macOS, it quickly receives a SIGKILL signal and exits. Could you please advise me on how to debug the wal-redo process? Or do you have any other suggestions for developers?

[knizhnik]

Hello, thank you for your work. I am a database enthusiast and would like to learn about the implementation details of Neon. However, when I debug the wal-redo process using lldb on macOS, it quickly receives a SIGKILL signal and exits. Could you please advise me on how to debug the wal-redo process? Or do you have any other suggestions for developers?

I can not propose better solution rather than temporary increasing wal_redo_timeout to some huge value enough for debugging

[softhuafei]

Thank you for the suggestion, but I'm still receiving SIGKILL. I tried setting wal_redo_timeout to 3 hours, but when I lldb the wal-redo process, it exits after only a few steps. Is there another mechanism that could be killing the wal-redo process?

[knizhnik]

There is also maybe_quiesce method which tries to shutdown walredo processes which have not perform any redo during compaction_period*10