Record and replay debugging

Software debugging technique

Record and replay debugging is the process of recording the execution of a software program so that it may be played back within a debugger to help diagnose and resolve defects.[1] The concept is analogous to the use of a flight data recorder to diagnose the cause of an airplane flight malfunction.[2]

Recording and replaying

Record and replay debuggers record application state at every step of the program's process and thread execution, including memory interactions, deterministic and non-deterministic inputs, system resource status, and store it to disk in a log.[3] The recording allows the program to be replayed again and again, and debugged exactly as it happened.

Usage

Recordings can be made in one location and replayed in another,[4] which makes it useful for remote debugging.

Record and replay debugging is particularly useful for debugging intermittent and non-deterministic defects, which can be difficult to reproduce.

Record and replay debugging technology is often fundamental to reverse debugging and time travel debugging.

Record and replay debuggers

  • GDB (GNU)[5]
  • LiveRecorder (Undo)[6]
  • rr (Mozilla)
  • TotalView's ReplayEngine (RogueWave)[7]
  • PyTrace for Python[8]

References

  1. ^ Mozilla (2017). "Engineering Record And Replay For Deployability Extended Technical Report". arXiv:1705.05937 [cs.PL].
  2. ^ Zicari, Roberto. "On Software Reliability. Interview with Barry Morris and Dale Vile". ODBMS Industry Watch. ODBMS Industry Watch. Retrieved 2 April 2019.
  3. ^ Undo, Ltd. "System and method for debugging of computer programs". Google Patents. US Patent Office. Retrieved 23 February 2016.
  4. ^ Undo, Ltd. "Remote recording". Undo Documentation. Undo, Ltd. Retrieved 1 October 2019.
  5. ^ "Process Record and Replay (Debugging with GDB)".
  6. ^ "LiveRecorder - Undo".
  7. ^ "TotalView for HPC".
  8. ^ "PyTrace Time Travel Debugger for Python".