Part IV · cs.DCFailure, Supervision, and Durable ExecutionMatthew Long · 28 pagesDownload PDFCodeTable of contents1 Introduction1.1 Contributions1.2 Relation to the rest of the series2 Preliminaries2.1 Imported objects2.2 The execution interface2.3 Recording the start value2.4 Observed closed systems3 The execution model3.1 Fail-stop failure3.2 An execution model as an agent3.3 Symbols introduced here4 Retry, timeout and backoff5 Supervision6 Persistence7 Logs and replay8 Supervision refinement9 Effectively-once execution10 Workflows10.1 The construction and its fragment10.2 Dependency shape11 Worked examples11.1 A durable pipeline11.2 A supervised worker group11.3 An N-shaped workflow11.4 A replayed model call12 Prior art13 Limitations14 Conclusion15 Code evidenceReferencesJavaScript is required to display this paper inline. Download the PDF instead.