Engineering, practice, product
Deep dives into the problems behind incident management and on-call, the practices that make being paged survivable, and what is new in TokayOps.
- Reliable notifications: one HTTP call and everything behind it "The on-call engineer must find out" looks like a single API call. Add the word "reliably" and it turns into a corridor of doors: timeouts, retries, duplicates, updates to a message that may not exist, and a second instance.
- On-call scheduling from scratch: what to decide before the first line of code An on-call schedule looks like a function of the current moment and behaves like a function of its own history. The design decisions that follow from that: revisions, phase anchors, calendar arithmetic, and overrides.