Tue Jul 30 09:29:52 EDT 2019

Debugging distributed systems

Debug and also doc = reconstruction of context.

For single-core there is usually a stack trace.  What is needed for
distributed is something similar that can trace the chain of effects.

If this is RPC it's essentially a distributed stack trace, but is
there an analogy for things that bounce around?  E.g. a rube-goldberg