Lambda distributed tracing showing massive gaps between spans

New Relic One > Explorer > Lambda functions > (select a lambda function) > Distributed tracing > (select a trace)

Sometimes the tracing for a lambda function will show massive gaps between the spans with no explanation.

e.g. I’m looking at one with a trace duration of 324 seconds, but the root span is only 728 ms. Drilling into in-process spans, I see an 874 ms initialization at the beginning of the trace, then a very large, big, enormous gap that’s unaccounted for, then a 547 ms invocation and 0.68 ms overhead at the end of the trace. The timeline chart at the top is literally blank except for those millisecond spans at the beginning and end of the trace.


Does anyone know what causes this? How can I find out where all that time went?