Error field -- please tell me exactly what field name(s) you want to generate; I considered this and I couldn't easily find documentation on what we consider to be errors.
I don't understand the "X rate" part of the custom fields. Can you elaborate?
Lambda: the way it's built, the traces are generated by individual goroutines that are running (well, mostly sleeping) for the full duration of a trace. Doing that in lambda is going to be expensive. At the moment, I can't quite think of a way to do this with lambda that makes sense.
Extras:
Most of the outliers I think are better handled by running a separate instance of loadgen with different parameters. Otherwise it's going to be a nightmare to configure.
But missing spans is maybe a "drop every 100th span" option?
I originally designed it as multi-service and then felt like it just got too complicated to specify. If you can give me a sample command line or two with some commentary on interpretation, I'd be happy to consider it again.
Explicit attributes is something I've considered. I'll chew on it a bit more. I think both explicit and automatic values are possible.
Thought I'd just load up some of the ideas we discussed in Slack so that I don't forget about them
Some extra ideas I thought of on top of things
These might be things we can do with running via multiple lambdas, but would be cool in the tool as well