Our current performance tests are fairly rudimentary. Driver behavior can't be customized. Latency numbers are subject to coordinated omission. There is no easy way to find max throughput. There is no easy way to introduce node failures or node health issues. Looking at results requires collecting CLI output and doing any analysis/graphing manually.
This can be solved with a new performance harness running in Fallout. Since this behavior can be built incrementally, this epic should encompass the efforts.
FYI I created with a broader scope, the tickets listed in this epic also belong in there but I didn’t want to move them.
Conversely, we could change this epic to a ticket with subtasks, would you be ok with that?
Done, sorry for the Jira fuss that it might have generated.