What problem does your feature solve?
Currently, request bodies for data-dependent endpoints are probabilistically generated from seed data, with the intent that requests will uniformly represent real RPC traffic. While this is good for general-purpose load testing patterns, it makes it challenging/impossible to exercise specific request shapes, edge cases/"unhappy paths" intentionally. This feature would enable request determinism in some unspecified capacity.
What would you like to see?
There'll need to be a considerable amount of design consideration put into this, but ideally, blaster would have a mechanism to deterministically map requests/input data to endpoints. There's a number of ways to go about this; initial ones that come to mind are:
- seeding request generation: for a given seed, the same request sequence is generated run to run
- extremely low overhead to implement as random functions already have seeded variants, but may be difficult to use to exercise specific request patterns
- hard-coded request body seeding
- requires users to provide correctly specified inputs
- replay capabilities: allow users to record specific request sequences and replay them repeatedly
- potentially difficult user experience
What alternatives are there?
N/A
What problem does your feature solve?
Currently, request bodies for data-dependent endpoints are probabilistically generated from seed data, with the intent that requests will uniformly represent real RPC traffic. While this is good for general-purpose load testing patterns, it makes it challenging/impossible to exercise specific request shapes, edge cases/"unhappy paths" intentionally. This feature would enable request determinism in some unspecified capacity.
What would you like to see?
There'll need to be a considerable amount of design consideration put into this, but ideally, blaster would have a mechanism to deterministically map requests/input data to endpoints. There's a number of ways to go about this; initial ones that come to mind are:
What alternatives are there?
N/A