schedule_task and data driven load test methodology

[daluu]

Three things:

1. How is TaskSet.schedule_task() generally used? It's in API doc but I don't really see any code samples of its use other than this one case in Support for scenarios #171 (comment)

2. Similar to schedule_task, using task decorators and the other conventions, is there a way to pass arguments to any given task? So not just "self" as the sole argument.

3. What approaches are suggested for data driven load tests or load testing at scale where the data varies as you scale up/down the load? e.g. each user/instance spawned runs with its own dataset. How to pass/provide that dataset on a per user/instance basis? Some way to provide args, add your own internal logic and get config via environment variables (or using statically defined config file name/paths)? Or adding custom kw args to locust to use with the tasks, etc. - which doesn't appear mentioned as an option, one would have to extend themselves?

For item 3, this is much easier to do in say JMeter (and I'd assume Gatling) for feeding data to each user/instance of load being generated.

[daluu]

Well I came across this researching use of locust more, for item 3: https://stackoverflow.com/questions/23016278/way-to-use-locust-io-by-supplying-user-list

Still those aren't necessarily optimal approaches, at least for sequential data access by say a user/hatch count.

[cyberw]

Closing this due to lack of activity. I hope you solved your issue, otherwise try slack!