I'm going to create load tests and I'd like to understand whether implementing my use case is possible via esrally. Basically what I'd like to emulate scenario where I create documents and then update them in single scenario. Do you know if it's possible? How can I pass document id for updates and things like that?
Firstly the bulk operation does support the ability to update documents that have been indexed. This is configured through the parameters
conflict-probability. You can modify how often these updates are biased to recent data through the parameter
recency - caution that impacting older documents will cause greater merge pressure thus increase disk io requirements.
If you need to build your own track, consider consulting the existing tracks which expose this parameter for inspiration. The event data track bulk update challenge also provides an update scenario but exposes some complexity behind a custom parameter source 1 2 3.