We've observed an issue in production where event batches grow well beyond the logging_max_buffer_size in StatsigOptions. In the worst case, they get so large they are rejected by the statsig backend and not retried.
The fix in this version is to split the full queue into logging_max_buffer_size chunks that are sent in multiple requests
Its possible that under high qps loads the sdk will still not be able to keep up
Included In This Release
- adbd162 tore-statsig
- fix: chunk logevent requests into batch_size (#374)
Full Changelog: 2.4.0...2.4.1