Currently we enforce loop wait (with sleeps) up to predefined cache.wait.put.max.ms configuration value, which is 20 ms, by default when cache is under pressure and scavengers are active. When timeout is over a put operation is being bailed out, resulting in rejected write. Rejected writes are unexpected outcomes, so we have to reconsider this approach.
Currently we enforce loop wait (with sleeps) up to predefined
cache.wait.put.max.ms
configuration value, which is 20 ms, by default when cache is under pressure and scavengers are active. When timeout is over a put operation is being bailed out, resulting in rejected write. Rejected writes are unexpected outcomes, so we have to reconsider this approach.