[INFERENCE] Remove node pinning logic #134630
Conversation
|
Pinging @elastic/ml-core (Team:ML) |
| * @param divisor - divisor to divide the initial requests per time unit by | ||
| */ | ||
| public synchronized void updateTokensPerTimeUnit(Integer divisor) { | ||
| private synchronized void updateTokensPerTimeUnit(Integer divisor) { |
There was a problem hiding this comment.
@jonathan-buttner is this code and the related updateRateLimitDivisor() functions still required?
There was a problem hiding this comment.
No I'm pretty sure we only added for the node pinning so we should be able to remove it
There was a problem hiding this comment.
Should we also remove the routed logic here: https://github.com/elastic/elasticsearch/blob/main/x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/inference/action/BaseInferenceActionRequest.java#L26
Good catch thanks. Removed in ab7e442 |
Developed for EIS and no longer required
Request routing logic was added to enable per node rate limiting with the Elastic Inference Service but that code remained behind a feature flag and was never enabled. EIS now implements rate limiting make the code redundant.