kkewwei opened a new issue, #16495: URL: https://github.com/apache/lucene/issues/16495
### Description Some of our production clusters rely heavily on `QueryCache`. Queries perform well when cache hits occur. However, segment merge operations evict the corresponding `QueryCache` entries. From enhanced logs, we observed up to 2.3GB of cache being cleared under extreme conditions. <img width="2578" height="150" alt="Image" src="https://github.com/user-attachments/assets/7da9b79c-0ded-4f86-9396-08d4f1c681e3"; /> New segments are created after merging, and `QueryCache` needs to be rebuilt for these new segments. The cache rebuilding process is resource-intensive. During this period, query latency rises drastically, and CPU on some nodes becomes fully saturated. This negatively impacts latency-sensitive services. <img width="1368" height="708" alt="Image" src="https://github.com/user-attachments/assets/c264c885-6105-48c1-9ae9-2d5efb9ba390"; /> <img width="826" height="698" alt="Image" src="https://github.com/user-attachments/assets/7d3739e1-0a28-47a2-8401-289e9d15d8f1"; /> ``` If we could introduce an pre-build cache for newly generated segments when there is remaining QueryCache` capacity. This can mitigate performance impacts on sensitive workloads. ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
