peferron commented on issue #4947: Historicals should only write to remote cache URL: https://github.com/apache/incubator-druid/issues/4947#issuecomment-472125502 One of your previous comments indicated that you were using local cache in this setting: > BTW there is hybrid cache and local L1 level is still used for get. Have you moved away from this configuration since then? I think the local cache could help if the remote cache is churning too much. Since the local cache only stores results for segments located on that historical, it may keep results longer than a remote cache that stores results for all segments in the cluster. Of course it's possible to simply give more memory to the remote cache instead. The admission cache that you linked is also interesting to reduce churn in the remote cache. But even in this situation, you can imagine a local cache being useful to keep results that haven't been deemed worth caching remotely yet. If another query for these results come in, then historicals could get results from the local cache instead of re-scanning the segments, and then potentially write these results to the remote cache if they are now considered worthy.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
