Github user ortutay commented on the issue:
https://github.com/apache/phoenix/pull/298
@JamesRTaylor @maryannxue updated for all comments in the PR ; this now
uses an exception flow for checking for presence of SQ cache, instead of the
previous RPC approach. I also a dded to the integration test to verify that the
SQ cache is properly used across queries.
Also, regarding expiring the cache--we have been using a technique
internally for this based on nonces. You can add a no-op condition to the query
such as "1=1" or "2=2" which changes the persistent cache key, and thus is a
way to force cache invalidation. You can do it on a daily basis, eg.
20180731=20180731, and this can also be used to pre-warm caches for upcoming
days as well. I'm planning to make a blog post explaining SQ cache + how to
rotate nonces once this change has been merged.
---
