mbwaheed commented on issue #1131: SOLR-14134: Add lazy and time-based evictiction of shared core concurrency metada… URL: https://github.com/apache/lucene-solr/pull/1131#issuecomment-570752800 > * This is done here because all ADDREPLICA paths flow through here logically and we want to evict entries before the replica gets marked ACTIVE (done here) This path is called on each ZK reconnect too and that made me think about, with new push and pull locks instances, what would happen to concurrency if any existing indexing might be running with older locks. That further led me to look into where the right place could be. The answer seems simple: It is when the SolrCore instance is created. And for regular eviction we do not need this tracking either, we can simply do that in SolrCore#close. The basic assumption is: if core container can hold so many SolrCore instances we can very easily hold this simple metadata. The problem of many cores is planned to be addressed by transient cores or zero replica design. With that I don't think this cache need to worry about its own size as long as it ties itself with SolrCore instances. Sorry I did not realize all this when I read the JIRA earlier.
---------------------------------------------------------------- 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: us...@infra.apache.org With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
