[GitHub] [lucene-solr] CaoManhDat commented on pull request #1470: SOLR-14354: Async or using threads in better way for HttpShardHandler

[GitBox <gi...@apache.org>]

CaoManhDat commented on pull request #1470:
URL: https://github.com/apache/lucene-solr/pull/1470#issuecomment-649949789


   Note that this PR including changes for removing the ability to use HttpShardHandler with a custom HttpClient. My reasons for doing that is
   * In current code, we use HttpShardHandlerFactory.LBClient (which can't be changed) when requests to multiple replicas (replicas are interchangeable to other, i.e for query cases), for query to single replica we will use provided HttpClient which can be changed by calling `HttpShardHandlerFactory.getShardHandler(HttpClient)`. But when people look into `getShardHandler(HttpClient)` method they will assume that all requests (single or multiple replicas) will use the provided client. 
   * The client that provided to `getShardHandler` is all places is updateShardHandler.defaultClient. That does not make sense since in the ref-guide mentioned that
   > **shardHandlerFactory**
   > socketTimeout
   The read timeout for intra-cluster query and administrative requests. The default is the same as the distribUpdateSoTimeout specified in the <solrcloud> section.
   therefore the client of httpShardHandlerFactory must be used instead of `updateShardHandler.defaultClient`
   * The changes also lead to less connections usage by switching from Http/1.1 client to Http/2


----------------------------------------------------------------
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org