[ https://issues.apache.org/jira/browse/DRILL-8307?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17605352#comment-17605352 ]
Charles Givre commented on DRILL-8307: -------------------------------------- [~dzamo] I don't know if you're planning on taking this one, but I had two thoughts here: # Would it be worth looking to see where else in Drill we use the Apache httpclient and switch them all over to OkHttp? # I started a branch ([https://github.com/cgivre/drill/tree/druid_evf)] which converts the Druid plugin to EVF. It was almost done. The remaining parts to finish were all in the ScanBatchCreator. Would you want to incorporate this work as well? > Druid storage plugin's use of Apache HttpClient is not thread safe > ------------------------------------------------------------------ > > Key: DRILL-8307 > URL: https://issues.apache.org/jira/browse/DRILL-8307 > Project: Apache Drill > Issue Type: Bug > Components: Storage - Other > Affects Versions: 1.20.2 > Reporter: James Turton > Priority: Major > Fix For: 1.20.3 > > > When multiple concurrent queries are run against a single Druid storage > plugin then an error such as is shown below is reported by the Apache > HttpClient used in that plugin. The Druid storage plugin uses a single static > HttpClient instance which should be replaced with something like > PoolingHttpClientConnectionManager for multithreaded access. > [1cdd2b75-1310-xxxx-xxxx-5a638567ed07:foreman] INFO > o.a.d.e.s.d.s.DruidSchemaFactory > - User Error Occurred: Failure while loading druid datasources for database > 'druid-egsmd300'. (Invalid use of BasicClientConnManager: connection still > allocated. > Make sure to release the connection before allocating another one.) > org.apache.drill.common.exceptions.UserException: DATA_READ ERROR: Failure > while loading druid datasources for database '<connection name>'. -- This message was sent by Atlassian Jira (v8.20.10#820010)