You are viewing a plain text version of this content. The canonical link for it is here.

Posted to notifications@iotdb.apache.org by "Xinyu Tan (Jira)" <ji...@apache.org> on 2021/07/12 08:28:00 UTC

[jira] [Created] (IOTDB-1487) Fix clientPool timeout in cluster module

Xinyu Tan created IOTDB-1487:
--------------------------------

             Summary: Fix clientPool timeout in cluster module
                 Key: IOTDB-1487
                 URL: https://issues.apache.org/jira/browse/IOTDB-1487
             Project: Apache IoTDB
          Issue Type: Improvement
            Reporter: Xinyu Tan


There's many ClientPools between nodes in cluster IoTDB that have multiple long TCP connections managed by thrift, and when coordinator has to communicate between nodes, it gets one client from the pool first, sends the RPC, and puts it back into the pool.
However, There is no retention mechanism for ClientPool between nodes. When a Thrift client is unused for a long time, its TCP connection may have been discarded by the operating system, routers, and so on, so there is a timeout when a client thread gets the client again for a write request. Therefore, we need to add some application layer heartbeat preservation mechanism for long TCP connections



--
This message was sent by Atlassian Jira
(v8.3.4#803005)