fix: Close stream connection before each retry (#1975)

GaoleMeng · gcf-owl-bot[bot] · web-flow · commit 54e9bb9116eb · 2023-02-06T11:38:19.000-08:00
* feat: Split writer into connection worker and wrapper, this is a prerequisite for multiplexing client * feat: add connection worker pool skeleton, used for multiplexing client * feat: add Load api for connection worker for multiplexing client * feat: add multiplexing support to connection worker. We will treat every new stream name as a switch of destinationt * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * feat: port the multiplexing client core algorithm and basic tests also fixed a tiny bug inside fake bigquery write impl for getting thre response from offset * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * feat: wire multiplexing connection pool to stream writer * feat: some fixes for multiplexing client * feat: fix some todos, and reject the mixed behavior of passed in client or not * feat: fix the bug that we may peek into the write_stream field but it's possible the proto schema does not contain this field * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * feat: fix the bug that we may peek into the write_stream field but it's possible the proto schema does not contain this field * feat: add getInflightWaitSeconds implementation * feat: Add schema comparision in connection loop to ensure schema update for the same stream name can be notified * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * feat: add schema update support to multiplexing * fix: fix windows build bug: windows Instant resolution is different with linux * fix: fix another failing tests for windows build * fix: fix another test failure for Windows build * feat: Change new thread for each retry to be a thread pool to avoid create/tear down too much threads if lots of retries happens * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * fix: add back the background executor provider that's accidentally removed * feat: throw error when use connection pool for explicit stream * fix: Add precision truncation to the passed in value from JSON float and double type. * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * modify the bom version * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * fix deadlockissue in ConnectionWorkerPool * fix: fix deadlock issue during close + append for multiplexing * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * fix: fix one potential root cause of deadlock issue for non-multiplexing case * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * Add timeout to inflight queue waiting, and also add some extra log * feat: allow java client lib handle switch table schema for the same stream name * 🦉 Updates from OwlBot post-processor See https://github.com/googleapis/repo-automation-bots/blob/main/packages/owl-bot/README.md * fix: close before retry connection * fix: close before retry connection --------- Co-authored-by: Owl Bot <gcf-owl-bot[bot]@users.noreply.github.com>
diff --git a/google-cloud-bigquerystorage/src/main/java/com/google/cloud/bigquery/storage/v1/ConnectionWorker.java b/google-cloud-bigquerystorage/src/main/java/com/google/cloud/bigquery/storage/v1/ConnectionWorker.java
@@ -243,6 +243,11 @@ public void run() {
 
   private void resetConnection() {
     log.info("Reconnecting for stream:" + streamName + " id: " + writerId);
+    if (this.streamConnection != null) {
+      // It's safe to directly close the previous connection as the in flight messages
+      // will be picked up by the next connection.
+      this.streamConnection.close();
+    }
     this.streamConnection =
         new StreamConnection(
             this.client,
