HBASE-27160 ClientZKSyncer.deleteDataForClientZkUntilSuccess should b…

…reak from the loop when deletion is succeeded (#4579) Signed-off-by: Bryan Beaudreault <bbeaudreault@apache.org>
apache · Jun 26, 2022 · d7f6861 · d7f6861
1 parent 42853aa
commit d7f6861
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 9 deletions.
diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/master/zksyncer/ClientZKSyncer.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/master/zksyncer/ClientZKSyncer.java
@@ -196,12 +196,16 @@ private void deleteDataForClientZkUntilSuccess(String node) throws InterruptedEx
       LOG.debug("Delete remote " + node + ", client zk wather: " + clientZkWatcher);
       try {
         ZKUtil.deleteNode(clientZkWatcher, node);
+        break;
       } catch (KeeperException e) {
+        if (e.code() == KeeperException.Code.NONODE) {
+          LOG.debug("Node is already deleted, give up", e);
+          break;
+        }
         LOG.debug("Failed to delete node from client ZK, will retry later", e);
         if (e.code() == KeeperException.Code.SESSIONEXPIRED) {
           reconnectAfterExpiration();
         }
-
       }
     }
   }

diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/client/TestSeparateClientZKCluster.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/client/TestSeparateClientZKCluster.java
@@ -133,15 +133,10 @@ public void testMasterSwitch() throws Exception {
       HMaster master = cluster.getMaster();
       master.stopMaster();
       LOG.info("Stopped master {}", master.getServerName());
-      while (master.isAlive()) {
-        Thread.sleep(200);
-      }
+      TEST_UTIL.waitFor(30000, () -> !master.isAlive());
       LOG.info("Shutdown master {}", master.getServerName());
-      while (cluster.getMaster() == null || !cluster.getMaster().isInitialized()) {
-        LOG.info("Get master {}",
-          cluster.getMaster() == null ? "null" : cluster.getMaster().getServerName());
-        Thread.sleep(200);
-      }
+      TEST_UTIL.waitFor(30000,
+        () -> cluster.getMaster() != null && cluster.getMaster().isInitialized());
       LOG.info("Got master {}", cluster.getMaster().getServerName());
       // confirm client access still works
       assertTrue(admin.balance(BalanceRequest.defaultInstance()).isBalancerRan());