diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/master/ServerManager.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/master/ServerManager.java index 639900eb745..5a48ce8d5b5 100644 --- a/hbase-server/src/main/java/org/apache/hadoop/hbase/master/ServerManager.java +++ b/hbase-server/src/main/java/org/apache/hadoop/hbase/master/ServerManager.java @@ -255,8 +255,7 @@ public class ServerManager { } /** - * Let the server manager know a regionserver is requesting configurations. - * Regionserver will not be added here, but in its first report. + * Let the server manager know a new regionserver has come online * @param request the startup request * @param ia the InetAddress from which request is received * @return The ServerName we know this server as. @@ -278,6 +277,10 @@ public class ServerManager { request.getServerStartCode()); checkClockSkew(sn, request.getServerCurrentTime()); checkIsDead(sn, "STARTUP"); + if (!checkAndRecordNewServer(sn, ServerLoad.EMPTY_SERVERLOAD)) { + LOG.warn("THIS SHOULD NOT HAPPEN, RegionServerStartup" + + " could not record the server: " + sn); + } return sn; } @@ -340,6 +343,7 @@ public class ServerManager { if (null == this.onlineServers.replace(sn, sl)) { // Already have this host+port combo and its just different start code? // Just let the server in. Presume master joining a running cluster. + // recordNewServer is what happens at the end of reportServerStartup. // The only thing we are skipping is passing back to the regionserver // the ServerName to use. Here we presume a master has already done // that so we'll press on with whatever it gave us for ServerName. diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java index a116f5ee677..5f5c9c599f3 100644 --- a/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java +++ b/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java @@ -1027,9 +1027,9 @@ public class HRegionServer extends HasThread implements this.rsHost = new RegionServerCoprocessorHost(this, this.conf); } - // Get configurations from the Master. Break if server is stopped or - // the clusterup flag is down or hdfs went wacky. Then start up all Services. - // Use RetryCounter to get backoff in case Master is struggling to come up. + // Try and register with the Master; tell it we are here. Break if server is stopped or the + // clusterup flag is down or hdfs went wacky. Once registered successfully, go ahead and start + // up all Services. Use RetryCounter to get backoff in case Master is struggling to come up. RetryCounterFactory rcf = new RetryCounterFactory(Integer.MAX_VALUE, this.sleeper.getPeriod(), 1000 * 60 * 5); RetryCounter rc = rcf.create(); @@ -1056,7 +1056,7 @@ public class HRegionServer extends HasThread implements rsQuotaManager.start(getRpcServer().getScheduler()); } - // Run mode. + // We registered with the Master. Go into run mode. long lastMsg = System.currentTimeMillis(); long oldRequestCount = -1; // The main run loop. @@ -1090,14 +1090,7 @@ public class HRegionServer extends HasThread implements } long now = System.currentTimeMillis(); if ((now - lastMsg) >= msgInterval) { - // Register with the Master now that our setup is complete. - if (tryRegionServerReport(lastMsg, now) && !online.get()) { - // Wake up anyone waiting for this server to online - synchronized (online) { - online.set(true); - online.notifyAll(); - } - } + tryRegionServerReport(lastMsg, now); lastMsg = System.currentTimeMillis(); } if (!isStopped() && !isAborted()) { @@ -1289,12 +1282,12 @@ public class HRegionServer extends HasThread implements } @InterfaceAudience.Private - protected boolean tryRegionServerReport(long reportStartTime, long reportEndTime) + protected void tryRegionServerReport(long reportStartTime, long reportEndTime) throws IOException { RegionServerStatusService.BlockingInterface rss = rssStub; if (rss == null) { // the current server could be stopping. - return false; + return; } ClusterStatusProtos.ServerLoad sl = buildServerLoad(reportStartTime, reportEndTime); try { @@ -1314,9 +1307,7 @@ public class HRegionServer extends HasThread implements // Couldn't connect to the master, get location from zk and reconnect // Method blocks until new master is found or we are stopped createRegionServerStatusStub(true); - return false; } - return true; } ClusterStatusProtos.ServerLoad buildServerLoad(long reportStartTime, long reportEndTime) @@ -1580,6 +1571,11 @@ public class HRegionServer extends HasThread implements ", sessionid=0x" + Long.toHexString(this.zooKeeper.getRecoverableZooKeeper().getSessionId())); + // Wake up anyone waiting for this server to online + synchronized (online) { + online.set(true); + online.notifyAll(); + } } catch (Throwable e) { stop("Failed initialization"); throw convertThrowableToIOE(cleanup(e, "Failed init"), @@ -2558,9 +2554,10 @@ public class HRegionServer extends HasThread implements } /* - * Run initialization using parameters passed us by the master. + * Let the master know we're here Run initialization using parameters passed + * us by the master. * @return A Map of key/value configurations we got from the Master else - * null if we failed during report. + * null if we failed to register. * @throws IOException */ private RegionServerStartupResponse reportForDuty() throws IOException { diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestGetReplicationLoad.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestGetReplicationLoad.java index 6ce8d76d595..bad92b80a9b 100644 --- a/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestGetReplicationLoad.java +++ b/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestGetReplicationLoad.java @@ -60,9 +60,8 @@ public class TestGetReplicationLoad { } @Override - protected boolean tryRegionServerReport(long reportStartTime, long reportEndTime) { + protected void tryRegionServerReport(long reportStartTime, long reportEndTime) { // do nothing - return true; } } diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestMasterMetrics.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestMasterMetrics.java index ead3a58e1a8..554e36ef944 100644 --- a/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestMasterMetrics.java +++ b/hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestMasterMetrics.java @@ -54,6 +54,12 @@ public class TestMasterMetrics { KeeperException, InterruptedException { super(conf, cp); } + + @Override + protected void tryRegionServerReport( + long reportStartTime, long reportEndTime) { + // do nothing + } } @BeforeClass diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/TestCompactionInDeadRegionServer.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/TestCompactionInDeadRegionServer.java index 32ba93fc703..879823b944b 100644 --- a/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/TestCompactionInDeadRegionServer.java +++ b/hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/TestCompactionInDeadRegionServer.java @@ -74,14 +74,13 @@ public class TestCompactionInDeadRegionServer { } @Override - protected boolean tryRegionServerReport(long reportStartTime, long reportEndTime) + protected void tryRegionServerReport(long reportStartTime, long reportEndTime) throws IOException { try { super.tryRegionServerReport(reportStartTime, reportEndTime); } catch (YouAreDeadException e) { // ignore, do not abort } - return true; } }