HBASE-8143 HBase on Hadoop 2 with local short circuit reads (ssr) causes OOM

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1545852 13f79535-47bb-0310-9956-ffa450edef68

hbase-common/src/main/resources/hbase-default.xml

@@ -991,6 +991,20 @@ possible configurations would overwhelm and obscure the important.
         datanode, performing block recovery to timeout on a dead datanode; usually
         dfs.socket.timeout. See the end of HBASE-8389 for more.</description>
   </property>
+  <property>
+    <name>hbase.dfs.client.read.shortcircuit.buffer.size</name>
+    <value>131072</value>
+    <description>If the DFSClient configuration
+    dfs.client.read.shortcircuit.buffer.size is unset, we will
+    use what is configured here as the short circuit read default
+    direct byte buffer size. DFSClient native default is 1MB; HBase
+    keeps its HDFS files open so number of file blocks * 1MB soon
+    starts to add up and threaten OOME because of a shortage of
+    direct memory.  So, we set it down from the default.  Make
+    it > the default hbase block size set in the HColumnDescriptor
+    which is usually 64k.
+    </description>
+  </property>
   <property>
     <name>hbase.regionserver.checksum.verify</name>
     <value>true</value>

hbase-server/src/main/java/org/apache/hadoop/hbase/master/HMaster.java

@@ -410,6 +410,7 @@ MasterServices, Server {
     this.conf = new Configuration(conf);
     // Disable the block cache on the master
     this.conf.setFloat(HConstants.HFILE_BLOCK_CACHE_SIZE_KEY, 0.0f);
+    FSUtils.setupShortCircuitRead(conf);
     // Server to handle client requests.
     String hostname = Strings.domainNamePointerToHostName(DNS.getDefaultHost(
       conf.get("hbase.master.dns.interface", "default"),

hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java

@@ -357,7 +357,6 @@ public class HRegionServer implements ClientProtos.ClientService.BlockingInterfa
 
   protected final Configuration conf;
 
-  private boolean useHBaseChecksum; // verify hbase checksums?
   private Path rootDir;
 
   protected final ReentrantReadWriteLock lock = new ReentrantReadWriteLock();
@@ -532,19 +531,7 @@ public class HRegionServer implements ClientProtos.ClientService.BlockingInterfa
     checkCodecs(this.conf);
     this.userProvider = UserProvider.instantiate(conf);
 
-    // do we use checksum verification in the hbase? If hbase checksum verification
+    FSUtils.setupShortCircuitRead(this.conf);
-    // is enabled, then we automatically switch off hdfs checksum verification.
-    this.useHBaseChecksum = conf.getBoolean(HConstants.HBASE_CHECKSUM_VERIFICATION, true);
-
-    // check that the user has not set the "dfs.client.read.shortcircuit.skip.checksum" property.
-    boolean shortCircuitSkipChecksum = conf.getBoolean(
-        "dfs.client.read.shortcircuit.skip.checksum", false);
-    if (shortCircuitSkipChecksum) {
-      LOG.warn("Configuration \"dfs.client.read.shortcircuit.skip.checksum\" should not " +
-          "be set to true." + (this.useHBaseChecksum ? " HBase checksum doesn't require " +
-          "it, see https://issues.apache.org/jira/browse/HBASE-6868." : ""));
-      assert !shortCircuitSkipChecksum; //this will fail if assertions are on
-    }
 
     // Config'ed params
     this.numRetries = this.conf.getInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER,
@@ -1240,8 +1227,10 @@ public class HRegionServer implements ClientProtos.ClientService.BlockingInterfa
       // accessors will be going against wrong filesystem (unless all is set
       // to defaults).
       FSUtils.setFsDefault(this.conf, FSUtils.getRootDir(this.conf));
-      // Get fs instance used by this RS
+      // Get fs instance used by this RS.  Do we use checksum verification in the hbase? If hbase
-      this.fs = new HFileSystem(this.conf, this.useHBaseChecksum);
+      // checksum verification enabled, then automatically switch off hdfs checksum verification.
+      boolean useHBaseChecksum = conf.getBoolean(HConstants.HBASE_CHECKSUM_VERIFICATION, true);
+      this.fs = new HFileSystem(this.conf, useHBaseChecksum);
       this.rootDir = FSUtils.getRootDir(this.conf);
       this.tableDescriptors = new FSTableDescriptors(this.fs, this.rootDir, true);
       this.hlog = setupWALAndReplication();

hbase-server/src/main/java/org/apache/hadoop/hbase/util/FSUtils.java

@@ -44,7 +44,6 @@ import java.util.regex.Pattern;
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 import org.apache.hadoop.classification.InterfaceAudience;
-import org.apache.hadoop.classification.InterfaceStability;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.BlockLocation;
 import org.apache.hadoop.fs.FSDataInputStream;
@@ -56,12 +55,12 @@ import org.apache.hadoop.fs.PathFilter;
 import org.apache.hadoop.fs.permission.FsAction;
 import org.apache.hadoop.fs.permission.FsPermission;
 import org.apache.hadoop.hbase.ClusterId;
-import org.apache.hadoop.hbase.TableName;
 import org.apache.hadoop.hbase.HColumnDescriptor;
 import org.apache.hadoop.hbase.HConstants;
 import org.apache.hadoop.hbase.HDFSBlocksDistribution;
 import org.apache.hadoop.hbase.HRegionInfo;
 import org.apache.hadoop.hbase.RemoteExceptionHandler;
+import org.apache.hadoop.hbase.TableName;
 import org.apache.hadoop.hbase.exceptions.DeserializationException;
 import org.apache.hadoop.hbase.fs.HFileSystem;
 import org.apache.hadoop.hbase.master.HMaster;
@@ -1876,4 +1875,40 @@ public abstract class FSUtils {
 
     LOG.info(overheadMsg);
   }
+
+  /**
+   * Do our short circuit read setup.
+   * Checks buffer size to use and whether to do checksumming in hbase or hdfs.
+   * @param conf
+   */
+  public static void setupShortCircuitRead(final Configuration conf) {
+    // Check that the user has not set the "dfs.client.read.shortcircuit.skip.checksum" property.
+    boolean shortCircuitSkipChecksum =
+      conf.getBoolean("dfs.client.read.shortcircuit.skip.checksum", false);
+    boolean useHBaseChecksum = conf.getBoolean(HConstants.HBASE_CHECKSUM_VERIFICATION, true);
+    if (shortCircuitSkipChecksum) {
+      LOG.warn("Configuration \"dfs.client.read.shortcircuit.skip.checksum\" should not " +
+        "be set to true." + (useHBaseChecksum ? " HBase checksum doesn't require " +
+        "it, see https://issues.apache.org/jira/browse/HBASE-6868." : ""));
+      assert !shortCircuitSkipChecksum; //this will fail if assertions are on
+    }
+    checkShortCircuitReadBufferSize(conf);
+  }
+
+  /**
+   * Check if short circuit read buffer size is set and if not, set it to hbase value.
+   * @param conf
+   */
+  public static void checkShortCircuitReadBufferSize(final Configuration conf) {
+    final int defaultSize = HConstants.DEFAULT_BLOCKSIZE * 2;
+    final int notSet = -1;
+    // DFSConfigKeys.DFS_CLIENT_READ_SHORTCIRCUIT_BUFFER_SIZE_KEY is only defined in h2
+    final String dfsKey = "dfs.client.read.shortcircuit.buffer.size";
+    int size = conf.getInt(dfsKey, notSet);
+    // If a size is set, return -- we will use it.
+    if (size != notSet) return;
+    // But short circuit buffer size is normally not set.  Put in place the hbase wanted size.
+    int hbaseSize = conf.getInt("hbase." + dfsKey, defaultSize);
+    conf.setIfUnset(dfsKey, Integer.toString(hbaseSize));
+  }
 }