diff --git a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt index a8599bb2cdc..ce6d1a2d753 100644 --- a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt +++ b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt @@ -66,6 +66,9 @@ Release 2.5.0 - UNRELEASED HDFS-6181. Fix the wrong property names in NFS user guide (brandonli) + HDFS-6180. dead node count / listing is very broken in JMX and old GUI. + (wheat9) + Release 2.4.1 - UNRELEASED INCOMPATIBLE CHANGES diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/DatanodeManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/DatanodeManager.java index c39d274b5f5..4326f94bac9 100644 --- a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/DatanodeManager.java +++ b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/DatanodeManager.java @@ -34,10 +34,6 @@ import org.apache.hadoop.hdfs.protocol.HdfsConstants.DatanodeReportType; import org.apache.hadoop.hdfs.server.blockmanagement.DatanodeDescriptor.BlockTargetPair; import org.apache.hadoop.hdfs.server.blockmanagement.DatanodeDescriptor.CachedBlocksList; import org.apache.hadoop.hdfs.server.namenode.CachedBlock; -import org.apache.hadoop.hdfs.server.namenode.HostFileManager; -import org.apache.hadoop.hdfs.server.namenode.HostFileManager.Entry; -import org.apache.hadoop.hdfs.server.namenode.HostFileManager.EntrySet; -import org.apache.hadoop.hdfs.server.namenode.HostFileManager.MutableEntrySet; import org.apache.hadoop.hdfs.server.namenode.NameNode; import org.apache.hadoop.hdfs.server.namenode.Namesystem; import org.apache.hadoop.hdfs.server.protocol.*; @@ -53,6 +49,7 @@ import org.apache.hadoop.util.Time; import java.io.IOException; import java.io.PrintWriter; import java.net.InetAddress; +import java.net.InetSocketAddress; import java.net.UnknownHostException; import java.util.*; @@ -211,13 +208,11 @@ public class DatanodeManager { // in the cache; so future calls to resolve will be fast. if (dnsToSwitchMapping instanceof CachedDNSToSwitchMapping) { final ArrayList locations = new ArrayList(); - for (Entry entry : hostFileManager.getIncludes()) { - if (!entry.getIpAddress().isEmpty()) { - locations.add(entry.getIpAddress()); - } + for (InetSocketAddress addr : hostFileManager.getIncludes()) { + locations.add(addr.getAddress().getHostAddress()); } dnsToSwitchMapping.resolve(locations); - }; + } final long heartbeatIntervalSeconds = conf.getLong( DFSConfigKeys.DFS_HEARTBEAT_INTERVAL_KEY, @@ -1198,46 +1193,45 @@ public class DatanodeManager { boolean listDeadNodes = type == DatanodeReportType.ALL || type == DatanodeReportType.DEAD; - ArrayList nodes = null; - final MutableEntrySet foundNodes = new MutableEntrySet(); + ArrayList nodes; + final HostFileManager.HostSet foundNodes = new HostFileManager.HostSet(); + final HostFileManager.HostSet includedNodes = hostFileManager.getIncludes(); + final HostFileManager.HostSet excludedNodes = hostFileManager.getExcludes(); + synchronized(datanodeMap) { nodes = new ArrayList(datanodeMap.size()); - Iterator it = datanodeMap.values().iterator(); - while (it.hasNext()) { - DatanodeDescriptor dn = it.next(); + for (DatanodeDescriptor dn : datanodeMap.values()) { final boolean isDead = isDatanodeDead(dn); - if ( (isDead && listDeadNodes) || (!isDead && listLiveNodes) ) { - nodes.add(dn); + if ((listLiveNodes && !isDead) || (listDeadNodes && isDead)) { + nodes.add(dn); } - foundNodes.add(dn); + foundNodes.add(HostFileManager.resolvedAddressFromDatanodeID(dn)); } } if (listDeadNodes) { - final EntrySet includedNodes = hostFileManager.getIncludes(); - final EntrySet excludedNodes = hostFileManager.getExcludes(); - for (Entry entry : includedNodes) { - if ((foundNodes.find(entry) == null) && - (excludedNodes.find(entry) == null)) { - // The remaining nodes are ones that are referenced by the hosts - // files but that we do not know about, ie that we have never - // head from. Eg. an entry that is no longer part of the cluster - // or a bogus entry was given in the hosts files - // - // If the host file entry specified the xferPort, we use that. - // Otherwise, we guess that it is the default xfer port. - // We can't ask the DataNode what it had configured, because it's - // dead. - DatanodeDescriptor dn = - new DatanodeDescriptor(new DatanodeID(entry.getIpAddress(), - entry.getPrefix(), "", - entry.getPort() == 0 ? defaultXferPort : entry.getPort(), - defaultInfoPort, defaultInfoSecurePort, defaultIpcPort)); - dn.setLastUpdate(0); // Consider this node dead for reporting - nodes.add(dn); + for (InetSocketAddress addr : includedNodes) { + if (foundNodes.matchedBy(addr) || excludedNodes.match(addr)) { + continue; } + // The remaining nodes are ones that are referenced by the hosts + // files but that we do not know about, ie that we have never + // head from. Eg. an entry that is no longer part of the cluster + // or a bogus entry was given in the hosts files + // + // If the host file entry specified the xferPort, we use that. + // Otherwise, we guess that it is the default xfer port. + // We can't ask the DataNode what it had configured, because it's + // dead. + DatanodeDescriptor dn = new DatanodeDescriptor(new DatanodeID(addr + .getAddress().getHostAddress(), addr.getHostName(), "", + addr.getPort() == 0 ? defaultXferPort : addr.getPort(), + defaultInfoPort, defaultInfoSecurePort, defaultIpcPort)); + dn.setLastUpdate(0); // Consider this node dead for reporting + nodes.add(dn); } } + if (LOG.isDebugEnabled()) { LOG.debug("getDatanodeListForReport with " + "includedNodes = " + hostFileManager.getIncludes() + diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/HostFileManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/HostFileManager.java new file mode 100644 index 00000000000..0b8d6c5bc16 --- /dev/null +++ b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/HostFileManager.java @@ -0,0 +1,217 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.hadoop.hdfs.server.blockmanagement; + +import com.google.common.annotations.VisibleForTesting; +import com.google.common.base.Function; +import com.google.common.base.Joiner; +import com.google.common.base.Preconditions; +import com.google.common.collect.HashMultimap; +import com.google.common.collect.Iterators; +import com.google.common.collect.Multimap; +import com.google.common.collect.UnmodifiableIterator; +import org.apache.commons.logging.Log; +import org.apache.commons.logging.LogFactory; +import org.apache.hadoop.hdfs.protocol.DatanodeID; +import org.apache.hadoop.util.HostsFileReader; + +import javax.annotation.Nullable; +import java.io.IOException; +import java.net.InetAddress; +import java.net.InetSocketAddress; +import java.net.URI; +import java.net.URISyntaxException; +import java.util.Collection; +import java.util.HashSet; +import java.util.Iterator; +import java.util.Map; + +/** + * This class manages the include and exclude files for HDFS. + *

+ * These files control which DataNodes the NameNode expects to see in the + * cluster. Loosely speaking, the include file, if it exists and is not + * empty, is a list of everything we expect to see. The exclude file is + * a list of everything we want to ignore if we do see it. + *

+ * Entries may or may not specify a port. If they don't, we consider + * them to apply to every DataNode on that host. The code canonicalizes the + * entries into IP addresses. + *

+ *

+ * The code ignores all entries that the DNS fails to resolve their IP + * addresses. This is okay because by default the NN rejects the registrations + * of DNs when it fails to do a forward and reverse lookup. Note that DNS + * resolutions are only done during the loading time to minimize the latency. + */ +class HostFileManager { + private static final Log LOG = LogFactory.getLog(HostFileManager.class); + private HostSet includes = new HostSet(); + private HostSet excludes = new HostSet(); + + private static HostSet readFile(String type, String filename) + throws IOException { + HostSet res = new HostSet(); + if (!filename.isEmpty()) { + HashSet entrySet = new HashSet(); + HostsFileReader.readFileToSet(type, filename, entrySet); + for (String str : entrySet) { + InetSocketAddress addr = parseEntry(type, filename, str); + if (addr != null) { + res.add(addr); + } + } + } + return res; + } + + @VisibleForTesting + static InetSocketAddress parseEntry(String type, String fn, String line) { + try { + URI uri = new URI("dummy", line, null, null, null); + int port = uri.getPort() == -1 ? 0 : uri.getPort(); + InetSocketAddress addr = new InetSocketAddress(uri.getHost(), port); + if (addr.isUnresolved()) { + LOG.warn(String.format("Failed to resolve address `%s` in `%s`. " + + "Ignoring in the %s list.", line, fn, type)); + return null; + } + return addr; + } catch (URISyntaxException e) { + LOG.warn(String.format("Failed to parse `%s` in `%s`. " + "Ignoring in " + + "the %s list.", line, fn, type)); + } + return null; + } + + static InetSocketAddress resolvedAddressFromDatanodeID(DatanodeID id) { + return new InetSocketAddress(id.getIpAddr(), id.getXferPort()); + } + + synchronized HostSet getIncludes() { + return includes; + } + + synchronized HostSet getExcludes() { + return excludes; + } + + // If the includes list is empty, act as if everything is in the + // includes list. + synchronized boolean isIncluded(DatanodeID dn) { + return includes.isEmpty() || includes.match + (resolvedAddressFromDatanodeID(dn)); + } + + synchronized boolean isExcluded(DatanodeID dn) { + return excludes.match(resolvedAddressFromDatanodeID(dn)); + } + + synchronized boolean hasIncludes() { + return !includes.isEmpty(); + } + + void refresh(String includeFile, String excludeFile) throws IOException { + HostSet newIncludes = readFile("included", includeFile); + HostSet newExcludes = readFile("excluded", excludeFile); + synchronized (this) { + includes = newIncludes; + excludes = newExcludes; + } + } + + /** + * The HostSet allows efficient queries on matching wildcard addresses. + *

+ * For InetSocketAddress A and B with the same host address, + * we define a partial order between A and B, A <= B iff A.getPort() == B + * .getPort() || B.getPort() == 0. + */ + static class HostSet implements Iterable { + // Host -> lists of ports + private final Multimap addrs = HashMultimap.create(); + + /** + * The function that checks whether there exists an entry foo in the set + * so that foo <= addr. + */ + boolean matchedBy(InetSocketAddress addr) { + Collection ports = addrs.get(addr.getAddress()); + return addr.getPort() == 0 ? !ports.isEmpty() : ports.contains(addr + .getPort()); + } + + /** + * The function that checks whether there exists an entry foo in the set + * so that addr <= foo. + */ + boolean match(InetSocketAddress addr) { + int port = addr.getPort(); + Collection ports = addrs.get(addr.getAddress()); + boolean exactMatch = ports.contains(port); + boolean genericMatch = ports.contains(0); + return exactMatch || genericMatch; + } + + boolean isEmpty() { + return addrs.isEmpty(); + } + + int size() { + return addrs.size(); + } + + void add(InetSocketAddress addr) { + Preconditions.checkArgument(!addr.isUnresolved()); + addrs.put(addr.getAddress(), addr.getPort()); + } + + @Override + public Iterator iterator() { + return new UnmodifiableIterator() { + private final Iterator> it = addrs.entries().iterator(); + + @Override + public boolean hasNext() { + return it.hasNext(); + } + + @Override + public InetSocketAddress next() { + Map.Entry e = it.next(); + return new InetSocketAddress(e.getKey(), e.getValue()); + } + }; + } + + @Override + public String toString() { + StringBuilder sb = new StringBuilder("HostSet("); + Joiner.on(",").appendTo(sb, Iterators.transform(iterator(), + new Function() { + @Override + public String apply(@Nullable InetSocketAddress addr) { + assert addr != null; + return addr.getAddress().getHostAddress() + ":" + addr.getPort(); + } + })); + return sb.append(")").toString(); + } + } +} diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/HostFileManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/HostFileManager.java deleted file mode 100644 index a157ce961b8..00000000000 --- a/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/namenode/HostFileManager.java +++ /dev/null @@ -1,358 +0,0 @@ -/** - * Licensed to the Apache Software Foundation (ASF) under one - * or more contributor license agreements. See the NOTICE file - * distributed with this work for additional information - * regarding copyright ownership. The ASF licenses this file - * to you under the Apache License, Version 2.0 (the - * "License"); you may not use this file except in compliance - * with the License. You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -package org.apache.hadoop.hdfs.server.namenode; - -import java.io.IOException; -import java.net.InetAddress; -import java.net.UnknownHostException; -import java.util.HashMap; -import java.util.HashSet; -import java.util.Iterator; -import java.util.Map; -import java.util.TreeMap; - -import org.apache.commons.logging.Log; -import org.apache.commons.logging.LogFactory; -import org.apache.hadoop.hdfs.protocol.DatanodeID; -import org.apache.hadoop.util.HostsFileReader; - -/** - * This class manages the include and exclude files for HDFS. - * - * These files control which DataNodes the NameNode expects to see in the - * cluster. Loosely speaking, the include file, if it exists and is not - * empty, is a list of everything we expect to see. The exclude file is - * a list of everything we want to ignore if we do see it. - * - * Entries may or may not specify a port. If they don't, we consider - * them to apply to every DataNode on that host. For example, putting - * 192.168.0.100 in the excludes file blacklists both 192.168.0.100:5000 and - * 192.168.0.100:6000. This case comes up in unit tests. - * - * When reading the hosts files, we try to find the IP address for each - * entry. This is important because it allows us to de-duplicate entries. - * If the user specifies a node as foo.bar.com in the include file, but - * 192.168.0.100 in the exclude file, we need to realize that these are - * the same node. Resolving the IP address also allows us to give more - * information back to getDatanodeListForReport, which makes the web UI - * look nicer (among other things.) See HDFS-3934 for more details. - * - * DNS resolution can be slow. For this reason, we ONLY do it when (re)reading - * the hosts files. In all other cases, we rely on the cached values either - * in the DatanodeID objects, or in HostFileManager#Entry. - * We also don't want to be holding locks when doing this. - * See HDFS-3990 for more discussion of DNS overheads. - * - * Not all entries in the hosts files will have an associated IP address. - * Some entries may be "registration names." The "registration name" of - * a DataNode is either the actual hostname, or an arbitrary string configured - * by dfs.datanode.hostname. It's possible to add registration names to the - * include or exclude files. If we can't find an IP address associated with - * a host file entry, we assume it's a registered hostname and act accordingly. - * The "registration name" feature is a little odd and it may be removed in the - * future (I hope?) - */ -public class HostFileManager { - private static final Log LOG = LogFactory.getLog(HostFileManager.class); - - public static class Entry { - /** - * This what the user put on the line before the colon, or the whole line - * if there is no colon. - */ - private final String prefix; - - /** - * This is the port which was specified after the colon. It is 0 if no - * port was given. - */ - private final int port; - - /** - * If we can resolve the IP address, this is it. Otherwise, it is the - * empty string. - */ - private final String ipAddress; - - /** - * Parse a hosts file Entry. - */ - static Entry parse(String fileName, String entry) throws IOException { - final String prefix; - final int port; - String ipAddress = ""; - - int idx = entry.indexOf(':'); - if (-1 == idx) { - prefix = entry; - port = 0; - } else { - prefix = entry.substring(0, idx); - String portStr = entry.substring(idx + 1); - try { - port = Integer.parseInt(portStr); - } catch (NumberFormatException e) { - throw new IOException("unable to parse port number for " + - "'" + entry + "'", e); - } - } - try { - // Let's see if we can resolve this prefix to an IP address. - // This may fail; one example is with a registered hostname - // which is not actually a real DNS name. - InetAddress addr = InetAddress.getByName(prefix); - ipAddress = addr.getHostAddress(); - } catch (UnknownHostException e) { - LOG.info("When reading " + fileName + ", could not look up " + - "IP address for " + prefix + ". We will assume this is a " + - "registration name.", e); - } - return new Entry(prefix, port, ipAddress); - } - - public String getIdentifier() { - return ipAddress.isEmpty() ? prefix : ipAddress; - } - - public Entry(String prefix, int port, String ipAddress) { - this.prefix = prefix; - this.port = port; - this.ipAddress = ipAddress; - } - - public String getPrefix() { - return prefix; - } - - public int getPort() { - return port; - } - - public String getIpAddress() { - return ipAddress; - } - - public String toString() { - StringBuilder bld = new StringBuilder(); - bld.append("Entry{").append(prefix).append(", port="). - append(port).append(", ipAddress=").append(ipAddress).append("}"); - return bld.toString(); - } - } - - public static class EntrySet implements Iterable { - /** - * The index. Each Entry appears in here exactly once. - * - * It may be indexed by one of: - * ipAddress:port - * ipAddress - * registeredHostname:port - * registeredHostname - * - * The different indexing strategies reflect the fact that we may or may - * not have a port or IP address for each entry. - */ - final TreeMap index = new TreeMap(); - - public boolean isEmpty() { - return index.isEmpty(); - } - - public Entry find(DatanodeID datanodeID) { - Entry entry; - int xferPort = datanodeID.getXferPort(); - assert(xferPort > 0); - String datanodeIpAddr = datanodeID.getIpAddr(); - if (datanodeIpAddr != null) { - entry = index.get(datanodeIpAddr + ":" + xferPort); - if (entry != null) { - return entry; - } - entry = index.get(datanodeIpAddr); - if (entry != null) { - return entry; - } - } - String registeredHostName = datanodeID.getHostName(); - if (registeredHostName != null) { - entry = index.get(registeredHostName + ":" + xferPort); - if (entry != null) { - return entry; - } - entry = index.get(registeredHostName); - if (entry != null) { - return entry; - } - } - return null; - } - - public Entry find(Entry toFind) { - int port = toFind.getPort(); - if (port != 0) { - return index.get(toFind.getIdentifier() + ":" + port); - } else { - // An Entry with no port matches any entry with the same identifer. - // In other words, we treat 0 as "any port." - Map.Entry ceil = - index.ceilingEntry(toFind.getIdentifier()); - if ((ceil != null) && - (ceil.getValue().getIdentifier().equals( - toFind.getIdentifier()))) { - return ceil.getValue(); - } - return null; - } - } - - public String toString() { - StringBuilder bld = new StringBuilder(); - - bld.append("HostSet("); - for (Map.Entry entry : index.entrySet()) { - bld.append("\n\t"); - bld.append(entry.getKey()).append("->"). - append(entry.getValue().toString()); - } - bld.append("\n)"); - return bld.toString(); - } - - @Override - public Iterator iterator() { - return index.values().iterator(); - } - } - - public static class MutableEntrySet extends EntrySet { - public void add(DatanodeID datanodeID) { - Entry entry = new Entry(datanodeID.getHostName(), - datanodeID.getXferPort(), datanodeID.getIpAddr()); - index.put(datanodeID.getIpAddr() + ":" + datanodeID.getXferPort(), - entry); - } - - public void add(Entry entry) { - int port = entry.getPort(); - if (port != 0) { - index.put(entry.getIdentifier() + ":" + port, entry); - } else { - index.put(entry.getIdentifier(), entry); - } - } - - void readFile(String type, String filename) throws IOException { - if (filename.isEmpty()) { - return; - } - HashSet entrySet = new HashSet(); - HostsFileReader.readFileToSet(type, filename, entrySet); - for (String str : entrySet) { - Entry entry = Entry.parse(filename, str); - add(entry); - } - } - } - - private EntrySet includes = new EntrySet(); - private EntrySet excludes = new EntrySet(); - - public HostFileManager() { - } - - public void refresh(String includeFile, String excludeFile) - throws IOException { - MutableEntrySet newIncludes = new MutableEntrySet(); - IOException includeException = null; - try { - newIncludes.readFile("included", includeFile); - } catch (IOException e) { - includeException = e; - } - MutableEntrySet newExcludes = new MutableEntrySet(); - IOException excludeException = null; - try { - newExcludes.readFile("excluded", excludeFile); - } catch (IOException e) { - excludeException = e; - } - synchronized(this) { - if (includeException == null) { - includes = newIncludes; - } - if (excludeException == null) { - excludes = newExcludes; - } - } - if (includeException == null) { - LOG.info("read includes:\n" + newIncludes); - } else { - LOG.error("failed to read include file '" + includeFile + "'. " + - "Continuing to use previous include list.", - includeException); - } - if (excludeException == null) { - LOG.info("read excludes:\n" + newExcludes); - } else { - LOG.error("failed to read exclude file '" + excludeFile + "'." + - "Continuing to use previous exclude list.", - excludeException); - } - if (includeException != null) { - throw new IOException("error reading hosts file " + includeFile, - includeException); - } - if (excludeException != null) { - throw new IOException("error reading exclude file " + excludeFile, - excludeException); - } - } - - public synchronized boolean isIncluded(DatanodeID dn) { - if (includes.isEmpty()) { - // If the includes list is empty, act as if everything is in the - // includes list. - return true; - } else { - return includes.find(dn) != null; - } - } - - public synchronized boolean isExcluded(DatanodeID dn) { - return excludes.find(dn) != null; - } - - public synchronized boolean hasIncludes() { - return !includes.isEmpty(); - } - - /** - * @return the includes as an immutable set. - */ - public synchronized EntrySet getIncludes() { - return includes; - } - - /** - * @return the excludes as an immutable set. - */ - public synchronized EntrySet getExcludes() { - return excludes; - } -} diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDatanodeRegistration.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDatanodeRegistration.java index 21cf746e623..0e5d974ddeb 100644 --- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDatanodeRegistration.java +++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDatanodeRegistration.java @@ -33,6 +33,7 @@ import org.apache.hadoop.test.GenericTestUtils; import org.apache.hadoop.util.VersionInfo; import org.junit.Test; +import java.net.InetAddress; import java.net.InetSocketAddress; import java.security.Permission; @@ -226,6 +227,7 @@ public class TestDatanodeRegistration { DatanodeRegistration mockDnReg = mock(DatanodeRegistration.class); doReturn(HdfsConstants.DATANODE_LAYOUT_VERSION).when(mockDnReg).getVersion(); + doReturn("127.0.0.1").when(mockDnReg).getIpAddr(); doReturn(123).when(mockDnReg).getXferPort(); doReturn("fake-storage-id").when(mockDnReg).getDatanodeUuid(); doReturn(mockStorageInfo).when(mockDnReg).getStorageInfo(); @@ -280,6 +282,7 @@ public class TestDatanodeRegistration { // Should succeed when software versions are the same and CTimes are the // same. doReturn(VersionInfo.getVersion()).when(mockDnReg).getSoftwareVersion(); + doReturn("127.0.0.1").when(mockDnReg).getIpAddr(); doReturn(123).when(mockDnReg).getXferPort(); rpcServer.registerDatanode(mockDnReg); diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDecommission.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDecommission.java index 25cec12e23a..afba5cad42a 100644 --- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDecommission.java +++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDecommission.java @@ -25,9 +25,7 @@ import java.io.IOException; import java.util.ArrayList; import java.util.Arrays; import java.util.Collection; -import java.util.HashMap; import java.util.Iterator; -import java.util.Map; import java.util.Random; import org.apache.commons.logging.Log; @@ -45,7 +43,6 @@ import org.apache.hadoop.hdfs.protocol.HdfsConstants.DatanodeReportType; import org.apache.hadoop.hdfs.protocol.LocatedBlock; import org.apache.hadoop.hdfs.protocol.LocatedBlocks; import org.apache.hadoop.hdfs.server.namenode.FSNamesystem; -import org.apache.hadoop.hdfs.server.namenode.HostFileManager; import org.apache.hadoop.hdfs.server.namenode.NameNode; import org.apache.hadoop.hdfs.server.namenode.NameNodeAdapter; import org.apache.hadoop.test.PathUtils; @@ -639,149 +636,6 @@ public class TestDecommission { assertEquals(bogusIp, info[1].getHostName()); } } - - @Test(timeout=360000) - public void testDuplicateHostsEntries() throws IOException, - InterruptedException { - Configuration hdfsConf = new Configuration(conf); - cluster = new MiniDFSCluster.Builder(hdfsConf) - .numDataNodes(1).setupHostsFile(true).build(); - cluster.waitActive(); - int dnPort = cluster.getDataNodes().get(0).getXferPort(); - - // pick some random ports that don't overlap with our DN's port - // or with each other. - Random random = new Random(System.currentTimeMillis()); - int port1 = dnPort; - while (port1 == dnPort) { - port1 = random.nextInt(6000) + 1000; - } - int port2 = dnPort; - while ((port2 == dnPort) || (port2 == port1)) { - port2 = random.nextInt(6000) + 1000; - } - - // Now empty hosts file and ensure the datanode is disallowed - // from talking to namenode, resulting in it's shutdown. - ArrayList nodes = new ArrayList(); - - // These entries will be de-duped by the NameNode, since they refer - // to the same IP address + port combo. - nodes.add("127.0.0.1:" + port1); - nodes.add("localhost:" + port1); - nodes.add("127.0.0.1:" + port1); - - // The following entries should not be de-duped. - nodes.add("127.0.0.1:" + port2); - nodes.add("127.0.30.1:" + port1); - writeConfigFile(hostsFile, nodes); - - refreshNodes(cluster.getNamesystem(0), hdfsConf); - - DFSClient client = getDfsClient(cluster.getNameNode(0), hdfsConf); - DatanodeInfo[] info = client.datanodeReport(DatanodeReportType.LIVE); - for (int i = 0 ; i < 5 && info.length != 0; i++) { - LOG.info("Waiting for datanode to be marked dead"); - Thread.sleep(HEARTBEAT_INTERVAL * 1000); - info = client.datanodeReport(DatanodeReportType.LIVE); - } - assertEquals("Number of live nodes should be 0", 0, info.length); - - // Test that non-live and bogus hostnames are considered "dead". - // The dead report should have an entry for (1) the DN that is - // now considered dead because it is no longer allowed to connect - // and (2) the bogus entries in the hosts file. - DatanodeInfo deadDns[] = client.datanodeReport(DatanodeReportType.DEAD); - HashMap deadByXferAddr = - new HashMap(); - for (DatanodeInfo dn : deadDns) { - LOG.info("DEAD DatanodeInfo: xferAddr = " + dn.getXferAddr() + - ", ipAddr = " + dn.getIpAddr() + - ", hostname = " + dn.getHostName()); - deadByXferAddr.put(dn.getXferAddr(), dn); - } - // The real DataNode should be included in the list. - String realDnIpPort = cluster.getDataNodes().get(0). - getXferAddress().getAddress().getHostAddress() + ":" + - cluster.getDataNodes().get(0).getXferPort(); - Assert.assertNotNull("failed to find real datanode IP " + realDnIpPort, - deadByXferAddr.remove(realDnIpPort)); - // The fake datanode with address 127.0.30.1 should be included in this list. - Assert.assertNotNull(deadByXferAddr.remove( - "127.0.30.1:" + port1)); - // Now look for the two copies of 127.0.0.1 with port1 and port2. - Iterator> iter = - deadByXferAddr.entrySet().iterator(); - boolean foundPort1 = false, foundPort2 = false; - while (iter.hasNext()) { - Map.Entry entry = iter.next(); - DatanodeInfo dn = entry.getValue(); - if (dn.getXferPort() == port1) { - foundPort1 = true; - iter.remove(); - } else if (dn.getXferPort() == port2) { - foundPort2 = true; - iter.remove(); - } - } - Assert.assertTrue("did not find a dead entry with port " + port1, - foundPort1); - Assert.assertTrue("did not find a dead entry with port " + port2, - foundPort2); - Assert.assertTrue(deadByXferAddr.isEmpty()); - } - - @Test(timeout=360000) - public void testIncludeByRegistrationName() throws IOException, - InterruptedException { - Configuration hdfsConf = new Configuration(conf); - final String registrationName = "--registration-name--"; - final String nonExistentDn = "127.0.0.40"; - hdfsConf.set(DFSConfigKeys.DFS_DATANODE_HOST_NAME_KEY, registrationName); - cluster = new MiniDFSCluster.Builder(hdfsConf) - .numDataNodes(1).checkDataNodeHostConfig(true) - .setupHostsFile(true).build(); - cluster.waitActive(); - - // Set up an includes file that doesn't have our datanode. - ArrayList nodes = new ArrayList(); - nodes.add(nonExistentDn); - writeConfigFile(hostsFile, nodes); - refreshNodes(cluster.getNamesystem(0), hdfsConf); - - // Wait for the DN to be marked dead. - DFSClient client = getDfsClient(cluster.getNameNode(0), hdfsConf); - while (true) { - DatanodeInfo info[] = client.datanodeReport(DatanodeReportType.DEAD); - if (info.length == 1) { - break; - } - LOG.info("Waiting for datanode to be marked dead"); - Thread.sleep(HEARTBEAT_INTERVAL * 1000); - } - - // Use a non-empty include file with our registration name. - // It should work. - int dnPort = cluster.getDataNodes().get(0).getXferPort(); - nodes = new ArrayList(); - nodes.add(registrationName + ":" + dnPort); - writeConfigFile(hostsFile, nodes); - refreshNodes(cluster.getNamesystem(0), hdfsConf); - cluster.restartDataNode(0); - - // Wait for the DN to come back. - while (true) { - DatanodeInfo info[] = client.datanodeReport(DatanodeReportType.LIVE); - if (info.length == 1) { - Assert.assertFalse(info[0].isDecommissioned()); - Assert.assertFalse(info[0].isDecommissionInProgress()); - assertEquals(registrationName, info[0].getHostName()); - break; - } - LOG.info("Waiting for datanode to come back"); - Thread.sleep(HEARTBEAT_INTERVAL * 1000); - } - } @Test(timeout=120000) public void testDecommissionWithOpenfile() throws IOException, InterruptedException { diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestHostFileManager.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestHostFileManager.java new file mode 100644 index 00000000000..5435572e00e --- /dev/null +++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/blockmanagement/TestHostFileManager.java @@ -0,0 +1,157 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.hadoop.hdfs.server.blockmanagement; + +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.hdfs.protocol.DatanodeID; +import org.apache.hadoop.hdfs.protocol.HdfsConstants; +import org.apache.hadoop.hdfs.server.namenode.FSNamesystem; +import org.junit.Assert; +import org.junit.Test; +import org.mockito.internal.util.reflection.Whitebox; + +import java.io.IOException; +import java.net.InetSocketAddress; +import java.util.Map; + +import static org.mockito.Mockito.doReturn; +import static org.mockito.Mockito.mock; + +public class TestHostFileManager { + private static InetSocketAddress entry(String e) { + return HostFileManager.parseEntry("dummy", "dummy", e); + } + + @Test + public void testDeduplication() { + HostFileManager.HostSet s = new HostFileManager.HostSet(); + // These entries will be de-duped, since they refer to the same IP + // address + port combo. + s.add(entry("127.0.0.1:12345")); + s.add(entry("localhost:12345")); + Assert.assertEquals(1, s.size()); + s.add(entry("127.0.0.1:12345")); + Assert.assertEquals(1, s.size()); + + // The following entries should not be de-duped. + s.add(entry("127.0.0.1:12346")); + Assert.assertEquals(2, s.size()); + s.add(entry("127.0.0.1")); + Assert.assertEquals(3, s.size()); + s.add(entry("127.0.0.10")); + Assert.assertEquals(4, s.size()); + } + + @Test + public void testRelation() { + HostFileManager.HostSet s = new HostFileManager.HostSet(); + s.add(entry("127.0.0.1:123")); + Assert.assertTrue(s.match(entry("127.0.0.1:123"))); + Assert.assertFalse(s.match(entry("127.0.0.1:12"))); + Assert.assertFalse(s.match(entry("127.0.0.1"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.1:12"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1:123"))); + Assert.assertFalse(s.match(entry("127.0.0.2"))); + Assert.assertFalse(s.match(entry("127.0.0.2:123"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.2"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.2:123"))); + + s.add(entry("127.0.0.1")); + Assert.assertTrue(s.match(entry("127.0.0.1:123"))); + Assert.assertTrue(s.match(entry("127.0.0.1:12"))); + Assert.assertTrue(s.match(entry("127.0.0.1"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.1:12"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1:123"))); + Assert.assertFalse(s.match(entry("127.0.0.2"))); + Assert.assertFalse(s.match(entry("127.0.0.2:123"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.2"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.2:123"))); + + s.add(entry("127.0.0.2:123")); + Assert.assertTrue(s.match(entry("127.0.0.1:123"))); + Assert.assertTrue(s.match(entry("127.0.0.1:12"))); + Assert.assertTrue(s.match(entry("127.0.0.1"))); + Assert.assertFalse(s.matchedBy(entry("127.0.0.1:12"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.1:123"))); + Assert.assertFalse(s.match(entry("127.0.0.2"))); + Assert.assertTrue(s.match(entry("127.0.0.2:123"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.2"))); + Assert.assertTrue(s.matchedBy(entry("127.0.0.2:123"))); + } + + @Test + @SuppressWarnings("unchecked") + public void testIncludeExcludeLists() throws IOException { + BlockManager bm = mock(BlockManager.class); + FSNamesystem fsn = mock(FSNamesystem.class); + Configuration conf = new Configuration(); + HostFileManager hm = mock(HostFileManager.class); + HostFileManager.HostSet includedNodes = new HostFileManager.HostSet(); + HostFileManager.HostSet excludedNodes = new HostFileManager.HostSet(); + + includedNodes.add(entry("127.0.0.1:12345")); + includedNodes.add(entry("localhost:12345")); + includedNodes.add(entry("127.0.0.1:12345")); + + includedNodes.add(entry("127.0.0.2")); + excludedNodes.add(entry("127.0.0.1:12346")); + excludedNodes.add(entry("127.0.30.1:12346")); + + Assert.assertEquals(2, includedNodes.size()); + Assert.assertEquals(2, excludedNodes.size()); + + doReturn(includedNodes).when(hm).getIncludes(); + doReturn(excludedNodes).when(hm).getExcludes(); + + DatanodeManager dm = new DatanodeManager(bm, fsn, conf); + Whitebox.setInternalState(dm, "hostFileManager", hm); + Map dnMap = (Map) Whitebox.getInternalState(dm, "datanodeMap"); + + // After the de-duplication, there should be only one DN from the included + // nodes declared as dead. + Assert.assertEquals(2, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.ALL).size()); + Assert.assertEquals(2, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + dnMap.put("uuid-foo", new DatanodeDescriptor(new DatanodeID("127.0.0.1", + "localhost", "uuid-foo", 12345, 1020, 1021, 1022))); + Assert.assertEquals(1, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + dnMap.put("uuid-bar", new DatanodeDescriptor(new DatanodeID("127.0.0.2", + "127.0.0.2", "uuid-bar", 12345, 1020, 1021, 1022))); + Assert.assertEquals(0, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + DatanodeDescriptor spam = new DatanodeDescriptor(new DatanodeID("127.0.0" + + ".3", "127.0.0.3", "uuid-spam", 12345, 1020, 1021, 1022)); + spam.setLastUpdate(0); + includedNodes.add(entry("127.0.0.3:12345")); + dnMap.put("uuid-spam", spam); + Assert.assertEquals(1, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + dnMap.remove("uuid-spam"); + Assert.assertEquals(1, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + excludedNodes.add(entry("127.0.0.3")); + Assert.assertEquals(0, dm.getDatanodeListForReport(HdfsConstants + .DatanodeReportType.DEAD).size()); + } +} diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/NNThroughputBenchmark.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/NNThroughputBenchmark.java index a79196c395b..d63439be797 100644 --- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/NNThroughputBenchmark.java +++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/server/namenode/NNThroughputBenchmark.java @@ -25,6 +25,7 @@ import java.util.Arrays; import java.util.EnumSet; import java.util.List; +import com.google.common.base.Preconditions; import org.apache.commons.logging.Log; import org.apache.commons.logging.LogFactory; import org.apache.commons.logging.impl.Log4JLogger; @@ -895,16 +896,9 @@ public class NNThroughputBenchmark implements Tool { long[] blockReportList; final int dnIdx; - /** - * Return a a 6 digit integer port. - * This is necessary in order to provide lexocographic ordering. - * Host names are all the same, the ordering goes by port numbers. - */ private static int getNodePort(int num) throws IOException { - int port = 100000 + num; - if (String.valueOf(port).length() > 6) { - throw new IOException("Too many data-nodes"); - } + int port = 1 + num; + Preconditions.checkState(port < Short.MAX_VALUE); return port; }