HDFS-3869. Expose non-file journal manager details in web UI. Contributed by Todd Lipcon.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/HDFS-3077@1380978 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Todd Lipcon 2012-09-05 04:16:02 +00:00
parent 8021d9199f
commit 72485f3112
10 changed files with 165 additions and 9 deletions

View File

@ -38,3 +38,5 @@ HDFS-3845. Fixes for edge cases in QJM recovery protocol (todd)
HDFS-3877. QJM: Provide defaults for dfs.journalnode.*address (eli) HDFS-3877. QJM: Provide defaults for dfs.journalnode.*address (eli)
HDFS-3863. Track last "committed" txid in QJM (todd) HDFS-3863. Track last "committed" txid in QJM (todd)
HDFS-3869. Expose non-file journal manager details in web UI (todd)

View File

@ -140,4 +140,10 @@ interface AsyncLogger {
* after this point, and any in-flight RPCs may throw an exception. * after this point, and any in-flight RPCs may throw an exception.
*/ */
public void close(); public void close();
/**
* Append an HTML-formatted report for this logger's status to the provided
* StringBuilder. This is displayed on the NN web UI.
*/
public void appendHtmlReport(StringBuilder sb);
} }

View File

@ -33,6 +33,7 @@ import org.apache.hadoop.hdfs.qjournal.protocol.QJournalProtocolProtos.SegmentSt
import org.apache.hadoop.hdfs.server.protocol.NamespaceInfo; import org.apache.hadoop.hdfs.server.protocol.NamespaceInfo;
import org.apache.hadoop.hdfs.server.protocol.RemoteEditLogManifest; import org.apache.hadoop.hdfs.server.protocol.RemoteEditLogManifest;
import org.apache.hadoop.ipc.RemoteException; import org.apache.hadoop.ipc.RemoteException;
import org.apache.jasper.compiler.JspUtil;
import com.google.common.annotations.VisibleForTesting; import com.google.common.annotations.VisibleForTesting;
import com.google.common.base.Preconditions; import com.google.common.base.Preconditions;
@ -190,6 +191,24 @@ class AsyncLoggerSet {
int size() { int size() {
return loggers.size(); return loggers.size();
} }
/**
* Append an HTML-formatted status readout on the current
* state of the underlying loggers.
* @param sb the StringBuilder to append to
*/
void appendHtmlReport(StringBuilder sb) {
sb.append("<table class=\"storage\">");
sb.append("<thead><tr><td>JN</td><td>Status</td></tr></thead>\n");
for (AsyncLogger l : loggers) {
sb.append("<tr>");
sb.append("<td>" + JspUtil.escapeXml(l.toString()) + "</td>");
sb.append("<td>");
l.appendHtmlReport(sb);
sb.append("</td></tr>\n");
}
sb.append("</table>");
}
/** /**
* @return the (mutable) list of loggers, for use in tests to * @return the (mutable) list of loggers, for use in tests to

View File

@ -84,7 +84,12 @@ public class IPCLoggerChannel implements AsyncLogger {
* The number of bytes of edits data still in the queue. * The number of bytes of edits data still in the queue.
*/ */
private int queuedEditsSizeBytes = 0; private int queuedEditsSizeBytes = 0;
/**
* The highest txid that has been successfully logged on the remote JN.
*/
private long highestAckedTxId = 0;
/** /**
* The maximum number of bytes that can be pending in the queue. * The maximum number of bytes that can be pending in the queue.
* This keeps the writer from hitting OOME if one of the loggers * This keeps the writer from hitting OOME if one of the loggers
@ -262,6 +267,9 @@ public class IPCLoggerChannel implements AsyncLogger {
public Void call() throws IOException { public Void call() throws IOException {
getProxy().journal(createReqInfo(), getProxy().journal(createReqInfo(),
segmentTxId, firstTxnId, numTxns, data); segmentTxId, firstTxnId, numTxns, data);
synchronized (IPCLoggerChannel.this) {
highestAckedTxId = firstTxnId + numTxns - 1;
}
return null; return null;
} }
}); });
@ -398,4 +406,14 @@ public class IPCLoggerChannel implements AsyncLogger {
public String toString() { public String toString() {
return "Channel to journal node " + addr; return "Channel to journal node " + addr;
} }
}
@Override
public synchronized void appendHtmlReport(StringBuilder sb) {
sb.append("Written txid ").append(highestAckedTxId);
long behind = committedTxId - highestAckedTxId;
assert behind >= 0;
if (behind > 0) {
sb.append(" (" + behind + " behind)");
}
}
}

View File

@ -109,4 +109,13 @@ class QuorumOutputStream extends EditLogOutputStream {
loggers.setCommittedTxId(firstTxToFlush + numReadyTxns - 1); loggers.setCommittedTxId(firstTxToFlush + numReadyTxns - 1);
} }
} }
@Override
public String generateHtmlReport() {
StringBuilder sb = new StringBuilder();
sb.append("Writing segment beginning at txid " + segmentTxId + "<br/>\n");
loggers.appendHtmlReport(sb);
return sb.toString();
}
} }

View File

@ -24,6 +24,7 @@ import static org.apache.hadoop.util.Time.now;
import org.apache.hadoop.classification.InterfaceAudience; import org.apache.hadoop.classification.InterfaceAudience;
import org.apache.hadoop.classification.InterfaceStability; import org.apache.hadoop.classification.InterfaceStability;
import org.apache.jasper.compiler.JspUtil;
/** /**
* A generic abstract class to support journaling of edits logs into * A generic abstract class to support journaling of edits logs into
@ -132,4 +133,12 @@ public abstract class EditLogOutputStream implements Closeable {
protected long getNumSync() { protected long getNumSync() {
return numSync; return numSync;
} }
/**
* @return a short HTML snippet suitable for describing the current
* status of the stream
*/
public String generateHtmlReport() {
return JspUtil.escapeXml(this.toString());
}
} }

View File

@ -26,6 +26,7 @@ import java.util.LinkedList;
import java.util.List; import java.util.List;
import java.util.PriorityQueue; import java.util.PriorityQueue;
import java.util.SortedSet; import java.util.SortedSet;
import java.util.concurrent.CopyOnWriteArrayList;
import org.apache.commons.logging.Log; import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory; import org.apache.commons.logging.LogFactory;
@ -147,7 +148,7 @@ public class JournalSet implements JournalManager {
return journal; return journal;
} }
private boolean isDisabled() { boolean isDisabled() {
return disabled; return disabled;
} }
@ -165,8 +166,12 @@ public class JournalSet implements JournalManager {
return required; return required;
} }
} }
private List<JournalAndStream> journals = Lists.newArrayList(); // COW implementation is necessary since some users (eg the web ui) call
// getAllJournalStreams() and then iterate. Since this is rarely
// mutated, there is no performance concern.
private List<JournalAndStream> journals =
new CopyOnWriteArrayList<JournalSet.JournalAndStream>();
final int minimumRedundantJournals; final int minimumRedundantJournals;
JournalSet(int minimumRedundantResources) { JournalSet(int minimumRedundantResources) {
@ -519,7 +524,6 @@ public class JournalSet implements JournalManager {
} }
} }
@VisibleForTesting
List<JournalAndStream> getAllJournalStreams() { List<JournalAndStream> getAllJournalStreams() {
return journals; return journals;
} }

View File

@ -48,6 +48,7 @@ import org.apache.hadoop.hdfs.server.blockmanagement.DatanodeManager;
import org.apache.hadoop.hdfs.server.common.JspHelper; import org.apache.hadoop.hdfs.server.common.JspHelper;
import org.apache.hadoop.hdfs.server.common.Storage; import org.apache.hadoop.hdfs.server.common.Storage;
import org.apache.hadoop.hdfs.server.common.Storage.StorageDirectory; import org.apache.hadoop.hdfs.server.common.Storage.StorageDirectory;
import org.apache.hadoop.hdfs.server.namenode.JournalSet.JournalAndStream;
import org.apache.hadoop.hdfs.server.protocol.NamenodeProtocols; import org.apache.hadoop.hdfs.server.protocol.NamenodeProtocols;
import org.apache.hadoop.http.HttpConfig; import org.apache.hadoop.http.HttpConfig;
import org.apache.hadoop.io.Text; import org.apache.hadoop.io.Text;
@ -61,6 +62,8 @@ import org.apache.hadoop.util.Time;
import org.apache.hadoop.util.VersionInfo; import org.apache.hadoop.util.VersionInfo;
import org.znerd.xmlenc.XMLOutputter; import org.znerd.xmlenc.XMLOutputter;
import com.google.common.base.Preconditions;
class NamenodeJspHelper { class NamenodeJspHelper {
static String getSafeModeText(FSNamesystem fsn) { static String getSafeModeText(FSNamesystem fsn) {
if (!fsn.isInSafeMode()) if (!fsn.isInSafeMode())
@ -213,6 +216,52 @@ class NamenodeJspHelper {
out.print("</table></div>\n"); out.print("</table></div>\n");
} }
/**
* Generate an HTML report containing the current status of the HDFS
* journals.
*/
void generateJournalReport(JspWriter out, NameNode nn,
HttpServletRequest request) throws IOException {
FSEditLog log = nn.getFSImage().getEditLog();
Preconditions.checkArgument(log != null, "no edit log set in %s", nn);
out.println("<h3> " + nn.getRole() + " Journal Status: </h3>");
out.println("<b>Current transaction ID:</b> " +
nn.getFSImage().getLastAppliedOrWrittenTxId() + "<br/>");
boolean openForWrite = log.isOpenForWrite();
out.println("<div class=\"dfstable\">");
out.println("<table class=\"storage\" title=\"NameNode Journals\">\n"
+ "<thead><tr><td><b>Journal Manager</b></td><td><b>State</b></td></tr></thead>");
for (JournalAndStream jas : log.getJournals()) {
out.print("<tr>");
out.print("<td>" + jas.getManager());
if (jas.isRequired()) {
out.print(" [required]");
}
out.print("</td><td>");
if (jas.isDisabled()) {
out.print("<span class=\"failed\">Failed</span>");
} else if (openForWrite) {
EditLogOutputStream elos = jas.getCurrentStream();
if (elos != null) {
out.println(elos.generateHtmlReport());
} else {
out.println("not currently writing");
}
} else {
out.println("open for read");
}
out.println("</td></tr>");
}
out.println("</table></div>");
}
void generateHealthReport(JspWriter out, NameNode nn, void generateHealthReport(JspWriter out, NameNode nn,
HttpServletRequest request) throws IOException { HttpServletRequest request) throws IOException {

View File

@ -60,8 +60,10 @@
<%= NamenodeJspHelper.getCorruptFilesWarning(fsn)%> <%= NamenodeJspHelper.getCorruptFilesWarning(fsn)%>
<% healthjsp.generateHealthReport(out, nn, request); %> <% healthjsp.generateHealthReport(out, nn, request); %>
<hr> <% healthjsp.generateJournalReport(out, nn, request); %>
<hr/>
<% healthjsp.generateConfReport(out, nn, request); %> <% healthjsp.generateConfReport(out, nn, request); %>
<hr>
<% <%
out.println(ServletUtil.htmlFooter()); out.println(ServletUtil.htmlFooter());
%> %>

View File

@ -17,17 +17,18 @@
*/ */
package org.apache.hadoop.hdfs.qjournal; package org.apache.hadoop.hdfs.qjournal;
import static org.junit.Assert.assertTrue; import static org.junit.Assert.*;
import static org.junit.Assert.fail;
import java.io.File; import java.io.File;
import java.io.IOException; import java.io.IOException;
import java.net.URL;
import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem; import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FileUtil; import org.apache.hadoop.fs.FileUtil;
import org.apache.hadoop.fs.Path; import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hdfs.DFSConfigKeys; import org.apache.hadoop.hdfs.DFSConfigKeys;
import org.apache.hadoop.hdfs.DFSTestUtil;
import org.apache.hadoop.hdfs.HdfsConfiguration; import org.apache.hadoop.hdfs.HdfsConfiguration;
import org.apache.hadoop.hdfs.MiniDFSCluster; import org.apache.hadoop.hdfs.MiniDFSCluster;
import org.apache.hadoop.hdfs.server.namenode.NameNode; import org.apache.hadoop.hdfs.server.namenode.NameNode;
@ -185,4 +186,41 @@ public class TestNNWithQJM {
"Unable to start log segment 1: too few journals", ioe); "Unable to start log segment 1: too few journals", ioe);
} }
} }
@Test
public void testWebPageHasQjmInfo() throws Exception {
conf.set(DFSConfigKeys.DFS_NAMENODE_NAME_DIR_KEY,
MiniDFSCluster.getBaseDirectory() + "/TestNNWithQJM/image");
conf.set(DFSConfigKeys.DFS_NAMENODE_EDITS_DIR_KEY,
mjc.getQuorumJournalURI("myjournal").toString());
MiniDFSCluster cluster = new MiniDFSCluster.Builder(conf)
.numDataNodes(0)
.manageNameDfsDirs(false)
.build();
try {
URL url = new URL("http://localhost:"
+ NameNode.getHttpAddress(cluster.getConfiguration(0)).getPort()
+ "/dfshealth.jsp");
cluster.getFileSystem().mkdirs(TEST_PATH);
String contents = DFSTestUtil.urlGet(url);
assertTrue(contents.contains("Channel to journal node"));
assertTrue(contents.contains("Written txid 2"));
// Stop one JN, do another txn, and make sure it shows as behind
// stuck behind the others.
mjc.getJournalNode(0).stopAndJoin(0);
cluster.getFileSystem().delete(TEST_PATH, true);
contents = DFSTestUtil.urlGet(url);
System.out.println(contents);
assertTrue(contents.contains("(1 behind)"));
} finally {
cluster.shutdown();
}
}
} }