MAPREDUCE-4989. JSONify DataTables input data for Attempts page. Contributed by Ravi Prakash

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1445448 13f79535-47bb-0310-9956-ffa450edef68
2013-02-13 01:15:48 +00:00 · 2013-02-13 01:15:48 +00:00 · c2d6407070
commit c2d6407070
parent 7e615c7f75
5 changed files with 132 additions and 140 deletions
--- a/hadoop-mapreduce-project/CHANGES.txt
+++ b/hadoop-mapreduce-project/CHANGES.txt
@ -696,6 +696,9 @@ Release 0.23.7 - UNRELEASED
    MAPREDUCE-4905. test org.apache.hadoop.mapred.pipes 
    (Aleksey Gorshkov via bobby)

+    MAPREDUCE-4989. JSONify DataTables input data for Attempts page (Ravi
+    Prakash via jlowe)
+
  OPTIMIZATIONS

    MAPREDUCE-4946. Fix a performance problem for large jobs by reducing the
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/main/java/org/apache/hadoop/mapreduce/v2/app/webapp/TaskPage.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-app/src/main/java/org/apache/hadoop/mapreduce/v2/app/webapp/TaskPage.java
@ -27,18 +27,11 @@

 import java.util.Collection;

+import org.apache.commons.lang.StringEscapeUtils;
 import org.apache.hadoop.http.HttpConfig;
 import org.apache.hadoop.mapreduce.v2.app.job.TaskAttempt;
 import org.apache.hadoop.mapreduce.v2.app.webapp.dao.TaskAttemptInfo;
-import org.apache.hadoop.util.StringUtils;
-import org.apache.hadoop.yarn.api.records.ContainerId;
-import org.apache.hadoop.yarn.util.Times;
 import org.apache.hadoop.yarn.webapp.SubView;
-import org.apache.hadoop.yarn.webapp.hamlet.Hamlet;
-import org.apache.hadoop.yarn.webapp.hamlet.Hamlet.TABLE;
-import org.apache.hadoop.yarn.webapp.hamlet.Hamlet.TBODY;
-import org.apache.hadoop.yarn.webapp.hamlet.Hamlet.TD;
-import org.apache.hadoop.yarn.webapp.hamlet.Hamlet.TR;
 import org.apache.hadoop.yarn.webapp.view.HtmlBlock;

 import com.google.inject.Inject;
@ -60,7 +53,7 @@ protected void render(Block html) {
          h2($(TITLE));
        return;
      }
-      TBODY<TABLE<Hamlet>> tbody = html.
+      html.
      table("#attempts").
        thead().
          tr().
@ -72,49 +65,46 @@ protected void render(Block html) {
            th(".tsh", "Started").
            th(".tsh", "Finished").
            th(".tsh", "Elapsed").
-            th(".note", "Note")._()._().
-        tbody();
+            th(".note", "Note")._()._();
+      // Write all the data into a JavaScript array of arrays for JQuery
+      // DataTables to display
+      StringBuilder attemptsTableData = new StringBuilder("[\n");
+
      for (TaskAttempt attempt : getTaskAttempts()) {
        TaskAttemptInfo ta = new TaskAttemptInfo(attempt, true);
-        String taid = ta.getId();
        String progress = percent(ta.getProgress() / 100);
-        ContainerId containerId = ta.getAssignedContainerId();

        String nodeHttpAddr = ta.getNode();
-        long startTime = ta.getStartTime();
-        long finishTime = ta.getFinishTime();
-        long elapsed = ta.getElapsedTime();
        String diag = ta.getNote() == null ? "" : ta.getNote();
-        TR<TBODY<TABLE<Hamlet>>> row = tbody.tr();
-        TD<TR<TBODY<TABLE<Hamlet>>>> nodeTd = row.
-          td(".id", taid).
-          td(".progress", progress).
-          td(".state", ta.getState()).td();
-        if (nodeHttpAddr == null) {
-          nodeTd._("N/A");
-        } else {
-          nodeTd.
-            a(".nodelink", url(HttpConfig.getSchemePrefix(),
-                               nodeHttpAddr), nodeHttpAddr);
-        }
-        nodeTd._();
-        if (containerId != null) {
-          String containerIdStr = ta.getAssignedContainerIdStr();
-          row.td().
-              a(".logslink", url(HttpConfig.getSchemePrefix(),
-              nodeHttpAddr, "node", "containerlogs",
-              containerIdStr, app.getJob().getUserName()), "logs")._();
-        } else {
-          row.td()._("N/A")._();
-        }
+        attemptsTableData.append("[\"")
+        .append(ta.getId()).append("\",\"")
+        .append(progress).append("\",\"")
+        .append(ta.getState().toString()).append("\",\"")

-        row.
-          td(".ts", Times.format(startTime)).
-          td(".ts", Times.format(finishTime)).
-          td(".dt", StringUtils.formatTime(elapsed)).
-          td(".note", diag)._();
+        .append(nodeHttpAddr == null ? "N/A" :
+          "<a class='nodelink' href='" + HttpConfig.getSchemePrefix() + nodeHttpAddr + "'>"
+          + nodeHttpAddr + "</a>")
+        .append("\",\"")
+
+        .append(ta.getAssignedContainerId() == null ? "N/A" :
+          "<a class='logslink' href='" + url(HttpConfig.getSchemePrefix(), nodeHttpAddr, "node"
+            , "containerlogs", ta.getAssignedContainerIdStr(), app.getJob()
+            .getUserName()) + "'>logs</a>")
+          .append("\",\"")
+
+        .append(ta.getStartTime()).append("\",\"")
+        .append(ta.getFinishTime()).append("\",\"")
+        .append(ta.getElapsedTime()).append("\",\"")
+        .append(StringEscapeUtils.escapeJavaScript(StringEscapeUtils.escapeHtml(
+          diag))).append("\"],\n");
      }
-      tbody._()._();
+      //Remove the last comma and close off the array of arrays
+      if(attemptsTableData.charAt(attemptsTableData.length() - 2) == ',') {
+        attemptsTableData.delete(attemptsTableData.length()-2, attemptsTableData.length()-1);
+      }
+      attemptsTableData.append("]");
+      html.script().$type("text/javascript").
+      _("var attemptsTableData=" + attemptsTableData)._();
    }

    protected boolean isValidRequest() {
@ -140,9 +130,24 @@ protected Collection<TaskAttempt> getTaskAttempts() {
  }

  private String attemptsTableInit() {
-    return tableInit().
-        // Sort by id upon page load
-        append(", aaSorting: [[0, 'asc']]").
-        append("}").toString();
+    return tableInit()
+    .append(", 'aaData': attemptsTableData")
+    .append(", bDeferRender: true")
+    .append(", bProcessing: true")
+    .append("\n,aoColumnDefs:[\n")
+
+    //logs column should not filterable (it includes container ID which may pollute searches)
+    .append("\n{'aTargets': [ 4 ]")
+    .append(", 'bSearchable': false }")
+
+    .append("\n, {'sType':'numeric', 'aTargets': [ 5, 6")
+    .append(" ], 'mRender': renderHadoopDate }")
+
+    .append("\n, {'sType':'numeric', 'aTargets': [ 7")
+    .append(" ], 'mRender': renderHadoopElapsedTime }]")
+
+    // Sort by id upon page load
+    .append("\n, aaSorting: [[0, 'asc']]")
+    .append("}").toString();
  }
 }
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTaskPage.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTaskPage.java
@ -29,6 +29,7 @@

 import java.util.Collection;

+import org.apache.commons.lang.StringEscapeUtils;
 import org.apache.hadoop.http.HttpConfig;
 import org.apache.hadoop.mapreduce.v2.api.records.TaskId;
 import org.apache.hadoop.mapreduce.v2.api.records.TaskType;
@ -110,13 +111,17 @@ protected void render(Block html) {
              th(".note", "Note");
      
       TBODY<TABLE<Hamlet>> tbody = headRow._()._().tbody();
-      for (TaskAttempt ta : getTaskAttempts()) {
+       // Write all the data into a JavaScript array of arrays for JQuery
+       // DataTables to display
+       StringBuilder attemptsTableData = new StringBuilder("[\n");
+
+       for (TaskAttempt ta : getTaskAttempts()) {
        String taid = MRApps.toString(ta.getID());

        String nodeHttpAddr = ta.getNodeHttpAddress();
        String containerIdString = ta.getAssignedContainerID().toString();
        String nodeIdString = ta.getAssignedContainerMgrAddress();
-        String nodeRackName = ta.getNodeRackName();        
+        String nodeRackName = ta.getNodeRackName();

        long attemptStartTime = ta.getLaunchTime();
        long shuffleFinishTime = -1;
@ -138,58 +143,43 @@ protected void render(Block html) {
        long attemptElapsed =
            Times.elapsed(attemptStartTime, attemptFinishTime, false);
        int sortId = ta.getID().getId() + (ta.getID().getTaskId().getId() * 10000);
-        
-        TR<TBODY<TABLE<Hamlet>>> row = tbody.tr();
-        TD<TR<TBODY<TABLE<Hamlet>>>> td = row.td();

-        td.br().$title(String.valueOf(sortId))._(). // sorting
-            _(taid)._().td(ta.getState().toString()).td().a(".nodelink",
-                HttpConfig.getSchemePrefix()+ nodeHttpAddr,
-                nodeRackName + "/" + nodeHttpAddr);
-        td._();
-        row.td().
-          a(".logslink",
-            url("logs", nodeIdString, containerIdString, taid, app.getJob()
-                .getUserName()), "logs")._();
-        
-        row.td().
-          br().$title(String.valueOf(attemptStartTime))._().
-            _(Times.format(attemptStartTime))._();
+        attemptsTableData.append("[\"")
+        .append(sortId + " ").append(taid).append("\",\"")
+        .append(ta.getState().toString()).append("\",\"")
+
+        .append("<a class='nodelink' href='" + HttpConfig.getSchemePrefix() + nodeHttpAddr + "'>")
+        .append(nodeRackName + "/" + nodeHttpAddr + "</a>\",\"")
+
+        .append("<a class='logslink' href='").append(url("logs", nodeIdString
+          , containerIdString, taid, app.getJob().getUserName()))
+          .append("'>logs</a>\",\"")
+
+          .append(attemptStartTime).append("\",\"");

        if(type == TaskType.REDUCE) {
-          row.td().
-            br().$title(String.valueOf(shuffleFinishTime))._().
-            _(Times.format(shuffleFinishTime))._();
-          row.td().
-          br().$title(String.valueOf(sortFinishTime))._().
-          _(Times.format(sortFinishTime))._();
+          attemptsTableData.append(shuffleFinishTime).append("\",\"")
+          .append(sortFinishTime).append("\",\"");
        }
-        row.
-            td().
-              br().$title(String.valueOf(attemptFinishTime))._().
-              _(Times.format(attemptFinishTime))._();
-        
+        attemptsTableData.append(attemptFinishTime).append("\",\"");
+
        if(type == TaskType.REDUCE) {
-          row.td().
-            br().$title(String.valueOf(elapsedShuffleTime))._().
-          _(formatTime(elapsedShuffleTime))._();
-          row.td().
-          br().$title(String.valueOf(elapsedSortTime))._().
-        _(formatTime(elapsedSortTime))._();
-          row.td().
-            br().$title(String.valueOf(elapsedReduceTime))._().
-          _(formatTime(elapsedReduceTime))._();
+          attemptsTableData.append(elapsedShuffleTime).append("\",\"")
+          .append(elapsedSortTime).append("\",\"")
+          .append(elapsedReduceTime).append("\",\"");
        }
-        
-        row.
-          td().
-            br().$title(String.valueOf(attemptElapsed))._().
-          _(formatTime(attemptElapsed))._().
-          td(".note", Joiner.on('\n').join(ta.getDiagnostics()));
-        row._();
+          attemptsTableData.append(attemptElapsed).append("\",\"")
+          .append(StringEscapeUtils.escapeJavaScript(StringEscapeUtils.escapeHtml(
+           Joiner.on('\n').join(ta.getDiagnostics())))).append("\"],\n");
      }
-      
-      
+       //Remove the last comma and close off the array of arrays
+       if(attemptsTableData.charAt(attemptsTableData.length() - 2) == ',') {
+         attemptsTableData.delete(attemptsTableData.length()-2, attemptsTableData.length()-1);
+       }
+       attemptsTableData.append("]");
+       html.script().$type("text/javascript").
+       _("var attemptsTableData=" + attemptsTableData)._();
+
      TR<TFOOT<TABLE<Hamlet>>> footRow = tbody._().tfoot().tr();
      footRow.
          th().input("search_init").$type(InputType.text).
@ -237,10 +227,6 @@ protected void render(Block html) {
      footRow._()._()._();
    }

-    private String formatTime(long elapsed) {
-      return elapsed < 0 ? "N/A" : StringUtils.formatTime(elapsed);
-    }
-    
    /**
     * @return true if this is a valid request else false.
     */
@ -292,24 +278,34 @@ private String attemptsTableInit() {
      TaskId taskID = MRApps.toTaskID($(TASK_ID));
      type = taskID.getTaskType();
    }
-    StringBuilder b = tableInit().
-      append(",aoColumnDefs:[");
+    StringBuilder b = tableInit()
+      .append(", 'aaData': attemptsTableData")
+      .append(", bDeferRender: true")
+      .append(", bProcessing: true")
+      .append("\n,aoColumnDefs:[\n")

-    b.append("{'sType':'title-numeric', 'aTargets': [ 0");
-    if(type == TaskType.REDUCE) {
-      b.append(", 7, 8, 9, 10");
-    } else { //MAP
-      b.append(", 5");
-    }
-    b.append(" ] }]");
+      //logs column should not filterable (it includes container ID which may pollute searches)
+      .append("\n{'aTargets': [ 3 ]")
+      .append(", 'bSearchable': false }")

-    // Sort by id upon page load
-    b.append(", aaSorting: [[0, 'asc']]");
+      .append("\n, {'sType':'numeric', 'aTargets': [ 0 ]")
+      .append(", 'mRender': parseHadoopAttemptID }")

-    b.append("}");
-    return b.toString();
+      .append("\n, {'sType':'numeric', 'aTargets': [ 4, 5")
+      //Column numbers are different for maps and reduces
+      .append(type == TaskType.REDUCE ? ", 6, 7" : "")
+      .append(" ], 'mRender': renderHadoopDate }")
+
+      .append("\n, {'sType':'numeric', 'aTargets': [")
+      .append(type == TaskType.REDUCE ? "8, 9, 10, 11" : "6")
+      .append(" ], 'mRender': renderHadoopElapsedTime }]")
+
+      // Sort by id upon page load
+      .append("\n, aaSorting: [[0, 'asc']]")
+      .append("}");
+      return b.toString();
  }
-  
+
  private String attemptsPostTableInit() {
    return "var asInitVals = new Array();\n" +
           "$('tfoot input').keyup( function () \n{"+
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTasksBlock.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTasksBlock.java
@ -140,6 +140,7 @@ public class HsTasksBlock extends HtmlBlock {
        attemptFinishTime = ta.getFinishTime();
        attemptElapsed = ta.getElapsedTime();
      }
+
      tasksTableData.append("[\"")
      .append("<a href='" + url("task", tid)).append("'>")
      .append(tid).append("</a>\",\"")
@ -205,9 +206,4 @@ public class HsTasksBlock extends HtmlBlock {

    footRow._()._()._();
  }
-
-  private String formatTime(long elapsed) {
-    return elapsed < 0 ? "N/A" : StringUtils.formatTime(elapsed);
-  }
-
 }
--- a/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTasksPage.java
+++ b/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-hs/src/main/java/org/apache/hadoop/mapreduce/v2/hs/webapp/HsTasksPage.java
@ -67,33 +67,25 @@ private String tasksTableInit() {
      type = MRApps.taskType(symbol);
    }
    StringBuilder b = tableInit().
-    append(", 'aaData': tasksTableData");
-    b.append(", bDeferRender: true");
-    b.append(", bProcessing: true");
+    append(", 'aaData': tasksTableData")
+    .append(", bDeferRender: true")
+    .append(", bProcessing: true")

-    b.append("\n, aoColumnDefs: [\n");
-    b.append("{'sType':'numeric', 'aTargets': [ 0 ]");
-    b.append(", 'mRender': parseHadoopID }");
+    .append("\n, aoColumnDefs: [\n")
+    .append("{'sType':'numeric', 'aTargets': [ 0 ]")
+    .append(", 'mRender': parseHadoopID }")

-    b.append(", {'sType':'numeric', 'aTargets': [ 4");
-    if(type == TaskType.REDUCE) {
-      b.append(", 9, 10, 11, 12");
-    } else { //MAP
-      b.append(", 7");
-    }
-    b.append(" ], 'mRender': renderHadoopElapsedTime }");
+    .append(", {'sType':'numeric', 'aTargets': [ 4")
+    .append(type == TaskType.REDUCE ? ", 9, 10, 11, 12" : ", 7")
+    .append(" ], 'mRender': renderHadoopElapsedTime }")

-    b.append("\n, {'sType':'numeric', 'aTargets': [ 2, 3, 5");
-    if(type == TaskType.REDUCE) {
-      b.append(", 6, 7, 8");
-    } else { //MAP
-      b.append(", 6");
-    }
-    b.append(" ], 'mRender': renderHadoopDate }]");
+    .append("\n, {'sType':'numeric', 'aTargets': [ 2, 3, 5")
+    .append(type == TaskType.REDUCE ? ", 6, 7, 8" : ", 6")
+    .append(" ], 'mRender': renderHadoopDate }]")

    // Sort by id upon page load
-    b.append("\n, aaSorting: [[0, 'asc']]");
-    b.append("}");
+    .append("\n, aaSorting: [[0, 'asc']]")
+    .append("}");
    return b.toString();
  }