svn merge -c 1602033 FIXES: HADOOP-10622. Shell.runCommand can deadlock. Contributed by Gera Shegalov

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1602034 13f79535-47bb-0310-9956-ffa450edef68

hadoop-common-project/hadoop-common/CHANGES.txt

@@ -195,6 +195,8 @@ Release 2.5.0 - UNRELEASED
     HADOOP-10656. The password keystore file is not picked by LDAP group mapping
     (brandonli)
 
+    HADOOP-10622. Shell.runCommand can deadlock (Gera Shegalov via jlowe)
+
   BREAKDOWN OF HADOOP-10514 SUBTASKS AND RELATED JIRAS
 
     HADOOP-10520. Extended attributes definition and FileSystem APIs for

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/util/Shell.java

@@ -526,12 +526,8 @@ abstract public class Shell {
       }
       // wait for the process to finish and check the exit code
       exitCode  = process.waitFor();
-      try {
-        // make sure that the error thread exits
-        errThread.join();
-      } catch (InterruptedException ie) {
-        LOG.warn("Interrupted while reading the error stream", ie);
-      }
+      // make sure that the error thread exits
+      joinThread(errThread);
       completed.set(true);
       //the timeout thread handling
       //taken care in finally block
@@ -560,13 +556,9 @@ abstract public class Shell {
       } catch (IOException ioe) {
         LOG.warn("Error while closing the input stream", ioe);
       }
-      try {
-        if (!completed.get()) {
-          errThread.interrupt();
-          errThread.join();
-        }
-      } catch (InterruptedException ie) {
-        LOG.warn("Interrupted while joining errThread");
+      if (!completed.get()) {
+        errThread.interrupt();
+        joinThread(errThread);
       }
       try {
         InputStream stderr = process.getErrorStream();
@@ -581,6 +573,19 @@ abstract public class Shell {
     }
   }
 
+  private static void joinThread(Thread t) {
+    while (t.isAlive()) {
+      try {
+        t.join();
+      } catch (InterruptedException ie) {
+        if (LOG.isWarnEnabled()) {
+          LOG.warn("Interrupted while joining on: " + t, ie);
+        }
+        t.interrupt(); // propagate interrupt
+      }
+    }
+  }
+
   /** return an array containing the command name & its parameters */ 
   protected abstract String[] getExecString();