SPARK-1890 and SPARK-1891- add admin and modify acls

It was easier to combine these 2 jira since they touch many of the same places. This pr adds the following: - adds modify acls - adds admin acls (list of admins/users that get added to both view and modify acls) - modify Kill button on UI to take modify acls into account - changes config name of spark.ui.acls.enable to spark.acls.enable since I choose poorly in original name. We keep backwards compatibility so people can still use spark.ui.acls.enable. The acls should apply to any web ui as well as any CLI interfaces. - send view and modify acls information on to YARN so that YARN interfaces can use (yarn cli for killing applications for example). Author: Thomas Graves <[email protected]> Closes apache#1196 from tgravescs/SPARK-1890 and squashes the following commits: 8292eb1 [Thomas Graves] review comments b92ec89 [Thomas Graves] remove unneeded variable from applistener 4c765f4 [Thomas Graves] Add in admin acls 72eb0ac [Thomas Graves] Add modify acls
japp2002 · Aug 5, 2014 · 1c5555a · 1c5555a
1 parent 2c0f705
commit 1c5555a
Show file tree

Hide file tree

Showing 8 changed files with 206 additions and 37 deletions.
diff --git a/core/src/main/scala/org/apache/spark/SecurityManager.scala b/core/src/main/scala/org/apache/spark/SecurityManager.scala
@@ -41,10 +41,19 @@ import org.apache.spark.deploy.SparkHadoopUtil
  * secure the UI if it has data that other users should not be allowed to see. The javax
  * servlet filter specified by the user can authenticate the user and then once the user
  * is logged in, Spark can compare that user versus the view acls to make sure they are
- * authorized to view the UI. The configs 'spark.ui.acls.enable' and 'spark.ui.view.acls'
+ * authorized to view the UI. The configs 'spark.acls.enable' and 'spark.ui.view.acls'
  * control the behavior of the acls. Note that the person who started the application
  * always has view access to the UI.
  *
+ * Spark has a set of modify acls (`spark.modify.acls`) that controls which users have permission
+ * to  modify a single application. This would include things like killing the application. By
+ * default the person who started the application has modify access. For modify access through
+ * the UI, you must have a filter that does authentication in place for the modify acls to work
+ * properly.
+ *
+ * Spark also has a set of admin acls (`spark.admin.acls`) which is a set of users/administrators
+ * who always have permission to view or modify the Spark application.
+ *
  * Spark does not currently support encryption after authentication.
  *
  * At this point spark has multiple communication protocols that need to be secured and
@@ -137,18 +146,32 @@ private[spark] class SecurityManager(sparkConf: SparkConf) extends Logging {
   private val sparkSecretLookupKey = "sparkCookie"
 
   private val authOn = sparkConf.getBoolean("spark.authenticate", false)
-  private var uiAclsOn = sparkConf.getBoolean("spark.ui.acls.enable", false)
+  // keep spark.ui.acls.enable for backwards compatibility with 1.0
+  private var aclsOn = sparkConf.getOption("spark.acls.enable").getOrElse(
+    sparkConf.get("spark.ui.acls.enable", "false")).toBoolean
+
+  // admin acls should be set before view or modify acls
+  private var adminAcls: Set[String] =
+    stringToSet(sparkConf.get("spark.admin.acls", ""))
 
   private var viewAcls: Set[String] = _
+
+  // list of users who have permission to modify the application. This should
+  // apply to both UI and CLI for things like killing the application.
+  private var modifyAcls: Set[String] = _
+
   // always add the current user and SPARK_USER to the viewAcls
-  private val defaultAclUsers = Seq[String](System.getProperty("user.name", ""),
+  private val defaultAclUsers = Set[String](System.getProperty("user.name", ""),
     Option(System.getenv("SPARK_USER")).getOrElse(""))
+
   setViewAcls(defaultAclUsers, sparkConf.get("spark.ui.view.acls", ""))
+  setModifyAcls(defaultAclUsers, sparkConf.get("spark.modify.acls", ""))
 
   private val secretKey = generateSecretKey()
   logInfo("SecurityManager: authentication " + (if (authOn) "enabled" else "disabled") +
-    "; ui acls " + (if (uiAclsOn) "enabled" else "disabled") +
-    "; users with view permissions: " + viewAcls.toString())
+    "; ui acls " + (if (aclsOn) "enabled" else "disabled") +
+    "; users with view permissions: " + viewAcls.toString() +
+    "; users with modify permissions: " + modifyAcls.toString())
 
   // Set our own authenticator to properly negotiate user/password for HTTP connections.
   // This is needed by the HTTP client fetching from the HttpServer. Put here so its
@@ -169,18 +192,51 @@ private[spark] class SecurityManager(sparkConf: SparkConf) extends Logging {
     )
   }
 
-  private[spark] def setViewAcls(defaultUsers: Seq[String], allowedUsers: String) {
-    viewAcls = (defaultUsers ++ allowedUsers.split(',')).map(_.trim()).filter(!_.isEmpty).toSet 
+  /**
+   * Split a comma separated String, filter out any empty items, and return a Set of strings
+   */
+  private def stringToSet(list: String): Set[String] = {
+    list.split(',').map(_.trim).filter(!_.isEmpty).toSet
+  }
+
+  /**
+   * Admin acls should be set before the view or modify acls.  If you modify the admin
+   * acls you should also set the view and modify acls again to pick up the changes.
+   */
+  def setViewAcls(defaultUsers: Set[String], allowedUsers: String) {
+    viewAcls = (adminAcls ++ defaultUsers ++ stringToSet(allowedUsers))
     logInfo("Changing view acls to: " + viewAcls.mkString(","))
   }
 
-  private[spark] def setViewAcls(defaultUser: String, allowedUsers: String) {
-    setViewAcls(Seq[String](defaultUser), allowedUsers)
+  def setViewAcls(defaultUser: String, allowedUsers: String) {
+    setViewAcls(Set[String](defaultUser), allowedUsers)
+  }
+
+  def getViewAcls: String = viewAcls.mkString(",")
+
+  /**
+   * Admin acls should be set before the view or modify acls.  If you modify the admin
+   * acls you should also set the view and modify acls again to pick up the changes.
+   */
+  def setModifyAcls(defaultUsers: Set[String], allowedUsers: String) {
+    modifyAcls = (adminAcls ++ defaultUsers ++ stringToSet(allowedUsers))
+    logInfo("Changing modify acls to: " + modifyAcls.mkString(","))
+  }
+
+  def getModifyAcls: String = modifyAcls.mkString(",")
+
+  /**
+   * Admin acls should be set before the view or modify acls.  If you modify the admin
+   * acls you should also set the view and modify acls again to pick up the changes.
+   */
+  def setAdminAcls(adminUsers: String) {
+    adminAcls = stringToSet(adminUsers)
+    logInfo("Changing admin acls to: " + adminAcls.mkString(","))
   }
 
-  private[spark] def setUIAcls(aclSetting: Boolean) { 
-    uiAclsOn = aclSetting 
-    logInfo("Changing acls enabled to: " + uiAclsOn)
+  def setAcls(aclSetting: Boolean) {
+    aclsOn = aclSetting
+    logInfo("Changing acls enabled to: " + aclsOn)
   }
 
   /**
@@ -224,22 +280,39 @@ private[spark] class SecurityManager(sparkConf: SparkConf) extends Logging {
    * Check to see if Acls for the UI are enabled
    * @return true if UI authentication is enabled, otherwise false
    */
-  def uiAclsEnabled(): Boolean = uiAclsOn
+  def aclsEnabled(): Boolean = aclsOn
 
   /**
    * Checks the given user against the view acl list to see if they have
-   * authorization to view the UI. If the UI acls must are disabled
-   * via spark.ui.acls.enable, all users have view access.
+   * authorization to view the UI. If the UI acls are disabled
+   * via spark.acls.enable, all users have view access. If the user is null
+   * it is assumed authentication is off and all users have access.
    *
    * @param user to see if is authorized
    * @return true is the user has permission, otherwise false
    */
   def checkUIViewPermissions(user: String): Boolean = {
-    logDebug("user=" + user + " uiAclsEnabled=" + uiAclsEnabled() + " viewAcls=" + 
+    logDebug("user=" + user + " aclsEnabled=" + aclsEnabled() + " viewAcls=" +
       viewAcls.mkString(","))
-    if (uiAclsEnabled() && (user != null) && (!viewAcls.contains(user))) false else true
+    if (aclsEnabled() && (user != null) && (!viewAcls.contains(user))) false else true
   }
 
+  /**
+   * Checks the given user against the modify acl list to see if they have
+   * authorization to modify the application. If the UI acls are disabled
+   * via spark.acls.enable, all users have modify access. If the user is null
+   * it is assumed authentication isn't turned on and all users have access.
+   *
+   * @param user to see if is authorized
+   * @return true is the user has permission, otherwise false
+   */
+  def checkModifyPermissions(user: String): Boolean = {
+    logDebug("user=" + user + " aclsEnabled=" + aclsEnabled() + " modifyAcls=" +
+      modifyAcls.mkString(","))
+    if (aclsEnabled() && (user != null) && (!modifyAcls.contains(user))) false else true
+  }
+
+
   /**
    * Check to see if authentication for the Spark communication protocols is enabled
    * @return true if authentication is enabled, otherwise false

diff --git a/core/src/main/scala/org/apache/spark/deploy/history/FsHistoryProvider.scala b/core/src/main/scala/org/apache/spark/deploy/history/FsHistoryProvider.scala
@@ -189,7 +189,9 @@ private[history] class FsHistoryProvider(conf: SparkConf) extends ApplicationHis
 
     if (ui != null) {
       val uiAclsEnabled = conf.getBoolean("spark.history.ui.acls.enable", false)
-      ui.getSecurityManager.setUIAcls(uiAclsEnabled)
+      ui.getSecurityManager.setAcls(uiAclsEnabled)
+      // make sure to set admin acls before view acls so properly picked up
+      ui.getSecurityManager.setAdminAcls(appListener.adminAcls)
       ui.getSecurityManager.setViewAcls(appListener.sparkUser, appListener.viewAcls)
     }
     (appInfo, ui)

diff --git a/core/src/main/scala/org/apache/spark/scheduler/ApplicationEventListener.scala b/core/src/main/scala/org/apache/spark/scheduler/ApplicationEventListener.scala
@@ -29,7 +29,7 @@ private[spark] class ApplicationEventListener extends SparkListener {
   var startTime = -1L
   var endTime = -1L
   var viewAcls = ""
-  var enableViewAcls = false
+  var adminAcls = ""
 
   def applicationStarted = startTime != -1
 
@@ -55,7 +55,7 @@ private[spark] class ApplicationEventListener extends SparkListener {
       val environmentDetails = environmentUpdate.environmentDetails
       val allProperties = environmentDetails("Spark Properties").toMap
       viewAcls = allProperties.getOrElse("spark.ui.view.acls", "")
-      enableViewAcls = allProperties.getOrElse("spark.ui.acls.enable", "false").toBoolean
+      adminAcls = allProperties.getOrElse("spark.admin.acls", "")
     }
   }
 }
diff --git a/core/src/main/scala/org/apache/spark/ui/jobs/JobProgressTab.scala b/core/src/main/scala/org/apache/spark/ui/jobs/JobProgressTab.scala
@@ -41,7 +41,7 @@ private[ui] class JobProgressTab(parent: SparkUI) extends WebUITab(parent, "stag
   def isFairScheduler = listener.schedulingMode.exists(_ == SchedulingMode.FAIR)
 
   def handleKillRequest(request: HttpServletRequest) =  {
-    if (killEnabled) {
+    if ((killEnabled) && (parent.securityManager.checkModifyPermissions(request.getRemoteUser))) {
       val killFlag = Option(request.getParameter("terminate")).getOrElse("false").toBoolean
       val stageId = Option(request.getParameter("id")).getOrElse("-1").toInt
       if (stageId >= 0 && killFlag && listener.activeStages.contains(stageId)) {

diff --git a/core/src/test/scala/org/apache/spark/SecurityManagerSuite.scala b/core/src/test/scala/org/apache/spark/SecurityManagerSuite.scala
@@ -31,7 +31,7 @@ class SecurityManagerSuite extends FunSuite {
     conf.set("spark.ui.view.acls", "user1,user2")
     val securityManager = new SecurityManager(conf);
     assert(securityManager.isAuthenticationEnabled() === true)
-    assert(securityManager.uiAclsEnabled() === true)
+    assert(securityManager.aclsEnabled() === true)
     assert(securityManager.checkUIViewPermissions("user1") === true)
     assert(securityManager.checkUIViewPermissions("user2") === true)
     assert(securityManager.checkUIViewPermissions("user3") === false)
@@ -41,23 +41,90 @@ class SecurityManagerSuite extends FunSuite {
     val conf = new SparkConf
     conf.set("spark.ui.view.acls", "user1,user2")
     val securityManager = new SecurityManager(conf);
-    securityManager.setUIAcls(true)
-    assert(securityManager.uiAclsEnabled() === true)
-    securityManager.setUIAcls(false)
-    assert(securityManager.uiAclsEnabled() === false)
+    securityManager.setAcls(true)
+    assert(securityManager.aclsEnabled() === true)
+    securityManager.setAcls(false)
+    assert(securityManager.aclsEnabled() === false)
 
     // acls are off so doesn't matter what view acls set to
     assert(securityManager.checkUIViewPermissions("user4") === true)
 
-    securityManager.setUIAcls(true)
-    assert(securityManager.uiAclsEnabled() === true)
-    securityManager.setViewAcls(ArrayBuffer[String]("user5"), "user6,user7")
+    securityManager.setAcls(true)
+    assert(securityManager.aclsEnabled() === true)
+    securityManager.setViewAcls(Set[String]("user5"), "user6,user7")
     assert(securityManager.checkUIViewPermissions("user1") === false)
     assert(securityManager.checkUIViewPermissions("user5") === true)
     assert(securityManager.checkUIViewPermissions("user6") === true)
     assert(securityManager.checkUIViewPermissions("user7") === true)
     assert(securityManager.checkUIViewPermissions("user8") === false)
     assert(securityManager.checkUIViewPermissions(null) === true)
   }
+
+  test("set security modify acls") {
+    val conf = new SparkConf
+    conf.set("spark.modify.acls", "user1,user2")
+
+    val securityManager = new SecurityManager(conf);
+    securityManager.setAcls(true)
+    assert(securityManager.aclsEnabled() === true)
+    securityManager.setAcls(false)
+    assert(securityManager.aclsEnabled() === false)
+
+    // acls are off so doesn't matter what view acls set to
+    assert(securityManager.checkModifyPermissions("user4") === true)
+
+    securityManager.setAcls(true)
+    assert(securityManager.aclsEnabled() === true)
+    securityManager.setModifyAcls(Set("user5"), "user6,user7")
+    assert(securityManager.checkModifyPermissions("user1") === false)
+    assert(securityManager.checkModifyPermissions("user5") === true)
+    assert(securityManager.checkModifyPermissions("user6") === true)
+    assert(securityManager.checkModifyPermissions("user7") === true)
+    assert(securityManager.checkModifyPermissions("user8") === false)
+    assert(securityManager.checkModifyPermissions(null) === true)
+  }
+
+  test("set security admin acls") {
+    val conf = new SparkConf
+    conf.set("spark.admin.acls", "user1,user2")
+    conf.set("spark.ui.view.acls", "user3")
+    conf.set("spark.modify.acls", "user4")
+
+    val securityManager = new SecurityManager(conf);
+    securityManager.setAcls(true)
+    assert(securityManager.aclsEnabled() === true)
+
+    assert(securityManager.checkModifyPermissions("user1") === true)
+    assert(securityManager.checkModifyPermissions("user2") === true)
+    assert(securityManager.checkModifyPermissions("user4") === true)
+    assert(securityManager.checkModifyPermissions("user3") === false)
+    assert(securityManager.checkModifyPermissions("user5") === false)
+    assert(securityManager.checkModifyPermissions(null) === true)
+    assert(securityManager.checkUIViewPermissions("user1") === true)
+    assert(securityManager.checkUIViewPermissions("user2") === true)
+    assert(securityManager.checkUIViewPermissions("user3") === true)
+    assert(securityManager.checkUIViewPermissions("user4") === false)
+    assert(securityManager.checkUIViewPermissions("user5") === false)
+    assert(securityManager.checkUIViewPermissions(null) === true)
+
+    securityManager.setAdminAcls("user6")
+    securityManager.setViewAcls(Set[String]("user8"), "user9")
+    securityManager.setModifyAcls(Set("user11"), "user9")
+    assert(securityManager.checkModifyPermissions("user6") === true)
+    assert(securityManager.checkModifyPermissions("user11") === true)
+    assert(securityManager.checkModifyPermissions("user9") === true)
+    assert(securityManager.checkModifyPermissions("user1") === false)
+    assert(securityManager.checkModifyPermissions("user4") === false)
+    assert(securityManager.checkModifyPermissions(null) === true)
+    assert(securityManager.checkUIViewPermissions("user6") === true)
+    assert(securityManager.checkUIViewPermissions("user8") === true)
+    assert(securityManager.checkUIViewPermissions("user9") === true)
+    assert(securityManager.checkUIViewPermissions("user1") === false)
+    assert(securityManager.checkUIViewPermissions("user3") === false)
+    assert(securityManager.checkUIViewPermissions(null) === true)
+
+  }
+
+
 }
 
diff --git a/docs/configuration.md b/docs/configuration.md
@@ -815,13 +815,13 @@ Apart from these, the following properties are also available, and may be useful
   </td>
 </tr>
 <tr>
-  <td><code>spark.ui.acls.enable</code></td>
+  <td><code>spark.acls.enable</code></td>
   <td>false</td>
   <td>
-    Whether Spark web ui acls should are enabled. If enabled, this checks to see if the user has
-    access permissions to view the web ui. See <code>spark.ui.view.acls</code> for more details.
-    Also note this requires the user to be known, if the user comes across as null no checks
-    are done. Filters can be used to authenticate and set the user.
+    Whether Spark acls should are enabled. If enabled, this checks to see if the user has
+    access permissions to view or modify the job.  Note this requires the user to be known, 
+    so if the user comes across as null no checks are done. Filters can be used with the UI
+    to authenticate and set the user.
   </td>
 </tr>
 <tr>
@@ -832,6 +832,23 @@ Apart from these, the following properties are also available, and may be useful
     user that started the Spark job has view access.
   </td>
 </tr>
+<tr>
+  <td><code>spark.modify.acls</code></td>
+  <td>Empty</td>
+  <td>
+    Comma separated list of users that have modify access to the Spark job. By default only the
+    user that started the Spark job has access to modify it (kill it for example).
+  </td>
+</tr>
+<tr>
+  <td><code>spark.admin.acls</code></td>
+  <td>Empty</td>
+  <td>
+    Comma separated list of users/administrators that have view and modify access to all Spark jobs.
+    This can be used if you run on a shared cluster and have a set of administrators or devs who
+    help debug when things work.
+  </td>
+</tr>
 </table>
 
 #### Spark Streaming

diff --git a/docs/security.md b/docs/security.md
@@ -8,8 +8,11 @@ Spark currently supports authentication via a shared secret. Authentication can
 * For Spark on [YARN](running-on-yarn.html) deployments, configuring `spark.authenticate` to `true` will automatically handle generating and distributing the shared secret. Each application will use a unique shared secret. 
 * For other types of Spark deployments, the Spark parameter `spark.authenticate.secret` should be configured on each of the nodes. This secret will be used by all the Master/Workers and applications.
 
-The Spark UI can also be secured by using [javax servlet filters](http://docs.oracle.com/javaee/6/api/javax/servlet/Filter.html) via the `spark.ui.filters` setting. A user may want to secure the UI if it has data that other users should not be allowed to see. The javax servlet filter specified by the user can authenticate the user and then once the user is logged in, Spark can compare that user versus the view ACLs to make sure they are authorized to view the UI. The configs `spark.ui.acls.enable` and `spark.ui.view.acls` control the behavior of the ACLs. Note that the user who started the application always has view access to the UI.
-On YARN, the Spark UI uses the standard YARN web application proxy mechanism and will authenticate via any installed Hadoop filters.
+The Spark UI can also be secured by using [javax servlet filters](http://docs.oracle.com/javaee/6/api/javax/servlet/Filter.html) via the `spark.ui.filters` setting. A user may want to secure the UI if it has data that other users should not be allowed to see. The javax servlet filter specified by the user can authenticate the user and then once the user is logged in, Spark can compare that user versus the view ACLs to make sure they are authorized to view the UI. The configs `spark.acls.enable` and `spark.ui.view.acls` control the behavior of the ACLs. Note that the user who started the application always has view access to the UI.  On YARN, the Spark UI uses the standard YARN web application proxy mechanism and will authenticate via any installed Hadoop filters.
+
+Spark also supports modify ACLs to control who has access to modify a running Spark application.  This includes things like killing the application or a task. This is controlled by the configs `spark.acls.enable` and `spark.modify.acls`. Note that if you are authenticating the web UI, in order to use the kill button on the web UI it might be necessary to add the users in the modify acls to the view acls also. On YARN, the modify acls are passed in and control who has modify access via YARN interfaces.
+
+Spark allows for a set of administrators to be specified in the acls who always have view and modify permissions to all the applications. is controlled by the config `spark.admin.acls`. This is useful on a shared cluster where you might have administrators or support staff who help users debug applications.
 
 If your applications are using event logging, the directory where the event logs go (`spark.eventLog.dir`) should be manually created and have the proper permissions set on it. If you want those log files secured, the permissions should be set to `drwxrwxrwxt` for that directory. The owner of the directory should be the super user who is running the history server and the group permissions should be restricted to super user group. This will allow all users to write to the directory but will prevent unprivileged users from removing or renaming a file unless they own the file or directory. The event log files will be created by Spark with permissions such that only the user and group have read and write access.