[
https://issues.apache.org/jira/browse/MAPREDUCE-3261?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Riccomini resolved MAPREDUCE-3261.
----------------------------------------
Resolution: Fixed
Sigh. Checked debug and everything looks fine in yarn. Appears to be a bug in
my app master's "ask for more/less" logic.
> AM unable to release containers
> -------------------------------
>
> Key: MAPREDUCE-3261
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3261
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: applicationmaster
> Environment: [criccomi@criccomi-ld hadoop-trunk]$ svn info
> Path: .
> URL: http://svn.apache.org/repos/asf/hadoop/common/trunk
> Repository Root: http://svn.apache.org/repos/asf
> Repository UUID: 13f79535-47bb-0310-9956-ffa450edef68
> Revision: 1185741
> Node Kind: directory
> Schedule: normal
> Last Changed Author: acmurthy
> Last Changed Rev: 1185488
> Last Changed Date: 2011-10-17 22:30:32 -0700 (Mon, 17 Oct 2011)
> Reporter: Chris Riccomini
> Fix For: 0.23.0, 0.24.0
>
>
> I'm probably doing something wrong here, but I can't figure it out.
> My ApplicationMaster is sending an AllocateRequest with ContainerIds to
> release. My ResourceManager logs say:
> 2011-10-25 10:02:52,236 WARN resourcemanager.RMAuditLogger
> (RMAuditLogger.java:logFailure(207)) - USER=criccomi IP=127.0.0.1
> OPERATION=AM Released Container TARGET=FifoScheduler RESULT=FAILURE
> DESCRIPTION=Trying to release container not owned by app or with invalid id
> PERMISSIONS=Unauthorized access or invalid container
> APPID=application_1319485153554_0028
> CONTAINERID=container_1319485153554_0028_01_000003
> The container ID is valid, as is the app id:
> [criccomi@criccomi-ld logs]$ pwd
> /tmp/logs
> [criccomi@criccomi-ld logs]$ find .
> .
> ./application_1319485153554_0028
> ./application_1319485153554_0028/container_1319485153554_0028_01_000002
> ./application_1319485153554_0028/container_1319485153554_0028_01_000002/stderr
> ./application_1319485153554_0028/container_1319485153554_0028_01_000002/stdout
> ./application_1319485153554_0028/container_1319485153554_0028_01_000001
> ./application_1319485153554_0028/container_1319485153554_0028_01_000001/stderr
> ./application_1319485153554_0028/container_1319485153554_0028_01_000001/stdout
> ./application_1319485153554_0028/container_1319485153554_0028_01_000003
> ./application_1319485153554_0028/container_1319485153554_0028_01_000003/stderr
> ./application_1319485153554_0028/container_1319485153554_0028_01_000003/stdout
> ./application_1319485153554_0028/container_1319485153554_0028_01_000006
> ./application_1319485153554_0028/container_1319485153554_0028_01_000006/stderr
> ./application_1319485153554_0028/container_1319485153554_0028_01_000006/stdout
> The containers are still running.
> My code to start a container, and then to release it:
> {code}
> // ugi = UserGroupInformation.getCurrentUser
> // security is not enabled
> def startContainer(packagePath: Path, container: Container, ugi:
> UserGroupInformation, env: Map[String, String], cmds: String*) {
> info("%s starting container %s %s %s %s %s" format (appAttemptId,
> packagePath, container, ugi, env, cmds))
> // connect to container manager (based on similar code in the
> ContainerLauncher in Hadoop MapReduce)
> val contToken = container.getContainerToken
> val address = container.getNodeId.getHost + ":" +
> container.getNodeId.getPort
> var user = ugi
> if (UserGroupInformation.isSecurityEnabled) {
> debug("%s security is enabled" format (appAttemptId))
> val hadoopToken = new
> Token[ContainerTokenIdentifier](contToken.getIdentifier.array,
> contToken.getPassword.array, new Text(contToken.getKind), new
> Text(contToken.getService))
> user = UserGroupInformation.createRemoteUser(address)
> user.addToken(hadoopToken)
> info("%s changed user to %s" format (appAttemptId, user))
> }
> val containerManager = user.doAs(new PrivilegedAction[ContainerManager] {
> def run(): ContainerManager = {
> return YarnRPC.create(conf).getProxy(classOf[ContainerManager],
> NetUtils.createSocketAddr(address), conf).asInstanceOf[ContainerManager]
> }
> })
> // set the local package so that the containers and app master are
> provisioned with it
> val packageResource = Records.newRecord(classOf[LocalResource])
> val packageUrl = ConverterUtils.getYarnUrlFromPath(packagePath)
> val fileStatus =
> packagePath.getFileSystem(conf).getFileStatus(packagePath)
> packageResource.setResource(packageUrl)
> packageResource.setSize(fileStatus.getLen)
> packageResource.setTimestamp(fileStatus.getModificationTime)
> packageResource.setType(LocalResourceType.ARCHIVE)
> packageResource.setVisibility(LocalResourceVisibility.APPLICATION)
> // start the container
> val ctx = Records.newRecord(classOf[ContainerLaunchContext])
> ctx.setEnvironment(env)
> ctx.setContainerId(container.getId())
> ctx.setResource(container.getResource())
> ctx.setUser(user.getShortUserName())
> ctx.setCommands(cmds.toList)
> ctx.setLocalResources(Collections.singletonMap("package",
> packageResource))
> debug("%s setting package to %s" format (appAttemptId, packageResource))
> debug("%s setting context to %s" format (appAttemptId, ctx))
> val startContainerRequest =
> Records.newRecord(classOf[StartContainerRequest])
> startContainerRequest.setContainerLaunchContext(ctx)
> containerManager.startContainer(startContainerRequest)
> }
> {code}
> -----
> {code}
> def sendResourceRequest(requests: List[ResourceRequest], release:
> List[ContainerId]): AMResponse = {
> info("%s sending resource request %s %s" format (appAttemptId, requests,
> release))
> val req = Records.newRecord(classOf[AllocateRequest])
> req.setResponseId(requestId)
> req.setApplicationAttemptId(appAttemptId)
> req.addAllAsks(requests)
> req.addAllReleases(release)
> requestId += 1
> debug("%s RM resource request %s" format (appAttemptId, req))
> resourceManager.allocate(req).getAMResponse
> }
> {code}
> I have double checked that my ContainerIds are accurate, and they are.
> Any idea what I'm doing wrong here?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
