无法在 Hadoop 集群上启动 H2O - ClassNotFound 异常

Question

我正在尝试在 Hadoop 集群上启动 H2O。遗憾的是它不起作用，并给我一个错误，即找不到类 water.hadoop.h2omapper。

Hadoop环境是2.6版本的HDP，包括5个节点，其中1个运行YARN资源管理器，3个节点是YARN客户端的数据节点。每个数据节点都有 32GB RAM 和 4 个 CPU 内核的资源。没有其他应用程序在其上运行。我在 Ambari 的每个节点上为每个 YARN 应用程序配置了最多 16GB 和 3 个内核。

我从终端启动 H2O 集群（在所有节点上都尝试过，到处都是相同的错误），输出如下：

[root@host3 h2o-3.14.0.6-hdp2.6]# sudo -u hdfs hadoop jar h2odriver.jar -nodes 3 -mapperXmx 6g -output h2o-test
Determining driver host interface for mapper->driver callback...
[Possible callback IP address: 192.168.20.35]
[Possible callback IP address: 127.0.0.1]
Using mapper->driver callback IP address and port: 192.168.20.35:46619
(You can override these with -driverif and -driverport/-driverportrange.)
Memory Settings:
mapreduce.map.java.opts:     -Xms6g -Xmx6g -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -Dlog4j.defaultInitOverride=true
Extra memory percent:        10
mapreduce.map.memory.mb:     6758
17/10/13 07:49:14 INFO client.RMProxy: Connecting to ResourceManager at host2/192.168.20.34:8050
17/10/13 07:49:14 INFO client.AHSProxy: Connecting to Application History server at host2/192.168.20.34:10200
17/10/13 07:49:15 WARN mapreduce.JobResourceUploader: No job jar file set.  User classes may not be found. See Job or Job#setJar(String).

17/10/13 07:49:15 INFO mapreduce.JobSubmitter: number of splits:3
17/10/13 07:49:15 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1507793796947_0002
17/10/13 07:49:15 INFO mapred.YARNRunner: Job jar is not present. Not adding any jar to the list of resources.
17/10/13 07:49:15 INFO impl.YarnClientImpl: Submitted application application_1507793796947_0002
17/10/13 07:49:15 INFO mapreduce.Job: The url to track the job: http://host2:8088/proxy/application_1507793796947_0002/
Job name 'H2O_86929' submitted
JobTracker job ID is 'job_1507793796947_0002'
For YARN users, logs command is 'yarn logs -applicationId application_1507793796947_0002'
Waiting for H2O cluster to come up...
17/10/13 07:49:29 INFO client.RMProxy: Connecting to ResourceManager at host2/192.168.20.34:8050
17/10/13 07:49:29 INFO client.AHSProxy: Connecting to Application History server at host2/192.168.20.34:10200

----- YARN cluster metrics -----
Number of YARN worker nodes: 3

----- Nodes -----
Node: http://host5:8042 Rack: /default-rack, RUNNING, 1 containers used, 4,0 / 16,0 GB used, 1 / 3 vcores used
Node: http://host4:8042 Rack: /default-rack, RUNNING, 0 containers used, 0,0 / 16,0 GB used, 0 / 3 vcores used
Node: http://host3:8042 Rack: /default-rack, RUNNING, 0 containers used, 0,0 / 16,0 GB used, 0 / 3 vcores used

----- Queues -----
Queue name:            default
Queue state:       RUNNING
Current capacity:  0,11
Capacity:          1,00
Maximum capacity:  1,00
Application count: 1
----- Applications in this queue -----
Application ID:                  application_1507793796947_0002 (H2O_86929)
    Started:                     hdfs (Fri Oct 13 07:49:15 CEST 2017)
    Application state:           FINISHED
    Tracking URL:                http://host2:8088/proxy/application_1507793796947_0002/
    Queue name:                  default
    Used/Reserved containers:    1 / 0
    Needed/Used/Reserved memory: 4,0 GB / 4,0 GB / 0,0 GB
    Needed/Used/Reserved vcores: 1 / 1 / 0

Queue 'default' approximate utilization: 4,0 / 48,0 GB used, 1 / 9 vcores used

----------------------------------------------------------------------

ERROR: Unable to start any H2O nodes; please contact your YARN administrator.

   A common cause for this is the requested container size (6,6 GB)
   exceeds the following YARN settings:

       yarn.nodemanager.resource.memory-mb
       yarn.scheduler.maximum-allocation-mb

---

Yarn 应用系统日志中对应的错误条目：

2017-10-13 07:49:24,505 FATAL [IPC Server handler 1 on 40503] org.apache.hadoop.mapred.TaskAttemptListenerImpl: Task: attempt_1507793796947_0002_m_000002_0 - exited : java.lang.RuntimeException: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2241)
at org.apache.hadoop.mapreduce.task.JobContextImpl.getMapperClass(JobContextImpl.java:186)
at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:745)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:341)
at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:170)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:422)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1866)
at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:164)
Caused by: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClassByName(Configuration.java:2147)
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2239)
... 8 more

2017-10-13 07:49:24,506 INFO [IPC Server handler 1 on 40503] org.apache.hadoop.mapred.TaskAttemptListenerImpl: Diagnostics report from attempt_1507793796947_0002_m_000002_0: Error: java.lang.RuntimeException: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2241)
at org.apache.hadoop.mapreduce.task.JobContextImpl.getMapperClass(JobContextImpl.java:186)
at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:745)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:341)
at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:170)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:422)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1866)
at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:164)
Caused by: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClassByName(Configuration.java:2147)
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2239)
... 8 more

2017-10-13 07:49:24,507 INFO [AsyncDispatcher event handler] org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl: Diagnostics report from attempt_1507793796947_0002_m_000002_0: Error: java.lang.RuntimeException: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2241)
at org.apache.hadoop.mapreduce.task.JobContextImpl.getMapperClass(JobContextImpl.java:186)
at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:745)
at org.apache.hadoop.mapred.MapTask.run(MapTask.java:341)
at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:170)
at java.security.AccessController.doPrivileged(Native Method)
at javax.security.auth.Subject.doAs(Subject.java:422)
at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1866)
at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:164)
Caused by: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found
at org.apache.hadoop.conf.Configuration.getClassByName(Configuration.java:2147)
at org.apache.hadoop.conf.Configuration.getClass(Configuration.java:2239)
... 8 more

完整日志可在here获取。

任何帮助将不胜感激。

最好的问候， 马库斯

Answer 1

您是否能够运行其他 YARN 作业？ （如 pi 示例。）

应用程序 ID 显示您的 Hadoop 集群于 2017 年 10 月 12 日星期四 07:36:36 UTC（即昨天）启动，这是该集群尝试运行的第一个（也是第二个）作业。

另外，集群中节点的大小真的非常小。

在我看来，所有这一切都像是您正在尝试成为自己的 Hadoop 管理员，但还没有让它发挥作用。 :)

继续尝试，当您的集群配置正确时，H2O 将运行。

Answer 2

根据你提供的部分日志，下面这行有助于理解一点：

2017-10-12 07:45:02,172 FATAL [IPC Server handler 1 on 39365] 
 org.apache.hadoop.mapred.TaskAttemptListenerImpl: Task: 
  attempt_1507726330188_0001_m_000002_0 - exited : 
    java.lang.RuntimeException: java.lang.ClassNotFoundException: Class water.hadoop.h2omapper not found

很明显，任务 ID #2 实例 #0（或第一个实例）由于缺少类而失败。因此，如果任务 #2 显示错误，则意味着另一个任务 #1 已经在运行。这也意味着作业已经开始或处于运行状态。所以问题发生在作业执行阶段。这意味着执行此特定任务的 DataNode 无法找到 H2O 驱动程序库。因此，可能由于某种原因，节点或节点上的文件系统不可用。如果您研究您的详细日志，您将能够了解为什么会发生这种情况。

[其他详细信息]

由于权限或文件系统问题，任何映射器都无法访问 h2odriver.jar，这就是您的作业未启动的原因。您应该确保正确修复 Hadoop 权限和可访问性，以便您可以在没有 root 且需要超级用户别名的情况下启动任何“hadoop”命令。