启动Hive CLI（基于Tez）因资源本地化失败求助

问题描述 #

启动Hive CLI时出现Tez资源本地化失败，报错日志如下：

2023-07-08 10:00:46,822 INFO  [main] tez.DagUtils (DagUtils.java:localizeResource(1159)) - Localizing resource because it does not exist: file:/home/hadoop/tez/tez.0.10 to dest: hdfs://HostName:9000/tmp/hive/hadoop/_tez_session_dir/9cfed3fe-ac8a-411f-a54e-900ede1b7304-resources/tez.0.10 
2023-07-08 10:00:49,788 INFO  [main] tez.DagUtils (DagUtils.java:localizeResource(1182)) - Looks like another thread or process is writing the same file 
2023-07-08 10:00:49,792 INFO  [main] tez.DagUtils (DagUtils.java:checkOrWaitForTheFile(1211)) - Waiting for the file hdfs://Hostname:9000/tmp/hive/hadoop/_tez_session_dir/9cfed3fe-ac8a-411f-a54e-900ede1b7304-resources/tez.0.10 (5 attempts, with 5000ms interval) 
2023-07-08 10:01:14,829 ERROR [main] tez.DagUtils (DagUtils.java:localizeResource(1190)) - Could not find the jar that was being uploaded 
Exception in thread "main" java.lang.RuntimeException: java.io.IOException: Previous writer likely failed to write hdfs://Hostname:9000/tmp/hive/hadoop/_tez_session_dir/9cfed3fe-ac8a-411f-a54e-900ede1b7304-resources/tez.0.10. Failing because I am unlikely to write too. 
	at org.apache.hadoop.hive.ql.session.SessionState.start(SessionState.java:683) 
	at org.apache.hadoop.hive.ql.session.SessionState.beginStart(SessionState.java:591) 
	at org.apache.hadoop.hive.cli.CliDriver.run(CliDriver.java:747) 
	at org.apache.hadoop.hive.cli.CliDriver.main(CliDriver.java:683) 
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) 
	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) 
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) 
	at java.lang.reflect.Method.invoke(Method.java:498) 
	at org.apache.hadoop.util.RunJar.run(RunJar.java:318) 
	at org.apache.hadoop.util.RunJar.main(RunJar.java:232) 
Caused by: java.io.IOException: Previous writer likely failed to write hdfs://Hostname:9000/tmp/hive/hadoop/_tez_session_dir/9cfed3fe-ac8a-411f-a54e-900ede1b7304-resources/tez.0.10. Failing because I am unlikely to write too. 
	at org.apache.hadoop.hive.ql.exec.tez.DagUtils.localizeResource(DagUtils.java:1191) 
	at org.apache.hadoop.hive.ql.exec.tez.DagUtils.addTempResources(DagUtils.java:1042) 
	at org.apache.hadoop.hive.ql.exec.tez.DagUtils.localizeTempFilesFromConf(DagUtils.java:931) 
	at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.ensureLocalResources(TezSessionState.java:610) 
	at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.openInternal(TezSessionState.java:287) 
	at org.apache.hadoop.hive.ql.exec.tez.TezSessionState.beginOpen(TezSessionState.java:256) 
	at org.apache.hadoop.hive.ql.session.SessionState.start(SessionState.java:680) 
	... 9 more 

已尝试将所有Tez JAR复制到/home/hadoop/tez/路径，但问题仍未解决，寻求技术帮助。

排查解决方案 #

1. 清理HDFS临时目录 #

手动删除残留的Tez会话临时文件，避免不完整文件或锁干扰：

# 删除指定会话的资源目录
hdfs dfs -rm -r /tmp/hive/hadoop/_tez_session_dir/9cfed3fe-ac8a-411f-a54e-900ede1b7304-resources/
# 或清理整个Tez会话目录
hdfs dfs -rm -r /tmp/hive/hadoop/_tez_session_dir/

2. 检查文件权限与所有权 #

确保本地Tez JAR和HDFS目标路径的权限正确，hadoop用户拥有读写权限：

# 检查本地文件权限
ls -l /home/hadoop/tez/tez.0.10
# 修正权限（若需要）
chown hadoop:hadoop /home/hadoop/tez/tez.0.10
chmod 644 /home/hadoop/tez/tez.0.10

# 检查HDFS路径权限
hdfs dfs -ls /tmp/hive/hadoop/
# 修正HDFS路径权限（若需要）
hdfs dfs -chown -R hadoop:hadoop /tmp/hive/hadoop/
hdfs dfs -chmod -R 755 /tmp/hive/hadoop/

3. 修正Tez配置指向 #

修改Hive配置，将tez.lib.uris指向HDFS上的Tez压缩包而非本地路径，避免重复本地化：
编辑hive-site.xml，确认配置：

<property>
  <name>tez.lib.uris</name>
  <value>hdfs://HostName:9000/apps/tez/tez.tar.gz</value>
</property>

提前将Tez压缩包上传至HDFS对应路径，确保所有节点可访问。

4. 验证Tez JAR完整性 #

检查本地Tez JAR是否损坏，对比MD5值：

# 生成本地文件MD5
md5sum /home/hadoop/tez/tez.0.10
# 若已上传HDFS，对比远程文件MD5
hdfs dfs -cat /tmp/hive/hadoop/_tez_session_dir/.../tez.0.10 | md5sum

若MD5不一致，重新上传完整的Tez JAR文件。

5. 关闭冲突会话 #

终止正在运行的Tez作业或Hive会话，避免线程竞争写文件：

# 查看当前YARN应用
yarn application -list
# 杀掉冲突的Tez应用
yarn application -kill <application-id>

内容的提问来源于stack exchange，提问作者Kaher