[KYLIN-4190] hiveproducer write() function throw exception because hive mertics table location path prefix is different with defaut fs when hdfs uses router-based federation - ASF JIRA

Voters

Watch issue

Watchers

Link

Clone

Update Comment Author

Replace String in Comment

Update Comment Visibility

Delete Comments

XML

Word

Printable

JSON

Details

Type: Bug
Status: Closed
Priority: Critical
Resolution: Fixed
Affects Version/s: v3.0.0-alpha
Fix Version/s: v3.0.0
Component/s: Metrics
Labels:
None

Description

Our hdfs cluster uses router-based federation（https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs-rbf/HDFSRouterFederation.html).

Opening the system cube configuration, hiveproducer write() function throw exception :

//代码占位符
ERROR [metrics-blocking-reservoir-scheduler-0] hive.HiveReservoirReporter:119 : Wrong FS: hdfs://DClusterNmg4/user/kylin/hive/hive_metrics_job_exception_qa/kday_date=2019-09-04, expected: hdfs://difed
java.lang.IllegalArgumentException: Wrong FS: hdfs://DClusterNmg4/user/kylin/hive/hive_metrics_job_exception_qa/kday_date=2019-09-04, expected: hdfs://difed
        at org.apache.hadoop.fs.FileSystem.checkPath(FileSystem.java:717)
        at org.apache.hadoop.hdfs.DistributedFileSystem.getPathName(DistributedFileSystem.java:197)
        at org.apache.hadoop.hdfs.DistributedFileSystem.access$000(DistributedFileSystem.java:109)
        at org.apache.hadoop.hdfs.DistributedFileSystem$23.doCall(DistributedFileSystem.java:1390)
        at org.apache.hadoop.hdfs.DistributedFileSystem$23.doCall(DistributedFileSystem.java:1386)
        at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81)
        at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:1402)
        at org.apache.hadoop.fs.FileSystem.exists(FileSystem.java:1494)
        at org.apache.kylin.metrics.lib.impl.hive.HiveProducer.write(HiveProducer.java:137)
        at org.apache.kylin.metrics.lib.impl.hive.HiveProducer.send(HiveProducer.java:122)
        at org.apache.kylin.metrics.lib.impl.hive.HiveReservoirReporter$HiveReservoirListener.onRecordUpdate(HiveReservoirReporter.java:117)
        at org.apache.kylin.metrics.lib.impl.BlockingReservoir.notifyListenerOfUpdatedRecord(BlockingReservoir.java:105)
        at org.apache.kylin.metrics.lib.impl.BlockingReservoir.onRecordUpdate(BlockingReservoir.java:93)
        at org.apache.kylin.metrics.lib.impl.BlockingReservoir.access$300(BlockingReservoir.java:33)
        at org.apache.kylin.metrics.lib.impl.BlockingReservoir$ReporterRunnable.run(BlockingReservoir.java:152)
        at java.lang.Thread.run(Thread.java:745)

This is because the default router namespace is hdfs://difed, and the actual federation namespaces are the hdfs://DClusterNmg4, the hdfs://DClusterNmg1, and the hdfs://DClusterNmg2...

So fs.defaultFS in core-sie.xml is hdfs ://difed, But this hive table location path is hdfs://DClusterNmg4/user/... . Then defaultFs.exists(hiveLocationPath) throw exception.

So we need to check if the prefix is same. If defaut fs is not a prefix of hive table location path, use location path get a new filesystem