开发NiFi的Hive到Teradata导出工具时，TDCH CLI测试报错求助

Hey there! Let's walk through troubleshooting your TDCH export issue— I’ve tackled plenty of Hive-to-Teradata migration kinks with TDCH before, so here’s a structured approach to dig into the problem:

First, let’s iron out the obvious gaps in your command:

• Complete the sourcetable parameter: Your command cuts off after -sourcetable— you need to specify the actual Hive table name here (e.g., -sourcetable customer_data). Missing this will throw a parameter parsing error immediately.
• Verify environment variables: Run echo $USERLIBTDCH and echo $LIBJARS to confirm these paths point to valid files/directories. Typos or missing paths here are a super common culprit.
• Check TDCH jar integrity: Use jar tf $USERLIBTDCH to list the contents of the TDCH jar. If you don’t see com.teradata.connector.common.tool.ConnectorExportTool in the output, your jar is either corrupted or the wrong TDCH package. Grab a fresh copy from Teradata’s official distribution.

Before blaming TDCH, rule out basic connection issues:

• Test JDBC access independently: Use a simple JDBC client (like sqlline or a quick Java snippet) with the same URL, username, and password to connect to Teradata. This will tell you if the problem is network-related (firewalls, DNS resolution), Teradata-side (user permissions, database availability), or missing JDBC drivers.
• Validate JDBC drivers in LIBJARS: Ensure $LIBJARS includes the Teradata JDBC driver jars (terajdbc4.jar and tdgssconfig.jar). These are mandatory for TDCH to talk to Teradata, and version mismatches between the driver and Teradata server can cause silent failures or connection errors.

Since you’re using -jobtype hcat, TDCH relies on HCatalog to interact with Hive:

• Test Hive table access: Run hive -e "SELECT COUNT(*) FROM testing.your_hive_table;" to confirm your user has read permissions on the source table, and Hive/HCatalog services are running properly. If this fails, TDCH will too.
• Check HCatalog dependencies: Make sure $LIBJARS includes HCatalog core jars (like hcatalog-core.jar, hive-metastore.jar) and their compatible Hadoop/Hive dependencies. Missing these will trigger ClassNotFoundException errors during job initialization.

TDCH’s default output is often too vague. Add the -verbose flag to your command to get full stack traces and debug logs:

hadoop jar $USERLIBTDCH com.teradata.connector.common.tool.ConnectorExportTool -libjars $LIBJARS -url jdbc:teradata://dns.com/DATABASE=dbName -username abc -password abc -jobtype hcat -sourcedatabase testing -sourcetable your_hive_table -verbose

The logs will point you directly to the root cause— whether it’s a permission denied error, missing dependency, or Teradata-side constraint.

TDCH is picky about version alignment with Hadoop, Hive, and Teradata:

• Ensure your TDCH version supports your Hadoop distribution (e.g., TDCH 2.0+ for Hadoop 3.x) and Teradata server version (e.g., Teradata 17.x requires TDCH 1.6+). Check the official TDCH documentation’s compatibility matrix to confirm all components play nice together.

内容的提问来源于stack exchange，提问作者knowone