Data Warehouse (ADS) Issues

This article discusses common Data Warehouse (ADS) issues and their solutions.

Connection Issues

When troubleshooting connection issues, always do the following as the first step: 

  • Ensure that you use the latest version of the Data Warehouse JDBC driver. You can download it from the Downloads page at https://secure.gooddata.com/downloads.html.

  • Review the error messages. The messages related to an unsuccessful connection to a Data Warehouse instance are usually propagated to the SQL client.   

Incorrect URL or hostname

Verify that your connection string is correct. To create a correct connection string, see Prepare the JDBC Connection String.

Authentication failed

Make sure that your username and password are correct. To validate it, try to log in to the GoodData platform (https://secure.gooddata.com or your white-labeled domain).

Data Warehouse instance not found or deleted

Verify that the instance ID is correct. You can check available Data Warehouse instances via the gray pages: https://secure.gooddata.com/gdc/datawarehouse/instances. If your domain is white-labeled, use your domain URL.

Error 403, or Authentication successful but cannot connect to Data Warehouse instance (access denied by server)

Your user is not assigned to the Data Warehouse instance that you are trying to connect to. Ask the owner or an admin of the instance to add your user to the Data Warehouse instance. For more information, see Adding a User in Data Warehouse.

Query Issues

Unable to run query

  • Make sure that you are successfully connected to the Data Warehouse instance.
  • Review the error message that you received. Usually, the reason is an incorrect query (for example, selecting a non-existing table), or wrong SQL syntax.
  • Verify whether you may have hit Resource Limitations.

Query did run but failed afterwards

Check whether your query may have hit one of Data Warehouse Resource Limitations.

  • Memory limit. Memory is shared among all your instances created by a single token. Check whether you are running any other memory intensive queries. If it is only one query that hit this limit, optimize the query.
  • Time limit. If a query runs for longer than 2 hours to execute, it gets terminated. To avoid this, optimize the query.

Use the EXPLAIN Keyword to analyze a query. Based on the retrieved query plan, make appropriate changes to your query.

When reviewing the results of running the EXPLAIN command, look for the following keywords:

  • RESEGMENT, BROADCAST, GLOBAL RESEGMENT GROUPS: Optimize segmentation of projections. For more information, see Segmentation.
  • HASH JOIN, ANALYTICAL -> Group Sort: Optimize HASH JOIN to MERGE JOIN in case of joining two big tables. Optimize the related projection ORDER BY clause to be compatible with the PARTITION BY clause, when using analytical functions. For more information, see Columns Sort Order.

Also, we recommend that you create an optimal projection for your tables. For more information about projections, see Physical Schema Design - Projections.

Query seems stuck

You can run up to four queries in parallel (see Resource Limitations). Additional queries are queued. You can check what is queued in the Data Warehouse monitoring tables.

The monitoring tables are part of the underlying Vertica database. Use the monitoring tables to troubleshoot query issues and getting the current and past information about your queries and the overall Data Warehouse status.

  • query_requests: Information about running and finished queries with details
  • resource_acquisitions: Information about resources (for example, memory consumed by a query)
  • resource_pool_status: Information about resource pools
  • resource_rejection_details: Information about rejected queries
  • resource_queues: Information about queued queries
Schedule Issues