When I tried to write output to hive using DB Loader node, sometimes it gives an error that the output table doesn’t exist in hive workspace, where that table actually exists. If I reset the connect and rerun the node, sometimes it fixes the issue, but sometimes it still has the same error. Does anyone know what might be the cause of this issue and how can I fix it?
Hi Karen,
this is related to a timeout when reading the metadata during node configuration. This only happens if Hive is not responding within the set timeout. To disable the timeout you can deselect the “Retrieve in configure” option of the Metadata section of the Advanced settings tab. We have a ticket for making this clearer in the node as well.
Bye
Tobias
@karen2180 could you give us more context about your settings? Which version of Hive are you using, how do you handle table statistics (if at all). Do you also use Impala? Is the table newly being created in the KNIME workflow or is it there all along?
How does your DB loader construct look? With Hive if you want to upload a file you would have to first create it and then upload the data to an existing file. Maybe you could provide us with a screenshot, and maybe also the Log files.
The table has been created for a while. The db loader node works most of the time if I run the workflow locally, but only works occasionally if I run it on the server.
For starters I would recommend to load the data into a temporary table on the cluster and then insert it into the final table. And do you have some sort of logfile or error message