Error while creating table with Glue catalog

[RC]

Hi, 

I have Databricks cluster earlier connected to hive metastore and we have started migrating to Glue catalog.

I'm facing an issue while creating table,

Path must be absolute: <table-name>-__PLACEHOLDER__

We have provided full access to glue and s3 in the respective role.

Below is the configuration which I'm using

spark.databricks.hive.metastore.glueCatalog.enabled true

I can create a hive table without any issue

Is there anything I'm missing here ?

[Kaniz]

 Hi @RC, The error message you're seeing suggests that the table path is not absolute. This could be due to how you create the table in the Glue Catalog. As per the given sources, when using AWS Glue Data Catalog as the metastore, it's recommended to specify the LOCATION clause to an absolute S3 bucket location while creating the table.

[RC]

hi @Kaniz ,

I have added "location" clause as well while defining the table.

But still facing issue, this happens only for delta table.

Attaching the screenshot

[Kaniz]

Hi @RC, it's important to remember that the LOCATION clause must specify a bucket location, such as s3://mybucket, when you call CREATE TABLE.

Here are the possible steps to troubleshoot the issue.

1. Check if the LOCATION specified in the CREATE TABLE statement is an S3 location.
2. Make sure the same mount point name is being used across all Databricks workspaces that share the Glue Catalog.
3. Ensure that the role associated with your Databricks cluster has the necessary permissions to access the S3 bucket specified in the LOCATION.

Sources:
- [Docs: aws-glue-metastore](https://docs.databricks.com/archive/external-metastores/aws-glue-metastore.html)
- [Docs: hive-metastore](https://docs.databricks.com/data-governance/unity-catalog/hive-metastore.html)
- [Docs: index](https://docs.databricks.com/archive/external-metastores/index.html)