Searchable: External table with name <...> does not exist in connection <...>

[ash-peng]

Hi,

We are encountering some errors and looking for help on solving them. The tool is Searchable deploy. It is giving out errors including "Tables do not exist" and "external with name... does not exist in connection".

The tables that do not exist are:

TS_GROUP_MEMBERSHIP
TS_ORG
VW_TS_BI_SERVER_ADVANCED_AUTHOR
TS_METADATA_OBJECT
TS_DEPENDENT_OBJECT
TS_CLUSTER
TS_DATA_SOURCE
TS_USER
TS_BI_SERVER
TS_GROUP
TS_METADATA_COLUMN

---

The external tables that do not exist are:

..ts_bi_server
..ts_cluster
..ts_column_synonym
..ts_data_source
..ts_dependent_object
..ts_group
..ts_xref_principal
..ts_group_privilege
..ts_metadata_column
..ts_metadata_object
..ts_org
..ts_xref_org
..ts_sharing_access
..ts_tag
..ts_tagged_object
..ts_user

---

When I log into ThoughtSpot in the browser with an admin account, all of the "TS: ..." titled items we have are worksheets and none of them are tables. An example would be "TS: Query Stats Worksheet".

Looking forward to your response, thanks!

[boonhapus]

Hey @ash-peng , have you run the commands searchable metadata and searchable bi-server to load the data to your database yet? These commands will also create the tables in the database as well.

searchable deploy can come last and only needs to be run once.

I would recommend using the latest version of CS Tools (v1.5.9) if you're on any dialect other than Snowflake.

[ash-peng]

@boonhapus We are using Microsoft Azure Synapse for our main database. We have a databricks connection, but it's not actively used. I checked the Syncer page and didn't find Azure Synapse... is it not workable with CS Tools?

[boonhapus]

@ash-peng yeah that's correct. I'll check tomorrow to see if we have a test bench for Synapse at ThoughtSpot, to see if I can write a Syncer that supports MSSQL, but my guess is it would be easier for you to export using CSV, and use your own data loading process into Synapse.

If you run the searchable metadata or searchable bi-server commands with --syncer mock:// , you should be able print the DDL for each table that would be loaded to the database.

[ash-peng]

@boonhapus Hi, thank you for the explanation of the workaround!

I chatted with my team lead today about all we've discussed, and their current preferred approach is to use Databricks - though our current data is in Synapse (we have Databricks connections that are unused). Could you verify (or not verify) that it is okay to have the actual ThoughtSpot data in one database and the metadata in another, because even though we are slowly moving ThoughtSpot data from Synapse to Databricks it will still take some time. Even if the data will be in Databricks it may end up in another schema/catalog. Thanks a bunch!

[boonhapus]

@ash-peng that all makes sense, good to hear about the move to dbx! It's a fantastic database.

CS Tools fetches information about ThoughtSpot metadata from the ThoughtSpot APIs. The data supporting those APIs actually resides in a ts-internal proprietary database that customers are not able to interact with directly.

There is no harm in storing the data that CS Tools fetches, in Databricks, and re-exposing it via its own Connection. This is actually the preferred method - creating a separate connection entirely.

[boonhapus]

If you would like, you can email me directly and include your lead(s) if it makes things easier for you to make the architecture decisions.

[email protected]

[ash-peng]

@boonhapus Thank you so much!!!