Connect your Amazon Redshift data warehouse (a provisioned cluster or a Serverless workgroup) to Tekst so Tekst can read the ERP data you keep there. For example, Process Intelligence can use SAP's sales, delivery and billing tables (such as vbak, vbap, likp, lips, vbrk, vbrp and vbfa) to analyze your processes. Tekst only reads: it never writes to your warehouse.
Before you start
Check the following before you connect:
- Your data is in one schema. Tekst connects to a single schema, so keep the tables it needs together in the same one.
-
Tekst can reach the warehouse over the internet on the database port. Allow Tekst's outbound IP addresses listed in Troubleshoot Firewall. On Redshift, this means the cluster or workgroup is publicly accessible and its VPC security group allows those addresses on port
5439. - You have the Admin or Contributor role in Tekst.
Step 1: Create a read-only user
Tekst signs in with a database username and password. IAM authentication is not supported.
Run the following as a Redshift administrator. Replace the placeholders with your own values.
CREATE USER readonly_user PASSWORD '<a strong password>';
GRANT USAGE ON SCHEMA <schema> TO readonly_user;
GRANT SELECT ON ALL TABLES IN SCHEMA <schema> TO readonly_user;
Tekst only sees tables this user can read.
Step 2: Connect in Tekst
- Go to Integrations and click Connect Integration.
- Find Redshift (category Data warehouse) and click Connect.
- In the Connect Redshift dialog, fill in all six fields:
-
Host: your cluster or workgroup endpoint, for example
cluster.region.redshift.amazonaws.com. Host name only: nohttps://, and leave out the:5439/devthat the AWS console shows after it. -
Port:
5439is filled in. -
Database: the database that holds your tables, for example
dev. -
Schema: the schema name, for example
public. -
Username:
readonly_user(or the user you created). - Password: that user's password.
-
Host: your cluster or workgroup endpoint, for example
- Click Connect. The button stays disabled until every field is filled in.
Expected result: Tekst checks that it can log in and see at least one table in the schema. The dialog closes, and the integration's configuration panel opens showing Connected. No success message appears.
Verify the connection later
Open the integration, go to the Advanced tab and click Verify connection. Tekst doesn't recheck the connection by itself, so verify again after you change the password or the security group.
What happens next
Tekst support reviews with you which tables Tekst loads and runs the first sync. Syncs don't run on a schedule, so the data in Tekst is as current as the last sync. Contact Tekst support for a new sync.
Change the password or user
Open the integration, go to the Settings tab and click Reconnect, then fill in all six fields again. With the same host, port, database and schema, Tekst updates the existing integration. Tell Tekst support, because their table setup is reset. A different host, port, database or schema creates a second integration.
Troubleshooting
Connect errors appear in red at the top of the Connect Redshift dialog. Text after "Could not connect to the database:" comes from Redshift.
- "Could not connect to the database: password authentication failed for user ..." - wrong username or password.
- "Could not connect to the database: write
CONNECT_TIMEOUT..." (after about 30 seconds) - Tekst can't reach the warehouse. Check that the cluster or workgroup is publicly accessible and that its security group allows Tekst's outbound IPs listed in Troubleshoot Firewall. -
Schema "<schema>" not found, or has no tables visible to this user- the schema name is wrong, or the user lacksUSAGEorSELECT.
Verify connection shows its result as a pop-up, not in the dialog:
- "Integration is not healthy - please check your credentials" - the password changed, the user was removed, or the security group changed. Fix it, then Reconnect.
If you have an issue following this guide or have any other questions, please submit a request or send your question to support@tekst.com.
0 comments
Please sign in to leave a comment.