Failover testing

This page describes how to run a failover test and what to check afterward. It applies to Automated disaster recovery on the DCT appliance. Use the same steps for a planned failover test and for an actual disaster.

How to test a failover

To test failover, promote the secondary to primary and confirm it operates correctly, then return it to a secondary role:

  1. Confirm replication is healthy first — on the secondary's Disaster Recovery tile, the replication state is Streaming with low lag.

  2. Promote the secondary to primary (see Performing a failover). The promoted appliance now operates as a primary.

  3. Run through the post-failover verification steps below to confirm each service you use is working.

  4. When you are done testing, return the appliance to a secondary role by configuring it to sync from the original primary again (see Configure a secondary). Its contents are replaced with the primary's, and replication resumes.

After you promote an appliance, increase its CPU and memory to match the workload it will now carry. While an appliance is a secondary it runs with minimum resources, so a freshly promoted appliance may need more CPU and memory before it can take full production load.

Pre-failover readiness checklist

Before a planned failover test:

  • Confirm whether you use the AI service — if so, be ready to re-upload its models after promotion.

  • Confirm whether you use LDAP over TLS and SAML/SSO — both need a step after promotion.

  • Make sure you have a local username and password or an API key that can sign in to the promoted appliance, in case SSO cannot reach it immediately after failover.

  • Confirm replication is healthy — replication state is Streaming with low lag — before starting.

Post-failover verification checklist

After promoting a secondary:

  • Core DCT and virtualization — confirm the promoted appliance is reachable and core operations (sign-in, listing engines and datasets) work. These are expected to work immediately.

  • Resources — increase CPU and memory to match the production workload the appliance will now carry.

  • AI service — confirm the AI models are present; re-upload any that are missing before relying on AI features.

  • LDAP over TLS — reconfigure the certificate material used to validate the LDAP server.

  • Network Security trust store — re-add any trusted certificates.

  • SAML / SSO — sign in with a local account or API key first, then update the identity provider to target the promoted appliance.

Related pages