Full Resynchronization of Codebeamer Databases for Disaster Recovery
The disaster recovery mechanism feature ensures data consistency between the Codebeamer database and the Indexing Service vector database when a desynchronization occurs.
When the indexing service and vector database are enabled, part of the data used for the Search Assistant is stored outside the Codebeamer core database. This separation introduces a risk that the two data sources can become inconsistent in rare failure scenarios, such as database rollback or data loss.
To resolve this issue, Codebeamer provides a disaster recovery mechanism. This feature involves manually initiating a full resynchronization between the Codebeamer database and Indexing Service vector database using the /api/cbai/v1/indexing/initiate-disaster-recovery REST API endpoint. For more information, see Disaster Recovery REST API Endpoint.
 |
The feature is available only to system administrators.
|
When to Use the Disaster Recovery Mechanism
Use the disaster recovery mechanism only in exceptional scenarios, such as:
• Data loss or rollback in the Codebeamer database
• Data loss or rollback in the vector database
• Any situation where the two databases are no longer synchronized
 |
This mechanism is not required during normal operations, such as when restarting Codebeamer or during network outages.
|
How Disaster Recovery works
When you initiate a full resynchronization:
• Codebeamer reindexes content based on the configured indexing settings and synchronizes it.
• The Indexing Service updates its internal state to match the current Codebeamer database.
• Missing or inconsistent data in the Indexing Service vector database is restored.
• Normal indexing operations continue during the process, and new content can still be added. However, the new content can be processed only after the disaster recovery is complete.
After the resynchronization completes, Codebeamer logs a message in the Codebeamer server log file indicating that disaster recovery has been completed and normal operations have resumed. Alternatively, check the Indexing Overview page under the AI System Admin tab. Disaster recovery is considered complete when no projects appear with status as Indexing. For more information, see Monitor AI Indexing Service Statistics.
Use Cases
The disaster recovery mechanism is typically used in the following scenarios:
• Recovery after database rollback — If either the Codebeamer database or the vector database is restored from a backup, newly created data may be missing in the other system. This can cause incomplete or incorrect search results. Use disaster recovery to resynchronize both systems and restore consistency.
• Restoring missing search results — When semantic search returns incomplete results due to missing indexed data in the vector database. Running disaster recovery restores the missing content in the vector database, ensuring correct search results.
• Initializing a clean environment — When starting with a clean Codebeamer database while the vector database still contains previous data, inconsistencies can occur. Triggering Disaster Recovery clears and rebuilds the indexed data to ensure alignment between systems.
Key Considerations and Limitations
• This mechanism is intended only for severe data inconsistency scenarios and should not be used during normal operations.
• Disaster Recovery must be initiated manually through the /api/cbai/v1/indexing/initiate-disaster-recovery REST API endpoint.
• The REST API endpoint requires a confirmation code to prevent accidental execution.
• The resynchronization process can take time depending on the size of indexed data.
• Re-triggering the resynchronization can restart the process from the beginning.
• The process includes delays for startup and completion checks.
• Existing vectors in the vector database are not re‑embedded, therefore no additional AI credits are consumed.
Impact on Search Assistant
• While disaster recovery is in progress, the search assistant may not return all relevant results. Wait until the resynchronization process completes before relying on search results.
• After the process completes, search results reflect the current state of Codebeamer and return accurate results.
Disaster Recovery REST API Endpoint
To initiate disaster recovery, the following REST API endpoint has been introduced in the Codebeamer AI 1.2 release:
/api/cbai/v1/indexing/initiate-disaster-recovery — Invokes a full resynchronization of the Codebeamer database and Indexing Service vector database in case of desynchronization.
URL
Input Parameters
|
Parameter
|
Description
|
Value
|
|---|---|---|
|
confirnmationCode
|
Confirmation code required to execute the API.
|
confirm
|
|
|
The confirmation code is hardcoded as confirm to prevent accidental invocation of the operation.
|
Steps to initiate disaster recovery using the API
1. Prepare the request URL:
◦ Replace <Your Codebeamer URL> with your actual Codebeamer server URL.
◦ Ensure the endpoint is: '/api/cbai/v1/indexing/initiate-disaster-recovery?confirmationCode=confirm'
2. Set the HTTP method:
Use the 'POST' method for this request.
3. Add the request header:
Include the following header to specify the response format: -H 'accept: application/json'
4. Provide authentication details
Replace <username> and <password> with valid Codebeamer credentials.
Use basic authentication in the format: -u '<username>:<password>'
The final request would look like this:
curl -X 'POST' '<Your Codebemaer URL>/api/cbai/v1/indexing/initiate-disaster-recovery?confirmationCode=confirm' -H 'accept: application/json' -u '<username>:<password>'
5. Execute the request
Send the POST request using a tool such as curl, Postman, or any REST client.
6. Verify the response
◦ Confirm that the response is returned in JSON format.
◦ Check the response to ensure the disaster recovery process was successfully initiated.