Upgrades¶
AgentCube connectors are upgraded by deploying a new container image version.
Upgrade Procedure¶
1. Check the New Version¶
Review the release notes for the new version to understand what has changed.
2. Pull and Deploy the New Image¶
Update your container deployment to use the new image version:
az containerapp update \
--name {app_name} \
--resource-group {resource_group} \
--image ghcr.io/caprusai/agentcube-essbase:{new_version} \
--revision-suffix "v{new_version}"
Tip
Always include a unique --revision-suffix to ensure the platform pulls the new image rather than using a cached version.
aws ecs update-service \
--service agentcube-essbase \
--task-definition agentcube-essbase:{new_revision} \
--force-new-deployment \
--region {region}
If the task definition references the image by tag, update the tag in the task definition and register a new revision before running the above command.
Pull the new image and restart the service:
Update the image tag in your docker-compose.yml, then:
kubectl set image deployment/agentcube-essbase \
agentcube-essbase=ghcr.io/caprusai/agentcube-essbase:{new_version} \
-n agentcube
Kubernetes will perform a rolling update, keeping the existing pod running until the new one is healthy.
OCI Container Instances do not support in-place image updates. Delete the existing instance and create a new one with the updated image tag using the original creation parameters.
3. Verify the Upgrade¶
Confirm:
versionmatches the new versionstatusishealthybackend_connectedistrue(basic mode) ornull(OIDC mode)
4. Test a Tool Call¶
From your AI platform, run a tool call (e.g., list_applications) to verify end-to-end functionality.
Version Numbering¶
Essbase and Planning connectors are versioned independently using semantic versioning:
| Format | Example |
|---|---|
| Essbase | essbase-v{major}.{minor}.{patch} |
| Planning | planning-v{major}.{minor}.{patch} |
The SDK version is embedded in each connector image and reported in the health endpoint.
Rollback¶
If the upgrade causes issues, see Rollbacks to revert to the previous version.
AI Platform Notes¶
Copilot Studio tool cache
After upgrading a connector, Copilot Studio may cache stale tool definitions. Disconnect and reconnect the MCP connector in your agent's Actions settings to refresh.
Claude.ai automatically picks up tool definition changes on the next conversation.