[🐛 Bug]: Failed to load Incoming Webhook settings

[kfkawalec]

Describe the bug
I can display the webhook settings and see the authorisation token.

To Reproduce
Go to Settings
Select Incoming Webhook tab

Screenshots

Additional context
Version 0.47.11.
Helm deployment.
MySQL database, most default settings.
After a fresh deployment, the webhook works - after some time, it stops working.
I checked #3400 and found that I could create a redeployment, but the settings stopped working again.
I cannot see any file called 'keep-webhook' inside the Docker container.

[dosubot]

Hi @kfkawalec! I'm Dosu and I’m helping the keep team.

This is a known issue in Helm/Kubernetes deployments when Keep uses local files for secret management. The error happens because the backend expects a file like ./keep-webhook to exist, but in multi-replica or ephemeral environments, these files aren't shared or persisted across containers. After redeployment or pod rescheduling, the file can disappear, causing the Incoming Webhook settings to fail to load—even though the database is intact. Redeploying temporarily fixes it, but the problem recurs because the underlying file-based secret manager is stateful and not suited for this environment.

The recommended fix is to set SECRET_MANAGER_TYPE=k8s in your deployment. This switches Keep to use Kubernetes-native secrets, making it stateless and resolving the missing file issue in multi-replica or ephemeral setups. This approach is confirmed by both user and maintainer reports as the solution for production Helm deployments with multiple replicas or rolling updates [source].

No changes to the database or manual file creation are needed—just update your Helm values or deployment environment to use SECRET_MANAGER_TYPE=k8s and redeploy. This will prevent the error from recurring.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}