Core

Considerations for production

• Make sure to enable monitoring & alerting and monitor the deployments for errors and resource usage. You can create alerts using prometheus alert manager which is installed if following the instructions on the monitoring guide.
• Enable Horizontal Pod Autoscalers and Pod Disruption Budgets where supported.
• Regularly rotate keys such as the categorizer.adminApp.admin.password or api.adminSecret.
• When upgrading a release, compare the generated manifests to what they were before to prevent accidental changes. You can do this easily using the Helm diff plugin.
• Deploy to a namespace using the Helm --namespace argument.
• When managing multiple deployments, use a different release name and/or namespace for each to prevent applying changes to the wrong deployment.
• If using non-managed users, consider setting a api.accounts.newAccountPolicy and api.accounts.forceEmailOtpOnUnknownIpField. You can set forceEmailOtpOnUnknownIpField to IP to force two factor authentication via email when the user attempts to sign in from a new IP without having to set up an ipinfo provider.
• Enable database backups. If using an inCluster database, see db.backups in values.yaml. If on AWS, the postgres operator supports backups directly. You can backup clickhouse by snapshotting the backing volume directly.