Configuration

Pass your settings in a values file with -f my-values.yaml. Anything you omit keeps the default from the chart's values.yaml. Use helm show values . to see every option with the maintainers' comments.

Settings you will set

Value	Default	Notes
namespace	glyph	Where everything lands. Must equal helm install --namespace and mongo.namespace.
global.domain	glyphapp.io	Domain part of the hostname.
global.subdomain	""	Subdomain part. The host is <subdomain>.<domain>. If left empty, the host renders as .glyphapp.io, which is invalid.
ingress.hosts	[]	Optional explicit host list, used instead of domain and subdomain. The application uses the first entry as its hostname.
ingress.tls.secretName	glyph-app-tls	TLS Secret in istio-system.
server.image.repository	glyph-server	Full repository path in your registry. Set tag to the release SENTRIQS provides.
frontend.image.repository	glyph-frontend	Full repository path in your registry. Set tag to the release SENTRIQS provides.
mongo.image.repository	glyph-mongodb	Set tag to the release SENTRIQS provides. Replaces only the mongod container. An empty repository falls back to the operator's default image.
server.secrets.localSecretName	glyph-app-local-secrets	Name of the application Secret.
global.environment	production	Passed to the server as ENV.

Set image repositories directly

Prefer full repository paths per image over global.imageRegistry. That value needs a trailing slash, applies only to the server and frontend images, and is also read by the SeaweedFS subchart, where it produces a malformed reference (registry//chrislusf/seaweedfs). The MongoDB and wait-for-mongo images ignore it. Override the object storage image with seaweedfs.image.registry and seaweedfs.image.repository, and the init container with server.waitForMongoImage.

Periodic jobs

All four jobs run the server image on a schedule. Each can be switched on and off and rescheduled under cronJobs.<name>.

Job	Default	Schedule	Needs
cleanup	On	Daily, 04:10	MongoDB, outbound email
archiver	Off	00:01, 08:01, 16:01	MongoDB, object storage
analytics	Off	06:10, 18:10	MongoDB, object storage, Snowflake
dailyMetrics	Off	Daily, 03:05	MongoDB, Snowflake

Enable archiver only after you have verified archiving in your own environment with SENTRIQS. Enable analytics and dailyMetrics only when real Snowflake credentials are in the application Secret.

Sizing and availability

The defaults are lean so a first install converges quickly. High availability is a values override, and the MongoDB and RabbitMQ operators handle the clustering for you.

Component	Replicas	Resources	Storage
Frontend	2	Request 100m CPU and 128Mi, limit 500m CPU and 512Mi	None
Server	2	Request 100m CPU and 512Mi, limit 1 CPU and 2Gi	None
MongoDB	1 member	Not set by the chart	20Gi (mongo.storage)
RabbitMQ	1	Request 100m CPU and 256Mi, limit 500m CPU and 512Mi (fixed)	10Gi (rabbitmq.storage)
SeaweedFS	1 each of master, volume, filer, S3	Not set by the subchart	1Gi, 10Gi, 10Gi

• High availability. Raise mongo.members and rabbitmq.replicaCount (for example to 3) and the operators form the replica set and cluster and roll out the change. The RabbitMQ operator is tested with 1, 3, or 5 replicas. SeaweedFS component counts are set under seaweedfs.master.replicas, seaweedfs.volume.replicas, seaweedfs.filer.replicas, and seaweedfs.s3.replicas.
• Server and frontend. Two replicas by default. Change server.replicaCount and frontend.replicaCount. Layer PodDisruptionBudgets, anti-affinity, or autoscaling on top through your own cluster policy.
• Disk. Volume sizes are set when the volume is created. Resizing later depends on your StorageClass.

Known limitations

• No external databases. Setting mongo.enabled or rabbitmq.enabled to false removes the matching server settings. It does not point the server at your own MongoDB or RabbitMQ, because the chart has no values for that.
• One hostname. The application is configured for a single host.
• Leave ingress.paths at its default. All traffic enters through the frontend, which proxies to the server.
• Bucket name is duplicated. storage.bucket and seaweedfs.s3.createBuckets[0].name must be kept equal by hand if you change either.
• skipAwsSecrets does not replace the Secret. Setting it to false adds an AWS Secrets Manager reference, but the Kubernetes Secret and all its keys are still required.
• No scheduling or extension hooks. There are no values for nodeSelector, tolerations, affinity, service account annotations, extra environment variables, or extra volumes. If you need them, ask SENTRIQS.
• No imagePullSecrets. See Prerequisites.

Defaults and names change between pre-1.0 releases. When in doubt, run helm show values . against the exact chart version you are installing.