Skip to content

Chart settings ​

Fluxify's Helm chart is configured with a values file, fluxify-values.yaml, passed to every helm install and helm upgrade with -f. This page lists every setting it can hold. Install on Kubernetes shows where the file is used.

A values file only needs the settings you change. Anything left out keeps its default.

TIP

To see the chart's full settings file, with a note on every line, for the release you run:

bash
helm show values oci://ghcr.io/fluxify-rest/charts/fluxify --version 0.4.0

Address ​

SettingWhat it doesDefault
urlThe address people open Fluxify at, e.g. https://fluxify.example.com. Sign-in only works at this address.http://localhost:8080
proxyUrlThe proxy's address as the admin reaches it, for sandboxes, the playground, call_route, Run and test suites (PROXY_URL). The admin calls development workers only through the proxy and lets it route by Host, so this is Traefik's in-cluster service, never a single worker. Leave it empty to have admin call the project's own public address instead.http://traefik.traefik.svc.cluster.local:80

Database ​

Set exactly one of these three.

SettingWhat it doesDefault
postgres.urlFrom.secretThe name of a Secret, in Fluxify's namespace, that holds the database address. CloudNativePG makes one named <database>-app.—
postgres.urlFrom.keyThe key inside that Secret.uri
secret.values.PG_URLThe database address itself: postgres://USER:PASSWORD@HOST:5432/DATABASE.—
postgres.bundledRun one Postgres pod inside the chart. No copies, no backups: for trying Fluxify only.false
postgres.storageDisk size for that built-in Postgres.5Gi
postgres.imageThe image for that built-in Postgres.postgres:17-alpine

Keys and passwords ​

All of Fluxify's keys and passwords live in one Secret, fluxify-env. By default the chart creates it and generates every value you leave empty. A generated value is kept across upgrades, and the Secret is kept even after helm uninstall.

SettingWhat it doesDefault
secret.createfalse to create fluxify-env yourself, e.g. from a secret manager. See below.true
secret.values.MASTER_ENCRYPTION_KEYLocks the passwords and keys stored in Fluxify. Must be openssl rand -base64 32. Never change it once set.generated
secret.values.BETTER_AUTH_SECRETSigns sign-in sessions.generated
secret.values.SYSTEM_ACCESS_KEYA key that lets scripts and other systems call Fluxify's API without a user.generated
secret.values.NATS_TOKENNATS's password. Must start with a letter: NATS reads one like 5e3… as a number and does not start.generated
secret.values.REDIS_PASSValkey's password.generated
secret.values.SEED_USER_PASSWORDThe first account's password, used on an empty database only.generated
secret.values.PG_PASSWORDThe built-in Postgres's password.generated
secret.values.PG_URLSee Database.—

WARNING

A values file with passwords in it is a secret itself. Do not commit it to git with them in. Leave them out to have them generated, or use your own Secret.

Fluxify's own settings ​

SettingWhat it doesDefault
envAny other Fluxify setting, by its name in env.example.—
admin.seedUserEmailThe first account's email, used on an empty database only.[email protected]

For example:

yaml
env:
  ENABLE_AI: "true"
  # Add workers sooner: at 50% CPU instead of 65%.
  ORCHESTRATOR_SCALE_CPU_PERCENT: "50"
  OTLP_LOGS_ENDPOINT: https://logs.example.com/v1/logs

Values under env must be text, so quote numbers and true/false.

To send Fluxify's logs to a dashboard, with a complete example, see Logs and monitoring.

Plugins folder ​

Off by default. Upload, npm and link plugins do not need it. To let people add plugins from a folder, give the chart a PersistentVolumeClaim that already holds the built plugins (one subfolder each). Every node must be able to mount it (ReadWriteMany or ReadOnlyMany).

SettingWhat it doesDefault
plugins.volume.existingClaimName of the claim, in Fluxify's namespace. Mounted at /plugins on the admin and the development worker, and read-only on every worker the orchestrator starts.— (off)
plugins.volume.readOnlyMount it read-only on the admin and the development worker.true
yaml
plugins:
  volume:
    existingClaim: fluxify-plugins

If the claim is missing or cannot be mounted, the pods stay Pending until it can.

Resources ​

How much CPU and memory Fluxify's own pods ask for. Production workers are sized per claim in the portal instead: see Sizing a worker.

SettingWhat it doesDefault
admin.resourcesThe portal and API. Test runs execute here, so it has a memory ceiling.250m CPU, 512Mi, at most 2Gi
orchestrator.resourcesThe orchestrator.100m CPU, 128Mi, at most 500m CPU and 512Mi
devWorker.resourcesThe development worker.100m CPU, 256Mi, at most 1Gi
yaml
admin:
  resources:
    requests: { cpu: "1", memory: 1Gi }
    limits: { memory: 4Gi }

Development worker ​

One pod for the development environment. It uses the same worker image as your production workers, serves every project, takes no license slot and needs no claim. It is the only worker the chart installs; the orchestrator does not manage it.

SettingWhat it doesDefault
devWorker.enabledRun the development worker. false if you do not use the development environment.true
devWorker.resourcesThe pod's CPU and memory.100m CPU, 256Mi, at most 1Gi

It is on by default so a Helm install matches the kit and the compose stack. With ingressRoute.enabled, Web traffic sends /_sandbox and /_/dev to it and nothing else, so sandboxes and routes on development can be called on your normal address, and on a project's subdomain (the rule matches any host; a project with a subdomain answers these paths only there). With the route turned off, send /_sandbox, /_/dev and /_/dev/… (not /_/devices) to the service fluxify-dev-worker, port 5600, yourself.

It starts once the orchestrator has written its first settings, a few seconds after the orchestrator is ready, and restarts when you run helm upgrade with changed settings.

yaml
devWorker:
  enabled: false

TIP

A development claim can take its place. A claim made for the Development environment answers exactly the same paths, and the orchestrator manages it for you — make one for every project in Instance settings → Orchestration, then turn this worker off. See Development claims. Both can run together while you switch: a project's own development claim answers first, then the claim for every project, and this worker only sees what neither covers.

Web traffic ​

SettingWhat it doesDefault
ingressRoute.enabledSend /_/admin (the portal, API and MCP server) and /.well-known/oauth-… (sign-in for MCP clients) to Fluxify through Traefik, and /_sandbox and /_/dev to the development worker. false to expose them your own way: send the first two to the service fluxify-admin, port 8080, and /_sandbox and /_/dev to fluxify-dev-worker, port 5600.true
ingressRoute.entryPointThe Traefik entry point it listens on.web

NATS ​

NATS is installed with Fluxify from the official NATS chart: three servers, so losing one loses nothing. Any setting of that chart goes under nats:.

SettingWhat it doesDefault
nats.enabledfalse to use a NATS you already run.true
nats.config.cluster.replicasHow many NATS servers. At least 3 for production.3
nats.config.jetstream.fileStore.pvc.sizeDisk per NATS server.10Gi
nats.container.image.tagNATS version. Fluxify needs 2.14 or newer and exits at start on anything older.2.15.0-alpine
external.natsUrlYour NATS, when nats.enabled is false, e.g. nats://nats.messaging.svc:4222. It must be NATS 2.14+ with JetStream on. Its token goes in secret.values.NATS_TOKEN.—
external.natsMonitoringEndpointYour NATS's monitoring address, host:8222, so workflow claims scale on waiting runs.—

Valkey ​

Valkey is installed with Fluxify from the official Valkey chart. Fluxify only keeps a cache in it, so one server is enough. Any setting of that chart goes under valkey:.

SettingWhat it doesDefault
valkey.enabledfalse to use a Valkey or Redis you already run.true
valkey.replica.enabledKeep copies on other machines, for faster recovery. Also set valkey.replica.persistence.size, e.g. 1Gi.false
external.redisHostYour Valkey or Redis, when valkey.enabled is false. Its password goes in secret.values.REDIS_PASS.—
external.redisPortIts port.6379
external.redisUserIts user, if it has users.—

Images and license ​

SettingWhat it doesDefault
image.tagRun another release's images than the chart's own. Best left alone.the chart's release
image.registryWhere the images come from, for a mirror.ghcr.io/fluxify-rest
image.pullPolicyWhen the cluster pulls images again.IfNotPresent
license.issuerKeySecretA Secret holding issuer.pub, for a self-signed license. Before 1.0 only.—

Example files ​

Production with CloudNativePG ​

Install on Kubernetes step 6's production file, with more room for the portal and API once many people use it:

yaml
url: https://fluxify.example.com

postgres:
  urlFrom:
    secret: fluxify-db-app

admin:
  resources:
    requests: { cpu: "1", memory: 1Gi }
    limits: { memory: 4Gi }

Your own NATS and Valkey ​

yaml
url: https://fluxify.example.com

postgres:
  urlFrom:
    secret: fluxify-db-app

nats:
  enabled: false
valkey:
  enabled: false

external:
  natsUrl: nats://nats.messaging.svc:4222
  natsMonitoringEndpoint: nats.messaging.svc:8222
  redisHost: valkey.cache.svc

Put their passwords in the Secret as NATS_TOKEN and REDIS_PASS: with --set-string secret.values.NATS_TOKEN=... on the command line, or in your own Secret.

Keys from your own Secret ​

With a secret manager (External Secrets, Sealed Secrets, Vault), create a Secret named fluxify-env in Fluxify's namespace before installing, and turn off the chart's:

yaml
secret:
  create: false

It must hold these keys:

KeyHow to make it
MASTER_ENCRYPTION_KEYopenssl rand -base64 32
BETTER_AUTH_SECRETopenssl rand -hex 32
SYSTEM_ACCESS_KEYopenssl rand -hex 32
NATS_TOKENn$(openssl rand -hex 31). It must start with a letter.
REDIS_PASSopenssl rand -hex 32
SEED_USER_PASSWORDA password you choose
PG_URLYour database address, unless you use postgres.urlFrom
PG_PASSWORDopenssl rand -hex 32, only with postgres.bundled

Trial on a laptop ​

yaml
url: http://localhost:8080

postgres:
  bundled: true

# One NATS server is enough to try things.
nats:
  config:
    cluster:
      enabled: false

Released under the Apache License 2.0. Enterprise features are under the Fluxify Enterprise Edition License.