Flux Instance CRD
FluxInstance is a declarative API for the installation, configuration and automatic upgrade of the Flux distribution.
A single custom resource of this kind can exist in a Kubernetes cluster
with the name flux
that must be created in the same namespace
where the flux-operator is deployed.
Example
The following example shows a FluxInstance custom resource that installs the upstream Flux distribution with all available components, and configures the flux-operator to automatically upgrade Flux to the latest stable version:
apiVersion: fluxcd.controlplane.io/v1
kind: FluxInstance
metadata:
name: flux
namespace: flux-system
annotations:
fluxcd.controlplane.io/reconcile: "enabled"
fluxcd.controlplane.io/reconcileEvery: "1h"
fluxcd.controlplane.io/reconcileTimeout: "3m"
spec:
distribution:
version: "2.x"
registry: "ghcr.io/fluxcd"
components:
- source-controller
- kustomize-controller
- helm-controller
- notification-controller
- image-reflector-controller
- image-automation-controller
cluster:
type: kubernetes
multitenant: false
networkPolicy: true
domain: "cluster.local"
storage:
class: "standard"
size: "10Gi"
kustomize:
patches:
- target:
kind: Deployment
name: "(kustomize-controller|helm-controller)"
patch: |
- op: add
path: /spec/template/spec/containers/0/args/-
value: --concurrent=10
- op: add
path: /spec/template/spec/containers/0/args/-
value: --requeue-dependency=5s
You can run this example by saving the manifest into fluxinstance.yaml
.
- Apply the resource on the cluster:
kubectl apply -f fluxinstance.yaml
- Run
kubectl get fluxinstance
to see the status of the resource:
$ kubectl -n flux-system get fluxinstance
NAME AGE READY STATUS REVISION
flux 59s True Reconciliation finished in 52s v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219
-
Run
kubectl describe fluxinstance
to see the reconciliation status components, conditions and events:$ kubectl -n flux-system describe fluxinstance flux Status: Components: Digest: sha256:161da425b16b64dda4b3cec2ba0f8d7442973aba29bb446db3b340626181a0bc Name: source-controller Repository: ghcr.io/fluxcd/source-controller Tag: v1.3.0 Digest: sha256:48a032574dd45c39750ba0f1488e6f1ae36756a38f40976a6b7a588d83acefc1 Name: kustomize-controller Repository: ghcr.io/fluxcd/kustomize-controller Tag: v1.3.0 Digest: sha256:a67a037faa850220ff94d8090253732079589ad9ff10b6ddf294f3b7cd0f3424 Name: helm-controller Repository: ghcr.io/fluxcd/helm-controller Tag: v1.0.1 Digest: sha256:c0fab940c7e578ea519097d36c040238b0cc039ce366fdb753947428bbf0c3d6 Name: notification-controller Repository: ghcr.io/fluxcd/notification-controller Tag: v1.3.0 Digest: sha256:aed795c7a8b85bca93f6d199d5a14bbefaf925ad5aa5316b32a716cfa4070d0b Name: image-reflector-controller Repository: ghcr.io/fluxcd/image-reflector-controller Tag: v0.32.0 Digest: sha256:ab5097213194f3cd9f0e68d8a937d94c4fc7e821f6544453211e94815b282aa2 Name: image-automation-controller Repository: ghcr.io/fluxcd/image-automation-controller Tag: v0.38.0 Conditions: Last Transition Time: 2024-06-03T12:20:57Z Message: Reconciliation finished in 52s Observed Generation: 1 Reason: ReconciliationSucceeded Status: True Type: Ready Last Applied Revision: v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219 Last Attempted Revision: v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219 Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal Progressing 6m20s flux-controller Installing revision v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219 Normal ReconciliationSucceeded 5m9s flux-controller Reconciliation finished in 52s
-
Run
kubectl logs
on the flux-operator pod to see the reconciliation logs:kubectl -n flux-system logs deployment/flux-operator
-
Run
kubectl events
to see the events generated by the flux-operator:kubectl -n flux-system events --for FluxInstance/flux
-
Run
kubectl delete
to remove the FluxInstance resource and to uninstall Flux without affecting any Flux-managed workloads:kubectl -n flux-system delete FluxInstance/flux
Writing a FluxInstance spec
As with all other Kubernetes config, a FluxInstance needs apiVersion
,
kind
, and metadata
fields. The name of a FluxInstance object must be a
valid DNS subdomain name.
A FluxInstance also needs a
.spec
section.
Distribution configuration
The .spec.distribution
field is required and specifies the Flux distribution to install.
Example using the upstream Flux distribution:
spec:
distribution:
version: "2.x"
registry: "ghcr.io/fluxcd"
Distribution version
The .spec.distribution.version
field is required and specifies the version of the Flux distribution to install.
The version field value must be a valid semver range or an exact version.
Example using a semver range to configure the automatic upgrade to the latest Flux minor version:
spec:
distribution:
version: "2.x"
Example using a semver range to configure the automatic upgrade
to the latest Flux patch version of the 2.3
series:
spec:
distribution:
version: "2.3.x"
Example using an exact version to install a specific Flux version:
spec:
distribution:
version: "2.3.0"
Distribution registry
The .spec.distribution.registry
field is required and specifies the container registry
where the Flux distribution images are pulled from.
Example using the upstream Flux distribution registry:
spec:
distribution:
version: "2.x"
registry: "ghcr.io/fluxcd"
Distribution image pull secret
The .spec.distribution.imagePullSecret
field is optional and specifies the name of the Kubernetes secret
that contains the credentials to pull the Flux distribution images from a private registry.
Example using the ControlPlane enterprise registry:
spec:
distribution:
version: "2.3.x"
registry: "ghcr.io/controlplaneio-fluxcd/distroless"
imagePullSecret: "flux-enterprise-auth"
The image pull secret must be created in the same namespace where the FluxInstance is deployed
and must be of type kubernetes.io/dockerconfigjson
.
Example generating a secret for the ControlPlane enterprise registry:
kubectl create secret docker-registry flux-enterprise-auth \
--namespace flux-system \
--docker-server=ghcr.io \
--docker-username=flux \
--docker-password=$ENTERPRISE_TOKEN
Distribution artifact
The .spec.distribution.artifact
field is optional and specifies the OCI artifact URL
containing the Flux distribution manifests. When specified, the operator will pull the
artifact on a regular interval to determine the latest Flux version available
including CVE patches and hotfixes.
Example using the official distribution artifact:
spec:
distribution:
version: "2.x"
registry: "ghcr.io/fluxcd"
artifact: "oci://ghcr.io/controlplaneio-fluxcd/flux-operator-manifests"
Components configuration
The .spec.components
field is optional and specifies the list of Flux components to install.
When not specified, the operator will install the default set of components for the Flux distribution:
spec:
components:
- source-controller
- kustomize-controller
- helm-controller
- notification-controller
Cluster configuration
The .spec.cluster
field is optional and specifies the Kubernetes cluster configuration.
Example using the OpenShift cluster configuration:
spec:
cluster:
type: openshift
multitenant: true
tenantDefaultServiceAccount: "flux"
networkPolicy: true
domain: "cluster.local"
Cluster type
The .spec.cluster.type
field is optional and specifies the type of the Kubernetes cluster.
This field is used to enable specific configuration for AKS, EKS, GKE and OpenShift clusters.
The supported values are kubernetes
(default), openshift
, aks
, eks
and gke
.
Cluster multitenant
The .spec.cluster.multitenant
field is optional and specifies whether to enable Flux
multi-tenancy lockdown.
The .spec.cluster.tenantDefaultServiceAccount
is optional and specifies the default
service account used by Flux when reconciling Kustomization
and HelmRelease
resources found in the tenant namespaces.
Cluster network policy
The .spec.cluster.networkPolicy
field is optional and specifies whether to restrict network access
to the Flux namespace from other namespaces. By default, network policy is enabled.
Cluster domain
The .spec.cluster.domain
field is optional and specifies the cluster internal domain name.
By default, the domain is set to cluster.local
.
Storage configuration
The .spec.storage
field is optional and specifies the persistent storage for Flux internal artifacts.
When specified, the operator will create a persistent volume claim named source-controller
with
the specified storage class and size and mount it to the Flux source-controller /data
volume.
Storage class
The .spec.storage.class
field is required and specifies the storage class to use for the persistent volume claim.
Example using the standard storage class:
spec:
storage:
class: "standard"
size: "10Gi"
Storage size
The .spec.storage.size
field is required and specifies the size of the persistent volume claim.
Sharding configuration
The .spec.sharding
field is optional and specifies the sharding configuration for the Flux controllers.
Example:
spec:
sharding:
key: "sharding.fluxcd.io/key"
shards:
- "shard1"
- "shard2"
For each shard, the operator will create a separate set of controllers, e.g.:
$ kubectl -n flux-system get deployments -l app.kubernetes.io/part-of=flux
NAME
source-controller
source-controller-shard1
source-controller-shard2
kustomize-controller
kustomize-controller-shard1
kustomize-controller-shard2
helm-controller
helm-controller-shard1
helm-controller-shard2
Note that only the source-controller
, kustomize-controller
and helm-controller
controllers
support sharding.
To assign a resource to a specific shard, add the sharding.fluxcd.io/key
label with the shard value,
e.g.: sharding.fluxcd.io/key: shard1
.
Sharding key
The .spec.sharding.key
field is optional and specifies the sharding key label to use for the Flux controllers.
By default, the key is set to sharding.fluxcd.io/key
.
Shards
The .spec.sharding.shards
field is required and specifies the list of sharding values to use for the Flux controllers.
Kustomize patches
The .spec.kustomize.patches
field is optional and specifies the Kustomize patches to apply to the Flux controllers.
Example:
spec:
kustomize:
patches:
- target:
kind: Deployment
name: "(kustomize-controller|helm-controller)"
patch: |
- op: add
path: /spec/template/spec/containers/0/args/-
value: --concurrent=10
- op: add
path: /spec/template/spec/containers/0/args/-
value: --requeue-dependency=5s
Reconciliation configuration
The reconciliation behaviour can be configured using the following annotations:
fluxcd.controlplane.io/reconcile
: Enable or disable the reconciliation loop. Default isenabled
, set todisabled
to pause the reconciliation.fluxcd.controlplane.io/reconcileEvery
: Set the reconciliation interval. Default is1h
.fluxcd.controlplane.io/reconcileTimeout
: Set the reconciliation timeout. Default is5m
.
Sync configuration
The .spec.sync
field is optional and specifies the Flux sync configuration.
When set, a Flux source and a Flux Kustomization are generated to sync
the cluster state with the source repository.
The Flux objects are created in the same namespace where the FluxInstance is deployed
using the namespace name as the Flux source and Kustomization name. The naming convention
matches the one used by flux bootstrap
to ensure compatibility with upstream, and
to allow transitioning a bootstrapped cluster to a FluxInstance managed one.
Sync fields:
kind
: The source kind, supported values areGitRepository
,OCIRepository
andBucket
.url
: The URL of the source repository, can be a Git repository HTTP/S or SSH address, an OCI repository address or a Bucket endpoint.ref
: The source reference, can be a Git ref name e.g.refs/heads/main
, an OCI tag e.g.latest
or a Bucket name.path
: The path to the source directory containing the kustomize overlay or plain Kubernetes manifests to sync from.pullSecret
: The name of the Kubernetes secret that contains the credentials to pull the source repository. This field is optional.interval
: The sync interval. This field is optional, when not set the default is1m
.
Sync from Git over HTTP/S
Example:
spec:
sync:
kind: GitRepository
url: "https://gitlab.com/my-group/my-fleet.git"
ref: "refs/heads/main"
path: "clusters/my-cluster"
pullSecret: "git-token-auth"
If the source repository is private, the Kubernetes secret must be created in the same namespace where the FluxInstance is deployed, and have the following format:
apiVersion: v1
kind: Secret
metadata:
name: git-token-auth
namespace: flux-system
type: Opaque
stringData:
username: "git-username"
password: "git-token"
To generate the secret with the Flux CLI:
flux create secret git git-token-auth \
--namespace flux-system \
--url=https://gitlab.com/my-group/my-fleet.git \
--username=git-username \
--password=git-token
Sync from Git over SSH
Example:
spec:
sync:
kind: GitRepository
url: "ssh://[email protected]/my-org/my-fleet.git"
ref: "refs/heads/main"
path: "clusters/my-cluster"
pullSecret: "git-ssh-auth"
If the source repository is private, the Kubernetes secret must be created in the same namespace where the FluxInstance is deployed, and have the following format:
apiVersion: v1
kind: Secret
metadata:
name: git-ssh-auth
namespace: flux-system
type: Opaque
stringData:
identity: |
-----BEGIN OPENSSH PRIVATE KEY-----
...
-----END OPENSSH PRIVATE KEY-----
known_hosts: |
github.com ecdsa-sha2-nistp256 AAAA...
To generate the secret with the Flux CLI:
flux create secret git git-ssh-auth \
--namespace flux-system \
--url=ssh://[email protected]/my-org/my-fleet.git \
--private-key-file=my-private.key
Sync from OCI over HTTP/S
Example:
spec:
sync:
kind: OCIRepository
url: "oci://ghcr.io/my-org/my-fleet-manifests"
ref: "latest"
path: "clusters/my-cluster"
pullSecret: "oci-token-auth"
If the container registry is private, the Kubernetes secret must be created
in the same namespace where the FluxInstance is deployed, and be of type kubernetes.io/dockerconfigjson
:
apiVersion: v1
kind: Secret
metadata:
name: oci-token-auth
namespace: flux-system
type: kubernetes.io/dockerconfigjson
data:
.dockerconfigjson: "base64-encoded-docker-config"
To generate the secret with the Flux CLI:
flux create secret oci oci-token-auth \
--namespace flux-system \
--url=ghcr.io \
--username=ghcr-username \
--password=ghcr-token
Sync from S3-compatible storage over HTTP/S
Example:
spec:
sync:
kind: Bucket
url: "minio.my-org.com"
ref: "my-bucket-fleet"
path: "clusters/my-cluster"
pullSecret: "bucket-auth"
If the Bucket is private, the Kubernetes secret must be created in the same namespace where the FluxInstance is deployed, and have the following format:
apiVersion: v1
kind: Secret
metadata:
name: bucket-auth
namespace: flux-system
type: Opaque
stringData:
accesskey: "my-accesskey"
secretkey: "my-secretkey"
Resources migration configuration
The .spec.migrateResources
field is optional and instructs the operator to migrate
the Flux custom resources stored in Kubernetes etcd to the latest API version as
specified in the Flux CRDs. The migration runs after the Flux distribution is upgraded
from a minor version to another and only when a new API version is introduced.
By default, the field value is set to true
. Note that disabling the migration may
result in upgrade failures due to deprecated API versions being removed in future Flux releases.
FluxInstance Status
Conditions
A FluxInstance enters various states during its lifecycle, reflected as Kubernetes Conditions. It can be reconciling while applying the resources on the cluster, it can be ready, or it can fail during reconciliation.
The FluxInstance API is compatible with the kstatus specification,
and reports Reconciling
and Stalled
conditions where applicable to
provide better (timeout) support to solutions polling the Kustomization to
become Ready
.
Reconciling FluxInstance
The flux-operator marks a FluxInstance as reconciling when it starts
the reconciliation of the same. The Condition added to the FluxInstance's
.status.conditions
has the following attributes:
type: Reconciling
status: "True"
reason: Progressing
|reason: ProgressingWithRetry
The Condition message
is updated during the course of the reconciliation to
report the action being performed at any particular moment such as
building manifests, detecting drift, etc.
The Ready
Condition's status
is also marked as Unknown
.
Ready FluxInstance
The flux-operator marks a FluxInstance as ready when the Flux configuration was built and applied on the cluster and all health checks are observed to be passing.
When the FluxInstance is "ready", the flux-operator sets a Condition with the
following attributes in the FluxInstance’s .status.conditions
:
type: Ready
status: "True"
reason: ReconciliationSucceeded
Failed FluxInstance
The flux-operator may get stuck trying to reconcile and apply a FluxInstance without completing. This can occur due to some of the following factors:
- The distribution artifact is not accessible.
- The distribution version is not available.
- The kustomization of the Flux components fails to build.
- Garbage collection fails.
- Running health checks fails.
When this happens, the flux-operator sets the Ready
Condition status to False
and adds a Condition with the following attributes to the FluxInstance’s
.status.conditions
:
type: Ready
status: "False"
reason: ArtifactFailed | BuildFailed | HealthCheckFailed | ReconciliationFailed
The message
field of the Condition will contain more information about why
the reconciliation failed.
While the FluxInstance has one or more of these Conditions, the flux-operator will continue to attempt a reconciliation with an exponential backoff, until it succeeds and the FluxInstance is marked as ready.
Components status
In order to provide visibility into the Flux components that are installed,
the flux-operator records the status of each component in the .status.components
field,
including the image repository, tag and digest.
Example:
Status:
Components:
Digest: sha256:161da425b16b64dda4b3cec2ba0f8d7442973aba29bb446db3b340626181a0bc
Name: source-controller
Repository: ghcr.io/fluxcd/source-controller
Tag: v1.3.0
Digest: sha256:48a032574dd45c39750ba0f1488e6f1ae36756a38f40976a6b7a588d83acefc1
Name: kustomize-controller
Repository: ghcr.io/fluxcd/kustomize-controller
Tag: v1.3.0
Inventory status
In order to perform operations such as drift detection, garbage collection, upgrades, etc.,
the flux-operator needs to keep track of all Kubernetes objects that are
reconciled as part of a FluxInstance. To do this, it maintains an inventory
containing the list of Kubernetes resource object references that have been
successfully applied and records it in .status.inventory
. The inventory
records are in the format Id: <namespace>_<name>_<group>_<kind>, V: <version>
.
Example:
Status:
Inventory:
Entries:
Id: flux-system_source-controller__ServiceAccount
V: v1
Id: flux-system_source-controller__Service
V: v1
Id: flux-system_source-controller_apps_Deployment
V: v1
Last applied revision
.status.lastAppliedRevision
is the last revision of the Flux distribution
that was successfully applied to the cluster.
The revision is in the format <version>@sha256:<digest>
.
The version is the Flux distribution exact semver version that was applied to the cluster.
The digest is the SHA256 hash of the Flux distribution manifests and customisations that was applied to the cluster.
Last attempted revision
.status.lastAttemptedRevision
is the last revision of the Flux distribution
that was attempted to be applied to the cluster.
Example:
Status:
Last Applied Revision: v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219
Last Attempted Revision: v2.3.0@sha256:4cc5babdb1279ad0177bf513292deadbfa3f7b7c3da0be7fa53b39ab434f7219
FluxInstance Metrics
The Flux Operator exports metrics for the FluxInstance resource. These metrics are refreshed every time the operator reconciles the instance.
Metrics:
flux_instance_info{uid, kind, name, exported_namespace, ready, suspended, registry, revision}
Labels:
uid
: The Kubernetes unique identifier of the resource.kind
: The kind of the resource (e.g.FluxInstance
).name
: The name of the resource (e.g.flux
).exported_namespace
: The namespace where the resource is deployed (e.g.flux-system
).ready
: The readiness status of the resource (e.g.True
,False
orUnkown
).reason
: The reason for the readiness status (e.g.Progressing
,BuildFailed
,HealthCheckFailed
, etc.).suspended
: The suspended status of the resource (e.g.True
orFalse
).registry
: The container registry used by the instance (e.g.ghcr.io/fluxcd
).revision
: The Flux revision installed by the instance (e.g.v2.3.0@sha256:75aa209c6a...
).
Example:
flux_instance_info{
exported_namespace="flux-system",
kind="FluxInstance",
name="flux",
ready="True",
reason="ReconciliationSucceeded",
registry="ghcr.io/fluxcd",
revision="v2.3.0@sha256:75aa209c6a2e25b97114ccf092246d02ab4363bc136edefc239d2a88da882b63",
suspended="False",
uid="16ca7202-9319-445b-99d0-617c25bda182"
}