[[TOC]]
Big Bang follows a GitOps approach to managing the Big Bang Kubernetes cluster configuration. Using GitOps, we must securely store secrets in Git using encryption. The private key, which is stored in key storage, is used by the continuous deployment tool to decrypt and deploy the secrets for use in the cluster.
Secrets Operations (SOPS) is used to securely encrypt values in YAML, JSON, ENV, INI and BINARY formats. Secrets, such as pull credentials or certificates, should be encrypted with SOPS prior to committing into a Git repository.
The private key used in SOPS should NEVER be stored in Git along side the encrypted secrets.
SOPS supports the ability to add multiple keys to the same file so multiple key pairs can use the same secret. This is useful for environments which may have different keys, but use the same secrets. For each key used, SOPS writes the public key, used to encrypt, and an encrypted copy of the data to the file. Decryption requires use of one of the private keys used. After editing, the embedded public keys are used to re-encrypt the file for all key pairs.
To setup Big Bang with SOPS, a key pair must be created. The private key is used for decryption and must be securely stored but accessible to the cluster. The public key is used for encryption. Follow the appropriate instructions below to create your key pair.
Key Management | Key Pair Instructions | Notes |
---|---|---|
GNU Privacy Guard (GPG)* | gpg --full-generate-key |
Use key type = RSA and RSA , keysize = 4096 , expiration = 0 |
Amazon Web Services (AWS) Key Management Service (KMS) | Link | Advanced setup help (e.g. roles, profiles, contexts) |
Google Cloud Platform (GCP) Key Management Service (KMS) | Link | |
HashiCorp Vault | Link |
*GPG is not recommended for production use because the private key can be misplaced or compromised too easily
SOPS uses .sops.yaml
as a configuration file for which keys to use for newly created files. Once a file is created, the key fingerprints are stored in the file and must be re-keyed to use any changes to .sops.yaml
.
-
Follow the SOPS instructions to configure
.sops.yaml
based on the encryption method you used. Multiple keys of the same type can be added using the block scalar yaml construct,>-
, and separating them by a comma and newline.If you are using the Big Bang sample files, make sure to remove the development Big Bang key.
-
Add the following regex to only encrypt data in the yaml files
creation_rules: - encrypted_regex: "^(data|stringData)$"
-
Save
.sops.yaml
in the root of folder of your configuration -
If you have existing secrets, use the following to re-key them with the configuration in
.sops.yaml
# You must have the old private key to rekey the file sops updatekeys <encrypted file>
This must be completed before deploying Big Bang or else deploying Secrets will fail.
-
Deploy your SOPS private key to a secret named
sops-gpg
in the clustergpg --export-secret-keys --armor <new key fingerprint> | kubectl create secret generic sops-gpg -n bigbang --from-file=yourkey.asc=/dev/stdin
-
Configure your KMS key(s) in your
.sops.yaml
by adding the target key's ARN to thekms
field within each creation rule.creation_rules: - encrypted_regex: "^(data|stringData)$" path_regex: ./dev/.* kms: "<kms_key_arn>"
-
Ensure your cluster (specifically the
flux-system/flux-controller
) has access to the specified key.-
For AWS deployments, this can be managed via IAM roles as described in the SOPS documentation.
-
For non-AWS deployments
-
Create an AWS user with appropriate permissions as described in the SOPS documentation.
-
Create a secret named
sops-aws-creds
in the cluster using the access creds from the target user:k create secret generic -n flux-system sops-aws-creds --from-literal=access_key_id=<key_id> --from-literal=access_key_secret=<key>
-
-
TBD - This article may help to automate secret consumption in Kubernetes.
TBD - This article may help to automate secret consumption in Kubernetes.
TBD - This article may help to automate secret consumption in Kubernetes.
Big Bang needs to know how to retrieve the private key so it can deploy the encrypted secrets from Git. Decryption configuration is placed in the top-level manifest (e.g. dev.yaml
, prod.yaml
) from the Big Bang template.
By default, the Kustomization
resource uses a Secret named sops-gpg
for the private key as shown here:
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: environment
spec:
decryption:
provider: sops
secretRef:
name: sops-gpg
Configure the Kustomization
resource to use sops for decryption:
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: environment
spec:
decryption:
provider: sops
Note, we are not providing the
secretRef
field, which is specific to GPG
If Big Bang is deployed within AWS, KMS key access can be handled via IAM roles and permissions on the cluster resources themselves. However, if the deployment is in a different environment from the KMS keys, AWS credentials may need to be provided via a secret as follows.
Configure the flux-system kustomize-controller
component with AWS credential environment variables using kustomize
. Specific instructions for doing this may vary by deployment and environment but an example is covered in the bigbang template repo. Broadly speaking, adding environment variables to the kustomize-controller
component can be accomplished by adding a patch to the flux/kustomization.yaml
for the target deployment or environment. An example of such a kustomization.yaml
is shown below:
bases:
- ../../base/flux
patchesStrategicMerge:
- |-
apiVersion: apps/v1
kind: Deployment
metadata:
name: kustomize-controller
namespace: flux-system
spec:
template:
spec:
containers:
- name: manager
env:
- name: AWS_ACCESS_KEY_ID
valueFrom:
secretKeyRef:
name: sops-aws-creds
key: access_key_id
- name: AWS_SECRET_ACCESS_KEY
valueFrom:
secretKeyRef:
name: sops-aws-creds
key: access_key_secret
Values should come from the
sops-aws-creds
secret created in AWS KMS above
TBD - Instructions on how to update for GCP, Vault