Private Alien Instance
A Private Alien Instance runs the Alien manager, the server that talks to your deployments, in your own AWS, Google Cloud, or Azure account. Your customers' deployment traffic and logs stay in your cloud, with no new subprocessor for them to review.
Architecture diagram. Deployments (Push: native identity, Pull: Operator) exchanges traffic with the Private Alien Instance (Rollouts, Heartbeats, Logs, Commands), via https 443.
One command starts the setup:
alien instances create --cloud aws --region us-east-1 --openThe manager rolls out releases, receives heartbeats and logs, and delivers commands to your deployments. Alien Cloud, the manager Alien hosts, is for trying Alien. For production, use a Private Alien Instance.
Alien still builds releases, runs the dashboard, and monitors the Private Alien Instance. New deployments connect to their cloud's default manager, so decide which one to use before the first customer deploys.
When to use one
- No new subprocessor. Your customers' deployments talk to your cloud, not to Alien's.
- Logs stay in your account. The manager stores deployment logs in your own bucket, and the browser reads them directly. The Logs page shows End-to-end encrypted.
- Values stay in your account. The Private Alien Instance stores deployment environment variables and secret inputs set after its values key exists, sealed with a key you own. See Deployment values.
- Your region and network. Choose the region, and give pull deployments a custom endpoint domain.
Private Alien Instances need the Pro plan. Some connections still go to Alien-hosted services; Network requirements lists them.
Create one
Open Settings → Private Alien Instances in your project and select Deploy Private Alien Instance.
Pick a cloud and a region, and start initial setup. Each cloud uses its own method:
| Cloud | Initial setup |
|---|---|
| AWS | Launch CloudFormation stack in your account. |
| Google Cloud | Sign in with Google and choose the project to host the manager. |
| Azure | Run Terraform in your Azure subscription with the generated main.tf. |
Follow the progress on the Private Alien Instance's page until it's Healthy. Alien deploys and updates it from then on.
The page shows four steps: Checking the account, Setting up the account, Creating the instance, and Waiting for the first heartbeat. If initial setup fails, it names the step and shows the error. Fix the cause in your account, then select Retry or Provision again.
The same from the CLI:
alien instances create --cloud aws --region us-east-1 --open
alien instances status mgr_9t4k2x --watchUse --setup cloudformation, google-oauth, or terraform to choose the method, and --network default to reuse the account's default network for a test instance. alien managers is an alias of alien instances.
Route deployments to it
Each project has a default manager per cloud. New deployments use it, and existing deployments stay on the manager they started with.
- Push deployments in the same cloud are reached through that cloud's native identity, such as an IAM role.
- Kubernetes pull deployments and Single machine pull deployments connect outbound to it over HTTPS.
Set the defaults on the same page, or from the CLI:
alien instances defaults set --cloud aws --manager mgr_9t4k2x
alien instances defaults showWhen a CLI command falls back to the Alien-hosted manager for a cloud, it prints a warning on stderr and continues. The warning never goes to stdout, so --json output stays parseable.
If you expected your Private Alien Instance, run alien instances defaults show and set the default for that cloud.
Setup CLI
The setup CLI your customer's admin downloads talks to your Private Alien Instance when every platform in the release uses the same Private Alien Instance as its default. Otherwise it talks to Alien's API.
A Private Alien Instance on an older release keeps the setup CLI on Alien's API until it updates. The address is fixed when the package is built, and changing a default doesn't rebuild packages that already exist.
Update a Private Alien Instance
Alien rolls out new releases to a Private Alien Instance like to any deployment. To apply one now, or to go to a specific release:
alien instances update mgr_9t4k2x
alien instances update mgr_9t4k2x --release rel_6w1pz3In the dashboard, open the Private Alien Instance's actions and select Update to the newest release.
When a release adds an optional resource that initial setup creates, such as the key that keeps deployment values in your account, the Private Alien Instance takes the release without it and keeps running. See Optional frozen resources.
To turn the resource on, run its initial setup again:
Open the Private Alien Instance in Settings → Private Alien Instances. Workspace admins see Action required: Update the CloudFormation stack.
Select View steps. The dashboard prepares the update: the newest template URL and a new Token for the stack.
In AWS, open the stack, choose Update stack, then Replace existing template, and paste the template URL. Set the parameters the steps list, such as InputValuesKey: true and the new Token, and keep the others as they are.
Submit the update.
The stack's original setup token has expired by then, so the update needs the new one. Each new token is valid for 7 days. Earlier ones aren't revoked, so an update a teammate already started with one still works.
The dashboard offers this only for a Private Alien Instance set up with CloudFormation on AWS. It also waits until the Private Alien Instance runs a release that can turn the values key on safely. Until then, the notice doesn't appear, and preparing the update over the API fails with PRIVATE_MANAGER_UPDATE_REQUIRED.
Deployment regions
On AWS, a Lambda function pulls its image from a registry in its own region. The Private Alien Instance's registry is in the instance's region. If your deployments run Lambda functions in other AWS regions, list them in the stack's Deployment regions parameter (InputDeploymentRegions), up to 10.
The registry then copies images to those regions. Before Alien creates, updates, or rolls back a function, it checks that the exact image is in the function's region.
- Set the parameter when you launch the stack, or later with a stack update as in Update a Private Alien Instance.
- Images pushed before you added a region aren't copied. Release again with a changed image to copy it.
- AWS allows one replication configuration per account and region. The stack keeps rules the account already has and adds its own.
Endpoint domain
Pull deployments and the browser reach the manager at its endpoint. Keep the Default endpoint, or choose a Custom domain from the verified domains in Settings → Domains.
The endpoint card shows the hostname, the Expected DNS record, and whether DNS and the certificate pass. Select Check again after you change DNS. The endpoint shows Active once pull targets can use it.
From the CLI, pass the domain's ID, or go back to the default endpoint:
alien instances set-domain mgr_9t4k2x --domain dom_4hq8vz
alien instances set-domain mgr_9t4k2x --clearClients that already use the old address, such as pull deployments, keep using it until you update them.
Deployment values
On Alien Cloud's hosted manager, Alien holds deployment values in its own account. A Private Alien Instance with a values key holds these in yours:
- environment variables, plain and secret
- secret stack inputs
- plain stack inputs that only you provide and that Alien doesn't need to plan the deployment
Plain inputs the deployer provides, inputs that turn a resource on or decide whether another input applies, and generated plain values stay with Alien, because setup templates and planning need them.
Values set before the key existed also stay with Alien. See Existing Private Alien Instances.
Each value is sealed with the values-key that initial setup creates in your account: an AWS KMS key, a Google Cloud KMS key, or an Azure Key Vault key. For these values, Alien keeps only metadata: the name, whether it's a secret, when it changed and who changed it, and a keyed digest that shows a value changed without revealing it.
- 1
Ask for a tokenAlien
Alien checks that you may change the deployment and returns a 5-minute token and the Private Alien Instance's URL.
- 2
Send the valueyour account
The dashboard, CLI or API sends the value to the Private Alien Instance with the token. It seals the value with your key.
- 3
Record metadataAlien
The Private Alien Instance returns the name, kind and a keyed digest. Alien stores only that.
The dashboard, the CLI, and the API send values straight to the Private Alien Instance, over https. Alien only issues the short-lived token and records the metadata it returns. When you edit an existing deployment, the form says Stored in Private Alien Instance: <name> and that values go from your browser straight to it.
In the CLI, this covers secret-kind --input and --secret-input, with or without --token. alien deployments create --env and --secret send only the variable names to Alien, and the values go to the Private Alien Instance.
- The Private Alien Instance decrypts a secret value only when it deploys it.
- Secret values are write-only: no client can read one back.
- Plain values it stores are read from the Private Alien Instance, not from Alien.
A request that sends one of these values to Alien instead, from an older CLI or an API call with the value in the request body, is refused with ALIEN_CLI_UPGRADE_REQUIRED. Update the CLI and run the command again.
Setup link values
Secrets and environment variables you enter when you create a setup link go to the Private Alien Instance too. It holds them until a deployment is created from the link, then copies them to that deployment. The link keeps them until you revoke or delete it.
- All platforms the link allows must use the same manager. Otherwise creating the link fails with
SETUP_LINK_VALUES_NEED_ONE_MANAGER. - Links for signed deployments send their values to Alien, as before.
- Links created before the Private Alien Instance held link values keep their values with Alien.
Generated secrets
A generated secret is created inside the Private Alien Instance by the first deploy that needs it, and is never regenerated. Entering a value replaces it. A secret Alien generated before the values key existed stays with Alien, and the Private Alien Instance doesn't generate a second one. A generated input can't be cleared; see Change inputs later.
Values still sent to Alien
- Answers that Alien reads to plan the deployment, such as the yes/no answers the CLI sends before it creates a deployment.
- Secret answers you type in the deployment portal yourself, in your own session.
- Values for a whole project or deployment group.
Existing Private Alien Instances
A Private Alien Instance set up before the values key existed has none, and Alien hosts its values, encrypted with the project key. To store new values in your account, run its initial setup again with Keep deployment values in your account (InputValuesKey) set to true.
On AWS, open the Private Alien Instance in Settings → Private Alien Instances, and select View steps on Action required: Update the CloudFormation stack. The notice appears once the Private Alien Instance runs a release that can turn the key on safely. The dashboard prepares this update only for a Private Alien Instance set up with CloudFormation.
Once the key exists, values set from then on are stored in the Private Alien Instance. Values its deployments already had stay with Alien, and those deployments keep deploying with them. Generated secrets are not regenerated. A new value you enter for an existing deployment is stored in the Private Alien Instance.
When the key is unavailable
If the key is disabled, deleted, or the Private Alien Instance's role can't use it, writing or deploying a value fails with VALUE_STORE_KEY_UNAVAILABLE. For the values it stores, the Private Alien Instance never deploys an empty value and never falls back to Alien's copy.
The Private Alien Instance caches each deployment's data key for up to 5 minutes, so a disabled key can take that long to block deployments that were using it. Re-enable the key and retry; deployments recover without re-entering values.
If the key is deleted, the values sealed with it can't be recovered, and the Private Alien Instance can't store new values until it's repaired. Contact Alien support.
If the Private Alien Instance's own store is lost, its values are lost too. Set them again from the dashboard or CLI.
When it isn't reachable
A Private Alien Instance that stored values can stop reporting that it can load them, for example after a rollback to an older build. Saving a value then fails with Private Alien Instance: <name> isn’t reachable. Try again. Nothing new is sent to Alien. Try again once the Private Alien Instance is healthy.
AWS Virtual Keys
The Private Alien Instance also runs an encryption gateway. New AWS Virtual Keys for deployments it manages use that gateway. Keys created earlier keep the gateway they started with.
Operate it
alien instances list
alien instances status mgr_9t4k2x
alien instances events mgr_9t4k2x --follow
alien instances retry mgr_9t4k2x
alien instances provision mgr_9t4k2x
alien instances delete mgr_9t4k2xretry restarts a failed initial setup or a failed rollout. provision runs provisioning again on the current release, keeping the resources it already created.
alien logs --manager mgr_9t4k2x searches the logs stored by one manager.
Deleting a Private Alien Instance doesn't move its deployments to another manager. Keep it running while it still manages deployments.