Fleet Operations

Fleet Operations lets you issue commands to Cloud OS instances from the Control Panel. Commands are queued, tracked through their lifecycle, and can target individual servers or groups of instances in bulk.

Command System

The Control Panel uses a command queue model. When you issue a command, it is placed in a queue and delivered to the target instance via the mTLS agent connection. The instance executes the command and reports the result back.

Command Lifecycle

Every command moves through these statuses:

┌─────────┐    ┌─────────┐    ┌───────────┐    ┌──────────┐
│ Pending │───►│ Running │───►│ Completed │    │  Failed  │
└─────────┘    └─────────┘    └───────────┘    └──────────┘
                    │                               ▲
                    └───────────────────────────────┘

Where to find it

Open Operations in the sidebar. The page has four sections, picked with the tabs at the top right (the URL keeps the choice as ?tab=):

App migrations live next door under Operations → Migrations.

Mass deploy and rolling update

1. Open Operations → Mass deploy
2. Tick the target servers (filter by name, region or tag; Select all picks every server)
3. Either type an app template and click Deploy, or enter a version (empty = latest) and click Start rolling update

Both start one bulk job, so you can follow the progress right there or under Jobs. For an update that should move through the fleet in stages and stop on errors, use a staged rollout instead.

Command Types

The Control Panel supports several categories of commands:

App Management

• Install App — install an application from the app template catalog on the target instance
• Uninstall App — remove an installed application
• Update App — update an application to a newer template version
• Start / Stop / Restart App — control the lifecycle of an installed application

System Operations

• Update Cloud OS — trigger a binary update on the target instance
• Restart Cloud OS — restart the Cloud OS service
• Run Health Check — request an immediate health report from the instance

Configuration

• Apply Config — push configuration changes to an instance
• Sync Settings — synchronize settings from a baseline configuration

Bulk Operations

You can act on many nodes at once from the Fleet page: tick the nodes you want, or apply the action to everything that matches the current filter (status, region, group, version, tenant, search).

Bulk Jobs

Every bulk action creates one job that fans out to the selected nodes (up to 10,000 per job). The Jobs list shows each job with its progress (done / total), how many succeeded and failed, who started it and when. Open a job to see the result of every node — failures are listed first with the node’s error message.

• Actions that need the node (restart, update, run command) complete when each node reports back; offline nodes pick the command up when they reconnect, and a node that never answers is marked failed after 24 hours.
• Cancel stops a running job: nodes that have not started yet are skipped and commands not yet picked up by a node are withdrawn. Commands a node already received still finish and are counted.
• A job ends as succeeded, partial (some nodes failed or were cancelled), failed (all failed) or cancelled.

Bulk actions need the fleet write permission, like single-node actions, and each job start and cancel is recorded in the audit log.

App Template Catalog Distribution

The Control Panel maintains a centralized catalog of app templates that can be distributed to Cloud OS instances. When you install an app via Fleet Operations, the template is delivered to the target instance along with the install command.

This allows you to maintain a curated set of approved applications and deploy them consistently across your fleet.

Command History

Operations → Jobs → Command history shows all previously issued commands with:

• Command type and parameters
• Target instance or group
• Status (pending, running, completed, failed)
• Timestamp of creation and completion
• Error details for failed commands

You can sort the history by status, type and time.

Queue Management

Canceling Commands

Pending commands that have not yet been delivered can be canceled from the command queue. Once a command reaches “running” status, it cannot be canceled from the Control Panel.

Retry Failed Commands

Failed commands can be retried with a single click. The Control Panel creates a new command entry with the same parameters.

Offline Instance Handling

Commands targeted at offline instances remain in “pending” status. When the instance reconnects and sends a heartbeat, pending commands are delivered automatically.

Next Steps

• Fleet Management — register and monitor your instances
• Governance — enforce policies and automate deployments