Fleet Operations¶

Fleet operations let you coordinate changes across many hosts safely — with batch control, abort thresholds, health checks between batches, and rollback on failure.

Rolling Restart¶

Restart a service across a fleet N hosts at a time:

Configure batch_size (how many hosts per batch) and abort_threshold (maximum failure rate before halting)
A health check command runs between batches — if it fails, the restart stops
If the failure rate across any batch exceeds the abort threshold, the campaign halts and triggers rollback on completed hosts

Canary Config Push¶

Deploy a config change to one host first, verify it, then roll to the rest:

Push config to the canary host
Run a verification command against the canary
If the verification passes, roll config to all remaining hosts
If verification fails, rollback restores the original file on the canary host

The original file is backed up before being replaced. Rollback restores it exactly.

Bulk File Distribution¶

Push files to a fleet in a single change request:

CA certificates
authorized_keys files
Config files
Scripts

Files are distributed to all target hosts via the Nexplane Agent. Each host reports success/failure independently.

Fleet Health Check¶

Before major operations (patch campaigns, rolling restarts), run a health check across all target hosts:

Disk usage
System load
Named service status
Pending-reboot flag (Windows: Windows Update pending reboot; Linux: /var/run/reboot-required)

Hosts failing the health check are flagged before the operation begins. You can abort or proceed with only healthy hosts.

Maintenance Windows¶

Define cron-scheduled windows during which approved change requests are allowed to execute:

Approved CRs arriving outside a maintenance window are queued automatically
When the window opens, queued CRs execute in order
Emergency CRs with the ir_responder role bypass window restrictions

Configure maintenance windows in Settings → Maintenance Windows or via POST /maintenance-windows/.