Advanced Usage

Operational patterns for managing deployments beyond the basics.

Deployment Locks

Prevent concurrent deployments with manual locks:

# Lock an environment (e.g., during maintenance)
dockflow lock acquire production -m "Database migration in progress"
 
# Check lock status
dockflow lock status production
 
# Release when done
dockflow lock release production

The dockflow deploy --force flag overrides locks when needed.

Selective Deployment

Deploy specific services without affecting others:

dockflow deploy production --services backend
dockflow deploy production --services frontend,worker

Skip the build phase to redeploy with existing images:

dockflow deploy production --skip-build

Preview what a deployment would do without executing:

dockflow deploy production --dry-run

Debug Workflow

When something goes wrong:

# 1. Check what's running
dockflow ps production
dockflow details production
 
# 2. Diagnose issues (shows why containers aren't starting)
dockflow diagnose production
 
# 3. Check logs
dockflow logs production app -f
dockflow logs production --since 10m
 
# 4. Shell into a container
dockflow bash production app
 
# 5. SSH to the server for manual inspection
dockflow ssh production "docker service ls"
 
# 6. Check deployment history
dockflow audit production
dockflow metrics production --history

Enable verbose Ansible output during deploy:

dockflow deploy production --debug

History Replication

Deployment history (audit log and metrics) is automatically replicated to all cluster nodes (managers and workers) after each deployment. This ensures:

• No data loss if a manager goes down
• Seamless history access after worker-to-manager promotion
• No fragmented history when manager failover occurs

History is read with automatic fallback: if the primary manager is unreachable, Dockflow tries other managers, then workers.

Manual Sync

If a node was offline during a deployment or a new node was added to the cluster, manually sync the full history:

dockflow history-sync production

This reads history from all reachable nodes, merges and deduplicates entries, then writes the complete dataset back to every node.