Each adapter declares object-specific discovery, apply, and verification capabilities. Built-in adapters live under dns_sync/plugins/adapters/. External adapters are loaded from DNS_SYNC_PLUGIN_PATH and must be explicitly enabled through ENABLED_PLUGINS.

Plugins

The generic reconciliation engine uses adapter plugins.

Built-in adapters:

npm - Nginx Proxy Manager proxy hosts and certificates

External adapters are loaded from DNS_SYNC_PLUGIN_PATH and must also be explicitly enabled through ENABLED_PLUGINS.

Example:

ENABLED_PLUGINS=npm,verizon
DNS_SYNC_PLUGIN_PATH=/plugins

The Verizon adapter lives under external_plugins/verizon and is deliberately not copied into the dns-sync container image. The external plugin directory is mounted read-only at runtime.

Container runtime

The container periodically executes the generic reconciliation engine:

python -m dns_sync reconcile

The default container mode is read-only.

SYNC_INTERVAL=1
DNS_SYNC_APPLY=false
DNS_SYNC_STATE_DB=/var/lib/dns-sync/state.sqlite3
DNS_SYNC_PLUGIN_PATH=/plugins
ENABLED_PLUGINS=npm,verizon

SYNC_INTERVAL is expressed in minutes. The interval begins after each reconciliation completes, so runs do not overlap.

To permit planned changes to be applied and verified:

DNS_SYNC_APPLY=true

An optional mutation budget can be specified:

DNS_SYNC_MAX_CHANGES=10

When DNS_SYNC_APPLY=false, reconciliation still performs discovery, normalization, desired-state reconciliation, planning, and state persistence, but no endpoint mutations are executed.

The SQLite database should be stored on persistent storage.

External plugins should be mounted read-only at /plugins.

Main environment

NPM_ENDPOINTS=192.168.1.3,192.168.1.4,192.168.1.5
PROXY_API_PORT=81
PROXY_API_SCHEME=http
DNS_TARGET_IP=192.168.1.101

VERIZON_ENDPOINTS=192.168.1.1

DOMAIN_MATCHING=false
DOMAIN=example.com

NPM_IDENTITY=...
NPM_SECRET=...
ROUTER_PASSWORD=...

Domain filtering

Filtering is performed centrally after adapter discovery and before reconciliation, so endpoint-specific adapters get identical behavior.

DOMAIN_MATCHING=true
DOMAIN=example.com

This includes the apex and every subdomain:

example.com
app.example.com
deep.app.example.com
*.example.com

It excludes unrelated names such as:

example.net
notexample.com
example.com.evil.net

When DOMAIN_MATCHING=false, DOMAIN is ignored and all discovered records are eligible for reconciliation.

When DOMAIN_MATCHING=true, DOMAIN is required.