Syncing Resources¶

The plugin imports your RIR's view of an organization into NetBox: the org itself, its points of contact, every network allocated to it, and (where applicable) the customers attached to reassigned networks. Sync is one-way (RIR -> NetBox) and idempotent: rows are upserted by handle.

Triggers¶

There are three ways to start a sync.

Manual, single config¶

From an RIRConfig detail view, click Sync. The view enqueues a SyncRIRConfigJob for the requesting user and redirects back. The job runs under the RQ worker and writes a row to Sync Logs for every operation.

Manual, bulk¶

From the RIRConfig list view, select rows and choose Sync Selected. The plugin enqueues one SyncRIRConfigJob per config. Configs without a RIRUserKey for the requesting user are skipped with a warning.

Scheduled¶

ScheduledRIRSyncJob is registered as a NetBox system job at the daily interval. On each run it:

1. Iterates RIRConfig.objects.filter(is_active=True).
2. For each config, picks the most-recently-used RIRUserKey (by checking which keys have synced-by relationships on RIROrganization, RIRContact, or RIRNetwork); falls back to the lowest-pk key for that config if no synced-by data exists.
3. Calls sync_rir_config with that key.
4. Logs failures per config and continues -- one failing config does not abort the rest.

Inspect or trigger it manually from System > Background Jobs.

What gets synced¶

sync_rir_config() orchestrates three phases:

1. Organization¶

backend.get_organization(rir_config.org_handle) is called. The result populates an RIROrganization row, upserted by handle. The org's address is normalised into an RIRAddress (deduplicated against existing addresses).

2. Contacts (POCs)¶

For every poc_link in the org's raw payload, backend.get_poc(handle) is called. Each POC becomes an RIRContact row (upserted by handle), linked back to the org. Contact addresses are also normalised into RIRAddress rows.

The contact_type is one of PERSON or ROLE (per ContactTypeChoices).

3. Networks (and customers)¶

The plugin iterates Aggregate.objects.filter(rir=rir_config.rir). For each NetBox aggregate:

1. Compute the aggregate's start/end IPv4 or IPv6 addresses.
2. Call backend.find_net(start, end) to locate the matching ARIN Net.
3. Upsert an RIRNetwork row (by handle) and link it to the aggregate.
4. If the Net has a customer_handle, fetch and upsert the matching RIRCustomer.

After the per-config job finishes, SyncRIRConfigJob enqueues one SyncPrefixesJob per (aggregate, parent_net) pair. That job walks every Prefix contained in the aggregate, calls find_net for each, and creates an RIRNetwork row whenever the result is a different handle from the parent (i.e. a real reassignment, not the parent net leaking through).

Auto-linking¶

When auto_link_networks is True (the default), a post-save signal on RIRNetwork reads the net_blocks from raw_data and tries to match them against existing NetBox Aggregate and Prefix records by exact CIDR. The first match wins and the FK is populated; existing links are not overwritten.

The signal lives in netbox_rir_manager/signals.py (auto_link_network). Disable it by setting auto_link_networks = False.

Sync logs¶

Every sync operation writes one row to RIRSyncLog:

Browse them under RIR Manager > Sync Logs, filter by config or status, or read them via /api/plugins/rir-manager/sync-logs/.

What is not synced¶

• ASN allocations. RIRBackend.get_asn exists for backend implementations but the ARIN backend currently returns None and the orchestrator does not call it.
• Backend-specific sync (backend.sync_resources) is a stub on the ARIN backend (return []); the orchestrator drives the read pipeline directly via get_organization, get_poc, and find_net.

Failure modes¶

• No API key for the requesting user: the manual sync view shows an error and redirects without enqueueing.
• backend.get_organization returns None: RIRSyncLog row with status error, object_type=organization. The job continues but skips contacts and networks for that config.
• backend.find_net returns None: no log row by default (the aggregate has no ARIN counterpart). Manual aggregate-level sync writes a skipped log row in this case for traceability.
• Transient connection errors: tenacity retries up to api_retry_count times with exponential backoff capped at api_retry_backoff * api_retry_count seconds. Permanent failures bubble up as None and are logged.

See also¶

• IPAM Integration -- the Sync buttons on Aggregate and Prefix pages.
• Reference: Background Jobs -- the full lifecycle of SyncRIRConfigJob, SyncPrefixesJob, ScheduledRIRSyncJob.