IRR / fastbgpq4¶

The plugin does not query Internet Routing Registries itself. It delegates to fastbgpq4, a REST service wrapping bgpq4. This page describes that contract. For the operator workflow, see IRR prefix lists.

Why a separate service¶

bgpq4 is a command-line tool, not a library. Running it behind a small HTTP service keeps bgpq4 off the NetBox hosts, adds result caching, lets large AS-SET expansions run asynchronously, and centralizes IRR queries in one place.

Deployment requirements¶

fastbgpq4 must be reachable over HTTP from the NetBox RQ worker. The web process never calls it.
The plugin sends no credentials. Put fastbgpq4 on a private network or behind a gateway that restricts access by source address.
Each IRRSource row holds the base URL of one fastbgpq4 instance. A trailing slash is ignored.

Request¶

For each address family, the client issues:

GET <irr_source.url>/api/v1/as-set/expand

Query parameter	Value
`target`	The config's source AS-SET.
`format`	`json`
`protocol`	`4` or `6`.
`sources`	The source's registry list, only when set.
`cache_ttl`	The source's cache TTL, only when set.

The address family comes from the netbox-routing prefix list being synced, so one sync of one prefix list is one expansion request.

Response¶

fastbgpq4 can answer in two ways, and the client handles both.

Immediate (HTTP 200). The prefixes are in the body:

{"data": {"nn": ["192.0.2.0/24", "198.51.100.0/24"]}}

Deferred (HTTP 202). The body carries a job_id. The client then polls:

GET <irr_source.url>/api/v1/jobs/<job_id>

until the job reports status of completed (prefixes under data) or failed (message under error), or until the poll limit below is reached.

Timeouts and retries¶

Setting	Value
HTTP timeout	30 seconds per request
Retries	Up to 3 attempts on connection errors and timeouts, with exponential backoff between 1 and 10 seconds
Poll interval	2 seconds
Poll limit	150 polls, about five minutes

These are fixed in netbox_peering_manager.irr_client and are not configurable through plugin settings.

Failure behavior¶

Any failure aborts the sync of that prefix list before its entries are touched, so a failed sync never leaves a prefix list half written.

Situation	Result
Connection refused, DNS failure, timeout	Retried, then the job fails.
HTTP 4xx or 5xx from fastbgpq4	The job fails.
Deferred job reports `failed`	The job fails with the fastbgpq4 error message.
Deferred job never completes	The job fails after the poll limit.
AS-SET expands to nothing	The job succeeds and the prefix list ends up empty.

The last row deserves attention. An AS-SET that was deleted from the registry, or a typo in its name, can legitimately expand to zero prefixes, and the prefix list is emptied accordingly. If a managed prefix list is used as an inbound filter, review job results after changing an AS-SET name.

What is written to netbox-routing¶

The returned prefixes replace the entries of the managed prefix list. The details are under netbox-routing integration.