Skip to content
Fresh IPv4 news just dropped — 🎉 see what you’re missing

Post-Transfer Checklist: Bringing a Block Online

Reviewed by Marek Dvořák, Network engineer

The registry update makes the block yours on paper. It does not make it routable, reachable, or correctly located. Work through these steps before the first announcement; most can be done the same day the record changes.

Before you start, have:

  • Access to your registry portal (LIR Portal, ARIN Online, MyAPNIC) with rights to create objects
  • The ASN that will originate the prefix — your own or your upstream’s; see how to get an ASN
  • Name servers for reverse DNS
  • The list of upstreams and the IRR database each one uses for filters

Replace the example values before running commands. 203.0.113.0/24 is documentation space and AS64500 is a documentation ASN; querying them cannot confirm your transfer or routing. Set your transferred prefix, intended origin ASN, and the WHOIS/RDAP endpoints of the registry now responsible for the block. Follow referrals if the registry has changed. The defaults below illustrate RIPE syntax; WHOIS and RDAP explains how to identify the authoritative registry and interpret other record formats.

PREFIX='203.0.113.0/24'             # Replace with your transferred prefix
ORIGIN_ASN='AS64500'                # Replace with the intended origin ASN
WHOIS_HOST='whois.ripe.net'         # Replace with the authoritative WHOIS server
RDAP_BASE='https://rdap.db.ripe.net' # Replace with the authoritative RDAP base URL
LOOKUP_IP="${PREFIX%/*}"

Later object templates, reverse-DNS queries, and reputation samples also contain example addresses. Replace their prefixes, ASN, maintainer, host addresses, and reverse zones with your own values. The expected results describe your substituted values.

Confirm the registry record

Check the record yourself rather than relying on the registry’s email:

whois -h "$WHOIS_HOST" "$PREFIX"
curl --fail-with-body -sS -L "$RDAP_BASE/ip/$PREFIX" \
  | jq '{handle, startAddress, endAddress, entities: [.entities[]? | {handle, roles}]}'

Expected result: the range matches the transferred block and the holder identifiers match your registry account, rather than a covering parent allocation. In a RIPE record, check your organisation ID, your maintainer in mnt-by, and the agreed resource status (ALLOCATED PA for an LIR receiving allocated space). For ARIN, use whois -h whois.arin.net "n + $LOOKUP_IP" and check OrgName and the NetType appropriate to your resource; Direct Allocation is one possible type. Other registries use different fields and status labels.

If the record also changed status — PA to PI, for example — check that the new status fits how you will use the block. PI space cannot be sub-assigned to customers.

Set contacts and the abuse mailbox

Inherited admin-c and tech-c references and old remarks: should go. In the RIPE Database the abuse contact comes from abuse-c on your organisation object or on the inetnum; at ARIN it is the abuse POC on your Org.

Test that abuse mail is delivered to a monitored mailbox. Unanswered abuse reports are how a clean block acquires a reputation problem in its first month.

Clear the previous holder’s routing objects

APNIC deletes route and domain objects associated with a transferred block. Other registries may leave objects behind, and third-party IRRs such as RADB never clean up automatically.

# Route objects anywhere in the IRR system for this prefix
whois -h whois.radb.net "$PREFIX"

# ROAs currently covering the prefix, as seen by validators
curl --fail-with-body -sS "https://stat.ripe.net/data/rpki-validation/data.json?resource=$ORIGIN_ASN&prefix=$PREFIX" \
  | jq '.data.status, .data.validating_roas'

Compare each record with the approved routing plan before deleting it. A different origin can be intentional during a migration or for a mitigation service. Remove obsolete or unauthorized records; retain approved additional origins only for their agreed purpose and period. More than one origin authorization for a prefix is not automatically an error.

Coordinate changes with the administrator who controls each record and confirm what the registry transfer does to certification. Do not assume the buyer can edit the seller’s account, or that every transfer permits a shared authorization window. For third-party IRR objects, contact the listed maintainer or the database operator. The origin-change example shows how to stage and retire permissions without confusing a portal update with observed routing.

Sign ROAs for your origin

All five registries offer hosted RPKI through their portals. Create a ROA for each origin:

FieldValue
Prefix203.0.113.0/24
Origin ASNAS64500
maxLength24

Prefer exact-prefix authorizations for the routes you intend to announce. The RPKI and ROA guide explains maximum lengths, covering records and how to verify the new origin before a change. Avoid broadening authorization solely to clear a warning.

For a block intentionally kept unannounced, an AS0 ROA can express that no origin is authorized. It is not a universal reachability control; review and replace it with the intended authorization before activation.

Create route objects

Create a route object in the IRR your upstreams use. The IRR object guide covers holder authorization, source selection and mirror checks. In the RIPE Database:

route:    203.0.113.0/24
descr:    Example Networks
origin:   AS64500
mnt-by:   EXAMPLE-MNT
source:   RIPE

ARIN offers an IRR in ARIN Online, and APNIC has its own IRR. Generate a local prefix list with bgpq4:

bgpq4 -4 -l "PL-$ORIGIN_ASN" "$ORIGIN_ASN"

Your prefix should appear in the generated list. If absent, check the IRR source and query settings, then confirm acceptance with your upstream. Use the same IRR server, source selection, and query options as your upstream when comparing results.

Delegate reverse DNS

Delegate the in-addr.arpa zone through the registry: a domain object in the RIPE Database, the reverse DNS section of ARIN Online or MyAPNIC. Each /24 is one zone.

dig +short NS 113.0.203.in-addr.arpa
dig +short -x 203.0.113.10

The first command should return your name servers, the second the PTR you configured. Mail servers in particular need forward-confirmed reverse DNS before anyone accepts their mail.

Publish a geofeed

A geofeed is a CSV file, defined in RFC 8805, that tells geolocation providers where each prefix is used:

203.0.113.0/24,NL,NL-NH,Amsterdam,

Publish it over HTTPS and reference it from the registry object as described in RFC 9632: the geofeed: attribute on the inetnum in the RIPE Database, or a Geofeed remark elsewhere. A block that moved countries keeps the previous location in some databases until the feed is picked up, so publish it on day one.

Check reputation before traffic

Sample addresses across the block against major blocklists. For this Spamhaus ZEN example, replace the documentation addresses and reversed octets with addresses from your block. Use a resolver permitted to query the service:

for i in 1 64 128 254; do
  printf '\n203.0.113.%s:\n' "$i"
  if ! dig +time=3 +tries=1 +noall +comments +answer \
    "$i.113.0.203.zen.spamhaus.org" A; then
    printf 'DNS lookup failed; listing state is unknown.\n' >&2
  fi
done

Read the DNS status header and every returned A record. A zero exit status from dig alone does not establish a negative result: DNS failures such as SERVFAIL can still produce a response.

ResponseInterpretation and next action
NXDOMAIN from a working, permitted resolverNegative answer: the sampled address is not listed in this zone at query time. This does not establish the reputation of the whole block.
NOERROR with a documented ZEN listing code, such as 127.0.0.2Positive listing. Read all returned codes to identify the lists involved; PBL codes 127.0.0.10 and 127.0.0.11 describe mail-sending policy, rather than proof of abuse.
An A record in 127.255.255.0/24Spamhaus query error, not a listing or a clean result. For public mirrors, 127.255.255.254 means a public/open resolver and 127.255.255.255 means excessive queries. Fix the access or rate issue and retry.
Timeout, SERVFAIL, REFUSED, no DNS response, or an unexpected answer such as NOERROR without A recordsInconclusive. Check the resolver, query name, and service access before retrying; do not record the address as clean.

Spamhaus requires query error codes to be interpreted separately from reputation results. Investigate confirmed inherited listings and follow the relevant list’s removal procedure before sending production traffic.

Announce and verify from outside

Ask each upstream to update its filters, announce the prefix, and then check what the internet actually sees:

curl --fail-with-body -sS "https://stat.ripe.net/data/routing-status/data.json?resource=$PREFIX" \
  | jq '.data.origins, .data.visibility'

Confirm the intended origin and validation state from independent observation points, then test actual traffic and the return path. Route collectors provide partial views, and a fixed global propagation time cannot be assumed. Use the prefix-announcement runbook to separate export, upstream acceptance and forwarding checks. Prefix filters may also reject reserved and bogon ranges.

Afterwards

Record the block in IPAM with its registry ID, transfer date, and any restriction window. A block received in the RIPE region cannot be transferred again for 24 months; your records should say so. See utilization and RIR audits for keeping registry assignments in step with real use.

Set up origin monitoring. An alert for your prefix being announced by any ASN other than yours catches hijacks and forgotten objects.

If the block is spare capacity rather than production space, leasing it out is an option once it is clean and routable.

Last updated on