<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: IPYE</title>
    <description>The latest articles on DEV Community by IPYE (@ipye_com).</description>
    <link>https://dev.to/ipye_com</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4208865%2F8c1ffd4d-c9be-4f04-824b-dc390f062016.png</url>
      <title>DEV Community: IPYE</title>
      <link>https://dev.to/ipye_com</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/ipye_com"/>
    <language>en</language>
    <item>
      <title>Track a Free /48 in NetBox from Claim Through Announcement</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 04:43:04 +0000</pubDate>
      <link>https://dev.to/ipye_com/track-a-free-48-in-netbox-from-claim-through-announcement-2ge7</link>
      <guid>https://dev.to/ipye_com/track-a-free-48-in-netbox-from-claim-through-announcement-2ge7</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;A single /48 looks too small to need an IP address management tool. But a /48 holds 65,536 /64s, and within a year it is usually split across LANs, VPN ranges, Kubernetes clusters and service addresses at several sites. If the plan lives in someone's head or a spreadsheet, overlaps and forgotten ranges follow. NetBox is a widely used open source tool for recording this. This article tracks a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt; in NetBox from the day you claim it to the day it is announced. Examples use &lt;code&gt;2001:db8:100::/48&lt;/code&gt; and &lt;code&gt;AS64496&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Model the source of the space
&lt;/h2&gt;

&lt;p&gt;In NetBox, create:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;An &lt;strong&gt;RIR&lt;/strong&gt; entry for the registry that holds the space. Mark it as not private.&lt;/li&gt;
&lt;li&gt;An &lt;strong&gt;Aggregate&lt;/strong&gt; for &lt;code&gt;2001:db8:100::/48&lt;/code&gt;, linked to that RIR, with the date you received it. In the description, record that the space is leased and where the LOA is kept.&lt;/li&gt;
&lt;li&gt;An &lt;strong&gt;ASN&lt;/strong&gt; object for &lt;code&gt;AS64496&lt;/code&gt;, linked to the same RIR.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Aggregates are the top of the tree. NetBox then shows how much of the /48 is used by the prefixes below it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Create child prefixes before you configure anything
&lt;/h2&gt;

&lt;p&gt;Make it a rule that a range exists in NetBox before it appears in any config. For example:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Prefix&lt;/th&gt;
&lt;th&gt;Status&lt;/th&gt;
&lt;th&gt;Role&lt;/th&gt;
&lt;th&gt;Site&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100::/56&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;active&lt;/td&gt;
&lt;td&gt;infrastructure&lt;/td&gt;
&lt;td&gt;FRA1&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100:20::/64&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;active&lt;/td&gt;
&lt;td&gt;LAN&lt;/td&gt;
&lt;td&gt;Office&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100:30::/64&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;active&lt;/td&gt;
&lt;td&gt;VPN clients&lt;/td&gt;
&lt;td&gt;Office&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100:1000::/56&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;reserved&lt;/td&gt;
&lt;td&gt;Kubernetes pods&lt;/td&gt;
&lt;td&gt;FRA1&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Mark ranges as &lt;code&gt;container&lt;/code&gt; when they will be split further. NetBox's "available prefixes" view then shows free /64s inside a container, which makes the next allocation quick.&lt;/p&gt;

&lt;h2&gt;
  
  
  Use the API from the start
&lt;/h2&gt;

&lt;p&gt;Creating prefixes by API means scripts and CI can do it too:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -s -X POST https://netbox.example.net/api/ipam/prefixes/ \
  -H &lt;span class="s2"&gt;"Authorization: Token $NETBOX_TOKEN"&lt;/span&gt; \
  -H &lt;span class="s2"&gt;"Content-Type: application/json"&lt;/span&gt; \
  -d &lt;span class="s1"&gt;'{"prefix": "2001:db8:100:40::/64", "status": "active", "description": "lab LAN"}'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Or let NetBox choose the next free /64 inside a container:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -s -X POST https://netbox.example.net/api/ipam/prefixes/&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;container-id&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;/available-prefixes/ \
  -H &lt;span class="s2"&gt;"Authorization: Token $NETBOX_TOKEN"&lt;/span&gt; \
  -H &lt;span class="s2"&gt;"Content-Type: application/json"&lt;/span&gt; \
  -d &lt;span class="s1"&gt;'{"prefix_length": 64, "description": "new VLAN"}'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The pynetbox library wraps the same API for Python scripts. Keep the token in a secret store, not in shell history.&lt;/p&gt;

&lt;h2&gt;
  
  
  Record the routing facts
&lt;/h2&gt;

&lt;p&gt;Store the facts other people need during an incident:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;the BGP session details: upstream ASN, neighbour address, location&lt;/li&gt;
&lt;li&gt;where the LOA is stored and when it was issued&lt;/li&gt;
&lt;li&gt;the ROA status and maxLength&lt;/li&gt;
&lt;li&gt;the IRR as-set name&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Recent NetBox versions support custom fields on any object, so you can add fields such as &lt;code&gt;loa_issued&lt;/code&gt; and &lt;code&gt;roa_maxlength&lt;/code&gt; to the aggregate. For BGP sessions, the netbox-bgp plugin adds a proper session model.&lt;/p&gt;

&lt;h2&gt;
  
  
  Lifecycle states
&lt;/h2&gt;

&lt;p&gt;Use statuses deliberately:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;reserved&lt;/strong&gt; when you have planned a range but nothing uses it yet&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;active&lt;/strong&gt; when it is configured and in use&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;deprecated&lt;/strong&gt; when you are moving things out&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Never delete a prefix the same day you stop using it. Mark it deprecated, check for traffic, then remove it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Catch drift
&lt;/h2&gt;

&lt;p&gt;NetBox is only useful if it matches reality. A simple drift check compares addresses on your hosts and routers with what NetBox has:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -j -6 addr show scope global &lt;span class="o"&gt;|&lt;/span&gt; jq -r &lt;span class="s1"&gt;'.[].addr_info[].local'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Feed that into a script that queries &lt;code&gt;/api/ipam/ip-addresses/?address=...&lt;/code&gt; and reports any address NetBox does not know. Run it nightly and send differences to a chat channel, or fail a CI job when a config change uses a range missing from NetBox.&lt;/p&gt;

&lt;h2&gt;
  
  
  Generate config from NetBox
&lt;/h2&gt;

&lt;p&gt;Once NetBox is trusted, generate the BGP outbound prefix-list from the aggregate rather than typing it. Then an export filter cannot drift from your records. The same goes for firewall sets and DNS zones.&lt;/p&gt;

&lt;h2&gt;
  
  
  Paperwork
&lt;/h2&gt;

&lt;p&gt;When the prefix moves to a new host, download a new LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; and update the aggregate's custom fields the same day.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, create the aggregate in NetBox first, and allocate every range through it. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Your First Bilateral Peering Session Advertising a Free /48</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 04:34:50 +0000</pubDate>
      <link>https://dev.to/ipye_com/your-first-bilateral-peering-session-advertising-a-free-48-4j5h</link>
      <guid>https://dev.to/ipye_com/your-first-bilateral-peering-session-advertising-a-free-48-4j5h</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Peering is a direct BGP session with another network in which each side sends only its own routes and its customers' routes, usually without payment. It shortens paths and keeps traffic off transit links. A small network with a single free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt; can peer too, and the first session is a good way to learn the habits that keep peering safe. Examples use your ASN &lt;code&gt;64496&lt;/code&gt;, the peer &lt;code&gt;AS64500&lt;/code&gt;, and exchange addresses from &lt;code&gt;2001:db8:7::/64&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Peering versus transit
&lt;/h2&gt;

&lt;p&gt;A transit provider sends you the whole Internet and carries your routes to everyone. A peer sends only its own and customer routes, and expects the same from you. If you leak transit routes to a peer, you are offering them free transit through your network, which can overload your links and cause a route leak visible across the Internet. Your export filter is the most important part of any peering session.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where to peer
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;At an internet exchange (IX):&lt;/strong&gt; you connect a port to the exchange fabric and can peer with many networks, either bilaterally or through the route servers.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Over a private interconnect:&lt;/strong&gt; a direct cross-connect or VLAN to one network.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Exchanges usually publish rules: one MAC address per port, no proxy ARP, no IPv6 router advertisements onto the fabric, and a fixed MTU, commonly 1500. Read them before you connect.&lt;/p&gt;

&lt;h2&gt;
  
  
  Find the peer's details
&lt;/h2&gt;

&lt;p&gt;Check the peer's PeeringDB record for its policy, its exchange addresses and its expected prefix count. Also publish your own record first, as peers will look you up.&lt;/p&gt;

&lt;h2&gt;
  
  
  Session configuration in FRR
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;router bgp 64496
 neighbor 2001:db8:7::500 remote-as 64500
 neighbor 2001:db8:7::500 description peer-AS64500-IX
 neighbor 2001:db8:7::500 ttl-security hops 1
 !
 address-family ipv6 unicast
  neighbor 2001:db8:7::500 activate
  neighbor 2001:db8:7::500 prefix-list OUR-PREFIXES out
  neighbor 2001:db8:7::500 prefix-list PEER-500-IN &lt;span class="k"&gt;in&lt;/span&gt;
  neighbor 2001:db8:7::500 maximum-prefix 20 80
 exit-address-family
!
ipv6 prefix-list OUR-PREFIXES seq 10 permit 2001:db8:100::/48
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;ttl-security hops 1&lt;/code&gt; (GTSM, RFC 5082) accepts BGP packets only from a directly connected neighbour, which blocks spoofed session attacks from further away. Both sides must enable it. Set &lt;code&gt;maximum-prefix&lt;/code&gt; slightly above the count in the peer's PeeringDB record.&lt;/p&gt;

&lt;h2&gt;
  
  
  Build the inbound filter from IRR
&lt;/h2&gt;

&lt;p&gt;Accept only what the peer is entitled to announce:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;bgpq4 -6 -l PEER-500-IN AS-PEER500EXAMPLE &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; peer500.txt
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Use the as-set listed in the peer's PeeringDB record, load the output into FRR, and regenerate it regularly. Add rules to reject bogons, your own prefix, and anything longer than /48. Also drop routes that are RPKI-invalid if your router uses RPKI.&lt;/p&gt;

&lt;h2&gt;
  
  
  The request email
&lt;/h2&gt;

&lt;p&gt;Keep it short:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Subject: IPv6 peering request AS64496 / AS64500 at &amp;lt;exchange name&amp;gt;

We would like to peer with you at &amp;lt;exchange name&amp;gt;.
Our ASN: 64496
Our IPv6 address on the exchange: 2001:db8:7::496
Prefixes: 1 (2001:db8:100::/48), as-set AS64496:AS-ALL
PeeringDB: &amp;lt;link to your record&amp;gt;
NOC: &amp;lt;your NOC address&amp;gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Bring the session up
&lt;/h2&gt;

&lt;p&gt;Once both sides configure the session:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast summary'&lt;/span&gt;
vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast neighbors 2001:db8:7::500 advertised-routes'&lt;/span&gt;
vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast neighbors 2001:db8:7::500 routes'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Confirm you send exactly one prefix and receive about the number the peer published. Then run &lt;code&gt;mtr -6&lt;/code&gt; to an address in the peer's space and check that traffic now takes the exchange path.&lt;/p&gt;

&lt;h2&gt;
  
  
  Keep it healthy
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Regenerate the inbound filter from IRR on a schedule.&lt;/li&gt;
&lt;li&gt;Watch session state and prefix count, and alert on big changes.&lt;/li&gt;
&lt;li&gt;Update your PeeringDB record when anything changes.&lt;/li&gt;
&lt;li&gt;Answer the peer's NOC promptly. Peering is a relationship as well as a session.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Paperwork
&lt;/h2&gt;

&lt;p&gt;Peers rarely ask for an LOA, but some do. Keep the one from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; handy, along with your validator output.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, publish your PeeringDB record, and send your first peering request to a network that lists an open policy. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>My Free /48 Is Filtered: A Triage Checklist That Starts With ROA and IRR</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 04:28:21 +0000</pubDate>
      <link>https://dev.to/ipye_com/my-free-48-is-filtered-a-triage-checklist-that-starts-with-roa-and-irr-185g</link>
      <guid>https://dev.to/ipye_com/my-free-48-is-filtered-a-triage-checklist-that-starts-with-roa-and-irr-185g</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;You announced your prefix, the BGP session is up, and parts of the Internet still cannot reach you. The cause is nearly always one of a short list, and the fastest way through is to check them in order, collecting evidence as you go. This checklist is written for a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;, with &lt;code&gt;2001:db8:100::/48&lt;/code&gt;, origin &lt;code&gt;AS64496&lt;/code&gt; and upstream &lt;code&gt;AS64497&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1: confirm what you send
&lt;/h2&gt;

&lt;p&gt;Start on your own router:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast neighbors 2001:db8:0:1::1 advertised-routes'&lt;/span&gt;
birdc show route &lt;span class="kd"&gt;export&lt;/span&gt; upstream1
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Use whichever matches your daemon. You should see exactly &lt;code&gt;2001:db8:100::/48&lt;/code&gt;. Common mistakes found here:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;nothing is advertised, because the aggregate is missing from the routing table (add a blackhole route for it)&lt;/li&gt;
&lt;li&gt;a /56 or /64 is advertised instead of the /48&lt;/li&gt;
&lt;li&gt;a typo in the prefix&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If the advertisement is wrong, fix it and stop here.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2: check RPKI
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;routinator validate --asn 64496 --prefix 2001:db8:100::/48
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Or use the RIPEstat RPKI validation widget. Results:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Valid:&lt;/strong&gt; move on.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Invalid, wrong origin:&lt;/strong&gt; the ROA authorises a different ASN. For the free /48 on IPYE's ARIN or APNIC accounts, the ROA follows the ASN set in the dashboard; check that it matches the ASN on your session.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Invalid, length:&lt;/strong&gt; you announce something longer than the ROA's maxLength. Announce the /48 instead.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Not found:&lt;/strong&gt; no ROA covers the prefix. Fewer networks drop not-found routes than invalid ones, but some do.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Networks that perform route origin validation drop invalid routes, which shows up as partial reachability: some networks reach you, others do not.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3: check the IRR
&lt;/h2&gt;

&lt;p&gt;Many upstreams build customer filters from IRR data:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;whois -h whois.radb.net -- &lt;span class="s1"&gt;'-i origin AS64496'&lt;/span&gt;
bgpq4 -6 AS64496
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If the route6 object is missing or names another origin, the upstream filter may drop your route before it ever leaves their network. The IRRexplorer tool from NLNOG shows IRR and RPKI status together and is a fast cross-check.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4: see what the world sees
&lt;/h2&gt;

&lt;p&gt;Look the prefix up in RIPEstat's routing status and in route collector data from RIPE RIS or RouteViews. Questions to answer:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;How many collector peers see the route? Very few suggests the upstream is not exporting it widely.&lt;/li&gt;
&lt;li&gt;Is the AS path what you expect, ending &lt;code&gt;64497 64496&lt;/code&gt;?&lt;/li&gt;
&lt;li&gt;Does any path show a different origin? That could be a hijack or a stale announcement.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Public looking glasses run by transit networks show whether specific large networks carry your route. If your direct upstream does not show it in their looking glass, the problem is at the first hop.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 5: check the data plane
&lt;/h2&gt;

&lt;p&gt;Control plane fine but traffic failing points to forwarding or filtering:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;mtr -6 -rwc 20 2001:db8:100::1
tcpdump -ni eth0 icmp6 or host 2001:db8:100::1
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Run &lt;code&gt;mtr&lt;/code&gt; from an outside host. If traffic reaches your router and replies never arrive, look at your firewall, reverse-path filtering, and the route back: a host with two uplinks may answer through the wrong one, where the source address is filtered.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 6: escalate with an evidence packet
&lt;/h2&gt;

&lt;p&gt;When the evidence points to the upstream or a remote network, send one complete message:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;Prefix: 2001:db8:100::/48   Origin: AS64496
Advertised to you: yes (output attached)
RPKI: valid (validator output attached)
IRR: route6 present, origin AS64496
Collectors: seen by N peers, missing via your network
Test: mtr from &amp;lt;source&amp;gt; attached
LOA: attached
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Fill in the real values from your checks. Attach the LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt;; network teams check it before changing filters.&lt;/p&gt;

&lt;h2&gt;
  
  
  Common causes worth rechecking
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Upstream export filter not yet updated for a new prefix.&lt;/li&gt;
&lt;li&gt;Announcing a longer prefix than the ROA allows.&lt;/li&gt;
&lt;li&gt;ASN changed but the announcement moved before validators picked up the new ROA.&lt;/li&gt;
&lt;li&gt;Firewall dropping ICMPv6 or inbound traffic.&lt;/li&gt;
&lt;li&gt;A stale route6 object with an old origin.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;Save this list in your runbook and run steps 1 to 3 after every change. If you are setting up from scratch, &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;claim a free /48 from IPYE&lt;/a&gt; and run the checklist before telling anyone the prefix is live. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Build a SLAAC LAN from One /64 of Your Free /48</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 04:00:34 +0000</pubDate>
      <link>https://dev.to/ipye_com/build-a-slaac-lan-from-one-64-of-your-free-48-4ho8</link>
      <guid>https://dev.to/ipye_com/build-a-slaac-lan-from-one-64-of-your-free-48-4ho8</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;SLAAC (Stateless Address Autoconfiguration) lets IPv6 hosts configure their own addresses from router advertisements, with no DHCP server. It is the simplest way to number a LAN, and the only method every major client operating system supports. This article builds a SLAAC LAN from one /64 of a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;. The edge announces &lt;code&gt;2001:db8:100::/48&lt;/code&gt; and the LAN uses &lt;code&gt;2001:db8:100:20::/64&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why exactly a /64
&lt;/h2&gt;

&lt;p&gt;SLAAC requires a /64 prefix. Hosts generate the last 64 bits of their address themselves, so a /56 or /80 on the LAN simply does not work with SLAAC. A /48 holds 65,536 /64s, so there is no reason to economise.&lt;/p&gt;

&lt;h2&gt;
  
  
  Router configuration with radvd
&lt;/h2&gt;

&lt;p&gt;On the LAN router, give the interface an address and advertise the prefix:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -6 addr add 2001:db8:100:20::1/64 dev lan0
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;





&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight conf"&gt;&lt;code&gt;&lt;span class="k"&gt;interface&lt;/span&gt; &lt;span class="nf"&gt;lan0&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="nc"&gt;AdvSendAdvert&lt;/span&gt; &lt;span class="n"&gt;on&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="nc"&gt;AdvLinkMTU&lt;/span&gt; &lt;span class="mi"&gt;1500&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="n"&gt;prefix&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="o"&gt;::/&lt;/span&gt;&lt;span class="mi"&gt;64&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
        &lt;span class="nc"&gt;AdvOnLink&lt;/span&gt; &lt;span class="n"&gt;on&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
        &lt;span class="nc"&gt;AdvAutonomous&lt;/span&gt; &lt;span class="n"&gt;on&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="no"&gt;RDNSS&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="no"&gt;DNSSL&lt;/span&gt; &lt;span class="n"&gt;lan&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;example&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;net&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;AdvAutonomous on&lt;/code&gt; is the flag that tells hosts to generate addresses. &lt;code&gt;RDNSS&lt;/code&gt; hands out the DNS server in the RA itself, which Android requires because it does not support DHCPv6 for addresses or DNS. Enable forwarding on the router with &lt;code&gt;sysctl -w net.ipv6.conf.all.forwarding=1&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;The same thing in systemd-networkd:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight ini"&gt;&lt;code&gt;&lt;span class="k"&gt;[Network]&lt;/span&gt;
&lt;span class="na"&gt;Address&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="o"&gt;/&lt;/span&gt;&lt;span class="mi"&gt;64&lt;/span&gt;
&lt;span class="na"&gt;IPv6SendRA&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;yes&lt;/span&gt;

&lt;span class="k"&gt;[IPv6SendRA]&lt;/span&gt;
&lt;span class="na"&gt;DNS&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;
&lt;span class="na"&gt;Domains&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;lan&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;example&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;net&lt;/span&gt;

&lt;span class="k"&gt;[IPv6Prefix]&lt;/span&gt;
&lt;span class="na"&gt;Prefix&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="o"&gt;::/&lt;/span&gt;&lt;span class="mi"&gt;64&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  What hosts do with it
&lt;/h2&gt;

&lt;p&gt;A typical client ends up with several addresses:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;a link-local &lt;code&gt;fe80::&lt;/code&gt; address&lt;/li&gt;
&lt;li&gt;a stable address for incoming connections, usually generated with RFC 7217 so it does not reveal the MAC address&lt;/li&gt;
&lt;li&gt;one or more temporary privacy addresses (RFC 8981) used for outgoing connections and rotated regularly&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;On Linux, check with &lt;code&gt;ip -6 addr show dev eth0&lt;/code&gt;. Addresses marked &lt;code&gt;temporary&lt;/code&gt; are privacy addresses. The stable address mode is set by &lt;code&gt;net.ipv6.conf.eth0.addr_gen_mode&lt;/code&gt;, where &lt;code&gt;2&lt;/code&gt; or &lt;code&gt;3&lt;/code&gt; means stable-privacy.&lt;/p&gt;

&lt;h2&gt;
  
  
  Servers on a SLAAC LAN
&lt;/h2&gt;

&lt;p&gt;Servers need predictable addresses for DNS. Options:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Assign a static address on the server, for example &lt;code&gt;2001:db8:100:20::53&lt;/code&gt;, alongside SLAAC.&lt;/li&gt;
&lt;li&gt;Use a stable-privacy address and publish it in DNS. It stays the same as long as the prefix and interface do not change.&lt;/li&gt;
&lt;li&gt;Use a token: &lt;code&gt;ip token set ::53 dev eth0&lt;/code&gt; makes the host combine the advertised prefix with a fixed interface identifier.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Turn off temporary addresses on servers (&lt;code&gt;net.ipv6.conf.eth0.use_tempaddr=0&lt;/code&gt;) so outbound connections come from the address you expect.&lt;/p&gt;

&lt;h2&gt;
  
  
  Adding DHCPv6 later
&lt;/h2&gt;

&lt;p&gt;SLAAC and DHCPv6 can coexist. Setting the O flag (&lt;code&gt;AdvOtherConfigFlag on;&lt;/code&gt; in radvd) tells hosts to fetch extra settings such as DNS from DHCPv6 while still building addresses with SLAAC. Setting the M flag asks for addresses from DHCPv6 as well, which Android ignores. Keep &lt;code&gt;AdvAutonomous on&lt;/code&gt; if any Android devices use the LAN, or they will end up without a global address.&lt;/p&gt;

&lt;h2&gt;
  
  
  Firewall the LAN
&lt;/h2&gt;

&lt;p&gt;Every host now has a global address. The router must default-deny new inbound connections to the LAN:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;chain forward {
  type filter hook forward priority 0; policy drop;
  ct state established,related accept
  meta l4proto ipv6-icmp accept
  iifname "lan0" accept
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Add explicit accept rules for anything you want to publish, and keep ICMPv6 open so path MTU discovery works.&lt;/p&gt;

&lt;h2&gt;
  
  
  Protect the RAs
&lt;/h2&gt;

&lt;p&gt;Any host on the LAN can send router advertisements. A misconfigured laptop running connection sharing can become everyone's default router. Managed switches offer RA Guard, which drops RAs from ports that are not router ports. On unmanaged networks, watch for rogue RAs with &lt;code&gt;rdisc6 lan0&lt;/code&gt; or &lt;code&gt;radvdump&lt;/code&gt;, which show every router advertising on the segment.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify
&lt;/h2&gt;

&lt;p&gt;From a client:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;rdisc6 eth0
ip -6 route show default
ping -6 -c 3 2001:db8:100::1
curl -6 https://ifconfig.co
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The last command should return an address inside &lt;code&gt;2001:db8:100:20::/64&lt;/code&gt;, which shows the edge is routing the LAN through your announced prefix.&lt;/p&gt;

&lt;h2&gt;
  
  
  Paperwork
&lt;/h2&gt;

&lt;p&gt;Only the /48 is announced, so the LAN needs no extra ROA. If the edge moves to a new host, download a fresh LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt;; the LAN addresses stay the same.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, pick one /64 for the LAN, and check with &lt;code&gt;rdisc6&lt;/code&gt; that only your router is advertising. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Blackhole and Static Route Hygiene for a Free /48 You Originate</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 03:55:24 +0000</pubDate>
      <link>https://dev.to/ipye_com/blackhole-and-static-route-hygiene-for-a-free-48-you-originate-183j</link>
      <guid>https://dev.to/ipye_com/blackhole-and-static-route-hygiene-for-a-free-48-you-originate-183j</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Blackhole routes and static routes are small, sharp tools. On an edge that originates a prefix, a correct blackhole keeps unused space from looping and protects you during attacks. A wrong one can withdraw your service, or announce routes you never meant to send. This article covers safe patterns for an edge announcing a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;, with &lt;code&gt;2001:db8:100::/48&lt;/code&gt; and &lt;code&gt;AS64496&lt;/code&gt; as examples.&lt;/p&gt;

&lt;h2&gt;
  
  
  Three kinds of discard route on Linux
&lt;/h2&gt;

&lt;p&gt;Linux has three route types that drop traffic:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -6 route add blackhole   2001:db8:100:99::/64   &lt;span class="c"&gt;# drop silently&lt;/span&gt;
ip -6 route add unreachable 2001:db8:100:98::/64   &lt;span class="c"&gt;# drop, send ICMPv6 unreachable&lt;/span&gt;
ip -6 route add prohibit    2001:db8:100:97::/64   &lt;span class="c"&gt;# drop, send administratively prohibited&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Use &lt;code&gt;blackhole&lt;/code&gt; for attack traffic. The other two generate ICMPv6 errors for every dropped packet, which adds load during a flood. They are better for unused space inside your own network, where a quick error helps people debugging.&lt;/p&gt;

&lt;h2&gt;
  
  
  The aggregate anchor is a good blackhole
&lt;/h2&gt;

&lt;p&gt;A blackhole for the whole /48 is the standard way to originate it:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -6 route add blackhole 2001:db8:100::/48
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The BGP daemon needs a route to advertise, and the blackhole provides one. Any more-specific route, such as a connected /64 or a WireGuard peer's range, wins by longest match, so real traffic is delivered. Packets for unused parts of the /48 are dropped at your edge instead of being sent back to the upstream's default route and looping until their hop limit runs out.&lt;/p&gt;

&lt;p&gt;The anchor is safe as long as your outbound policy allows only the /48 itself.&lt;/p&gt;

&lt;h2&gt;
  
  
  The dangerous one: redistributing statics
&lt;/h2&gt;

&lt;p&gt;The most common blackhole accident is &lt;code&gt;redistribute static&lt;/code&gt; on the BGP process. Every static route then becomes a candidate for export, including /64 blackholes you added during an attack and any test route someone left behind. Avoid redistribution and use &lt;code&gt;network&lt;/code&gt; statements, or filter it explicitly:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;ipv6 prefix-list OUT-V6 seq 10 permit 2001:db8:100::/48
!
route-map REDIST-STATIC permit 10
 match ipv6 address prefix-list OUT-V6
!
router bgp 64496
 address-family ipv6 unicast
  redistribute static route-map REDIST-STATIC
  neighbor 2001:db8:0:1::1 prefix-list OUT-V6 out
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Two layers of filtering, at redistribution and at the neighbour, mean a single mistake does not leak.&lt;/p&gt;

&lt;p&gt;Check what you actually send:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast neighbors 2001:db8:0:1::1 advertised-routes'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;It should list exactly one prefix.&lt;/p&gt;

&lt;h2&gt;
  
  
  Local discard versus remote blackholing
&lt;/h2&gt;

&lt;p&gt;A local blackhole stops traffic at your router. During a volumetric attack that is too late, because the traffic has already filled the link from your upstream. Remote-triggered blackholing (RTBH) asks the upstream to drop it before it reaches you, usually by sending a /128 with the upstream's blackhole community. That sacrifices the attacked address to save everything else. Ask your upstream whether they support it and how, and test it once on an unused /128 outside an emergency.&lt;/p&gt;

&lt;h2&gt;
  
  
  Label every manual route
&lt;/h2&gt;

&lt;p&gt;Plain &lt;code&gt;ip route&lt;/code&gt; entries carry no description. Use FRR's static routes, which can carry a tag, and keep a comment in the configuration:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;! ticket 4711, attack on 2001:db8:100:99::/64, remove after review
ipv6 route 2001:db8:100:99::/64 blackhole tag 4711
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;A route tag of the ticket number makes it easy to list all emergency routes with &lt;code&gt;show ipv6 route tag 4711&lt;/code&gt; and to find their reason later.&lt;/p&gt;

&lt;h2&gt;
  
  
  Make removal part of adding
&lt;/h2&gt;

&lt;p&gt;Emergency blackholes tend to stay forever. A small timer helps:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;systemd-run --on-active=6h --unit=bh-4711 \
  vtysh -c &lt;span class="s1"&gt;'conf t'&lt;/span&gt; -c &lt;span class="s1"&gt;'no ipv6 route 2001:db8:100:99::/64 blackhole tag 4711'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If the attack is still going after six hours, re-add it deliberately. That forces a human decision instead of a silent permanent outage for whatever later lives in that /64.&lt;/p&gt;

&lt;h2&gt;
  
  
  Watch out for uRPF interactions
&lt;/h2&gt;

&lt;p&gt;If you use strict reverse-path filtering, a blackhole for a source range also drops traffic from that range, which is the basis of source-based RTBH. That is useful against an attack from a known source, and surprising when you did not intend it. Document whether your edge uses strict or loose mode.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify after every change
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;code&gt;advertised-routes&lt;/code&gt; still shows only the /48.&lt;/li&gt;
&lt;li&gt;A public route collector still shows the /48 with origin &lt;code&gt;AS64496&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;The blackholed range is silent, and everything else in the /48 still answers from an external host.&lt;/li&gt;
&lt;/ol&gt;

&lt;h2&gt;
  
  
  Paperwork
&lt;/h2&gt;

&lt;p&gt;None of this changes your ROA or LOA. Keep the LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; and your upstream's RTBH instructions in the same runbook.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;Audit your static routes today with &lt;code&gt;ip -6 route show type blackhole&lt;/code&gt;, and &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;claim a free /48 from IPYE&lt;/a&gt; to practise these patterns on a lab edge. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Provider-Assigned IPv6 vs Announcing Your Own Free /48: When to Switch</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 03:49:27 +0000</pubDate>
      <link>https://dev.to/ipye_com/provider-assigned-ipv6-vs-announcing-your-own-free-48-when-to-switch-39l3</link>
      <guid>https://dev.to/ipye_com/provider-assigned-ipv6-vs-announcing-your-own-free-48-when-to-switch-39l3</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Every VPS comes with provider-assigned (PA) IPv6, usually a /64 and sometimes a /56. It costs nothing extra and needs no BGP. For many workloads it is the right choice. But PA addresses belong to the provider, and they stay behind when you leave. This article looks at when it makes sense to switch from PA addresses to announcing your own prefix, such as a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;. Examples use PA &lt;code&gt;2001:db8:aaaa:1::/64&lt;/code&gt; and your own &lt;code&gt;2001:db8:100::/48&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  When PA is fine
&lt;/h2&gt;

&lt;p&gt;Stay on PA addresses if most of these are true:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;You have one server, or a few, at one provider.&lt;/li&gt;
&lt;li&gt;Clients reach you by DNS name, not by address.&lt;/li&gt;
&lt;li&gt;Nobody allowlists your addresses in their firewall.&lt;/li&gt;
&lt;li&gt;You can rebuild at another provider in an afternoon and update DNS.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;In that situation, your own prefix adds a BGP session to maintain and little benefit.&lt;/p&gt;

&lt;h2&gt;
  
  
  Signs that PA has started to hurt
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Partners allowlist your addresses.&lt;/strong&gt; Each change means emails and waiting.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Addresses are baked into configs&lt;/strong&gt; you do not control: client VPN profiles, IoT devices, SPF records at customers.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You run services across providers&lt;/strong&gt; and want them in one consistent range.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You have moved providers before&lt;/strong&gt; and the renumbering was painful.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;You need more than one /64&lt;/strong&gt; and the provider will not give you a larger block.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Each of these costs a little every time an address changes. Owning the addresses turns that recurring cost into a one-time setup.&lt;/p&gt;

&lt;h2&gt;
  
  
  What you take on
&lt;/h2&gt;

&lt;p&gt;Announcing your own prefix requires:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;an ASN, or a provider that announces the prefix for you&lt;/li&gt;
&lt;li&gt;a VPS or server with BGP support&lt;/li&gt;
&lt;li&gt;an LOA, a ROA and an IRR object&lt;/li&gt;
&lt;li&gt;monitoring for visibility and hijacks&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;With the free /48, setup is instant, the LOA comes from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt;, and the ROA and route objects are created automatically. The ASN and the BGP-capable host remain your part.&lt;/p&gt;

&lt;h2&gt;
  
  
  Run both side by side
&lt;/h2&gt;

&lt;p&gt;You do not have to switch in one step. A Linux host can hold PA and own addresses at the same time:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -6 addr add 2001:db8:aaaa:1::10/64 dev eth0      &lt;span class="c"&gt;# PA, from the provider&lt;/span&gt;
ip -6 addr add 2001:db8:100::10/128 dev lo          &lt;span class="c"&gt;# own prefix&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Keep SSH and management on the PA address, so a BGP problem never locks you out. Move public services to the own prefix one at a time by adding AAAA records for the new address, then removing the old ones.&lt;/p&gt;

&lt;h2&gt;
  
  
  Source address selection
&lt;/h2&gt;

&lt;p&gt;With two global addresses, Linux picks the source for outbound connections using RFC 6724 rules, which usually means the address on the outgoing interface. Partners who allowlist your new addresses will see the PA source unless you set it explicitly:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip -6 route replace default via fe80::1 dev eth0 src 2001:db8:100::10
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Applications can also bind to a source address directly. Test with &lt;code&gt;curl -6 https://ifconfig.co&lt;/code&gt; from the host, which shows the address the outside world sees.&lt;/p&gt;

&lt;h2&gt;
  
  
  A worked flip
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Bring up BGP and announce the /48 with no services on it.&lt;/li&gt;
&lt;li&gt;Confirm visibility in public route collectors and run an external ping to &lt;code&gt;2001:db8:100::10&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Add AAAA records for the new address alongside the PA ones.&lt;/li&gt;
&lt;li&gt;Ask partners to add the new address to their allowlists.&lt;/li&gt;
&lt;li&gt;Switch the outbound source.&lt;/li&gt;
&lt;li&gt;After a quiet period, remove the PA AAAA records.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Keep the PA addresses configured as a fallback for some weeks. If the BGP session has trouble, you can put the PA AAAA records back.&lt;/p&gt;

&lt;h2&gt;
  
  
  Moving later
&lt;/h2&gt;

&lt;p&gt;Once services live in your own range, changing provider becomes a BGP change rather than a renumbering project, and ASN and region changes in the IPYE dashboard apply automatically.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;List every place your current PA addresses appear. If that list is long, &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;claim a free /48 from IPYE&lt;/a&gt; and run both side by side. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Fill Out PeeringDB When Your First Prefix Is a Free /48</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 03:42:58 +0000</pubDate>
      <link>https://dev.to/ipye_com/fill-out-peeringdb-when-your-first-prefix-is-a-free-48-51nl</link>
      <guid>https://dev.to/ipye_com/fill-out-peeringdb-when-your-first-prefix-is-a-free-48-51nl</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;PeeringDB is the shared database where networks publish how to peer with them: which exchanges and data centres they are in, what their peering policy is, and how many prefixes to expect. Many operators set max-prefix limits and decide whether to peer at all by reading it. If your network starts with a single free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;, a small, accurate PeeringDB record makes the first peering requests much easier. Examples use &lt;code&gt;AS64496&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why bother with one prefix
&lt;/h2&gt;

&lt;p&gt;A record helps even before you peer with anyone:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Peers and route servers read your expected prefix count to set max-prefix limits. Without it, they guess.&lt;/li&gt;
&lt;li&gt;Your NOC contact is where other operators look when your network causes a problem.&lt;/li&gt;
&lt;li&gt;Some exchange points and route servers build filters from the IRR as-set listed in PeeringDB.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Get access
&lt;/h2&gt;

&lt;p&gt;Create a user account on peeringdb.com, then request affiliation with your ASN. PeeringDB verifies the request against the contact information in the RIR's record for the ASN, usually by email to the registered contact. Make sure that address works before you start. If no organisation exists yet, the affiliation request creates one.&lt;/p&gt;

&lt;h2&gt;
  
  
  The minimum useful record
&lt;/h2&gt;

&lt;p&gt;Fill in these fields carefully:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Field&lt;/th&gt;
&lt;th&gt;What to enter&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Name&lt;/td&gt;
&lt;td&gt;your network's public name&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Website&lt;/td&gt;
&lt;td&gt;your main site&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IRR as-set&lt;/td&gt;
&lt;td&gt;your as-set, for example &lt;code&gt;AS64496:AS-CUSTOMERS&lt;/code&gt;, or &lt;code&gt;AS64496&lt;/code&gt; if you have no customers&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IPv6 prefixes&lt;/td&gt;
&lt;td&gt;the number you announce, here &lt;code&gt;1&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;IPv4 prefixes&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;0&lt;/code&gt; until you have IPv4 space&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Network type&lt;/td&gt;
&lt;td&gt;the honest category, such as Content or NSP&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Peering policy&lt;/td&gt;
&lt;td&gt;Open, Selective or Restrictive&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Policy URL&lt;/td&gt;
&lt;td&gt;a page with your peering terms, if you have one&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Then add at least one role contact for NOC and one for policy, with addresses that a person reads.&lt;/p&gt;

&lt;p&gt;Set the prefix counts slightly above what you announce today if you expect growth, but not far above. If you set &lt;code&gt;1000&lt;/code&gt; with one prefix, peers lose the protection max-prefix is meant to give.&lt;/p&gt;

&lt;h2&gt;
  
  
  Facilities and exchanges
&lt;/h2&gt;

&lt;p&gt;Add a facility only where you have equipment or a cross-connect. Add an exchange connection only once you have a port and IP addresses from that exchange. A record that claims presence you do not have wastes other operators' time and damages trust.&lt;/p&gt;

&lt;h2&gt;
  
  
  Traffic levels
&lt;/h2&gt;

&lt;p&gt;The traffic field is a range, and it should reflect real measurements. A new network with a single /48 normally sits in the lowest band. That is fine; plenty of small networks peer actively.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verify the record
&lt;/h2&gt;

&lt;p&gt;The API is open for reading:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -s &lt;span class="s1"&gt;'https://www.peeringdb.com/api/net?asn=64496'&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; jq &lt;span class="s1"&gt;'.data[0] | {name, info_prefixes6, irr_as_set, policy_general}'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Check the IRR as-set too:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;bgpq4 -6 -l OUR-PREFIXES AS64496
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The output should list your /48. If it is empty, the route6 object is missing or the as-set does not include your ASN. For a free /48 on IPYE's ARIN or APNIC accounts, route objects are created automatically; confirm they resolve from the as-set you published.&lt;/p&gt;

&lt;h2&gt;
  
  
  Keep it current
&lt;/h2&gt;

&lt;p&gt;Review the record whenever something changes: a new prefix, a new exchange, a new NOC address. A quick check every few months is enough for a small network. Stale contact addresses are the most common problem, and they matter most during incidents.&lt;/p&gt;

&lt;h2&gt;
  
  
  Your first peering request
&lt;/h2&gt;

&lt;p&gt;When you contact a network about peering, include your ASN, the exchange or facility, your IPv6 address on the exchange, your max-prefix value and a link to your PeeringDB record. Keep it short. If they ask for proof of the prefix, attach the LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, check your as-set with &lt;code&gt;bgpq4&lt;/code&gt;, and publish a PeeringDB record that lists exactly what you announce. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>nftables Hardening on Day One for a Fresh Free /48 Edge</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 01:45:53 +0000</pubDate>
      <link>https://dev.to/ipye_com/nftables-hardening-on-day-one-for-a-fresh-free-48-edge-285i</link>
      <guid>https://dev.to/ipye_com/nftables-hardening-on-day-one-for-a-fresh-free-48-edge-285i</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The moment a new /48 is announced, scanners start probing it. IPv6 space is too large to sweep completely, but hosts with predictable addresses such as &lt;code&gt;::1&lt;/code&gt; or addresses published in DNS are found quickly. If you are bringing up an edge for a free /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;, write the firewall before the BGP session, not after. This article builds a compact nftables ruleset for a Linux edge that originates &lt;code&gt;2001:db8:100::/48&lt;/code&gt; and forwards it to a few servers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Start from default drop
&lt;/h2&gt;

&lt;p&gt;Use one &lt;code&gt;inet&lt;/code&gt; table so IPv4 and IPv6 share the same rules where that makes sense:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;table inet edge {
  set mgmt6 { type ipv6_addr; flags interval; elements = { 2001:db8:ffff::/64 } }
  set web6  { type ipv6_addr; elements = { 2001:db8:100:10::80 } }

  chain input {
    type filter hook input priority 0; policy drop;
    ct state established,related accept
    ct state invalid drop
    iif lo accept
    meta l4proto ipv6-icmp jump icmp6_in
    tcp dport 179 ip6 saddr 2001:db8:0:1::1 accept
    tcp dport 22 ip6 saddr @mgmt6 accept
  }

  chain forward {
    type filter hook forward priority 0; policy drop;
    ct state established,related accept
    ct state invalid drop
    meta l4proto ipv6-icmp jump icmp6_fwd
    ip6 daddr @web6 tcp dport { 80, 443 } accept
  }

  chain output { type filter hook output priority 0; policy accept; }
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Two details matter. BGP on port 179 is accepted only from the upstream's neighbour address. SSH is accepted only from a management set, which you can update without touching the rule itself with &lt;code&gt;nft add element inet edge mgmt6 { 2001:db8:ffff:1::/64 }&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  ICMPv6 is not optional
&lt;/h2&gt;

&lt;p&gt;IPv6 depends on ICMPv6 for neighbour discovery and path MTU discovery. Blocking it all breaks the network in slow, confusing ways: connections hang when a large packet needs to be fragmented, and neighbours stop resolving. RFC 4890 lists what to allow. A practical version:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;chain icmp6_in {
  icmpv6 type { destination-unreachable, packet-too-big, time-exceeded, parameter-problem } accept
  icmpv6 type { nd-neighbor-solicit, nd-neighbor-advert, nd-router-advert } ip6 hoplimit 255 accept
  icmpv6 type echo-request limit rate 10/second accept
  drop
}
chain icmp6_fwd {
  icmpv6 type { destination-unreachable, packet-too-big, time-exceeded, parameter-problem } accept
  icmpv6 type echo-request limit rate 50/second accept
  drop
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The hop limit check of 255 on neighbour discovery is a cheap guard: legitimate ND packets are sent with hop limit 255, so anything lower has been routed from elsewhere and should not be trusted. Neighbour discovery is not forwarded, so it does not appear in the forward chain.&lt;/p&gt;

&lt;h2&gt;
  
  
  Anti-spoofing at the edge
&lt;/h2&gt;

&lt;p&gt;Your edge should never receive packets from the Internet that claim a source inside your own /48, and it should never send packets with a source outside it. Add these at the top of the forward chain:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;iifname "eth0" ip6 saddr 2001:db8:100::/48 drop
oifname "eth0" ip6 saddr != 2001:db8:100::/48 drop
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Adjust the interface name to your WAN. The second rule is BCP 38 for your own network, and it stops a compromised server from spoofing traffic out of your edge.&lt;/p&gt;

&lt;h2&gt;
  
  
  Load it safely
&lt;/h2&gt;

&lt;p&gt;Never apply a new ruleset over SSH without a way back. A simple pattern:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;nft -c -f /etc/nftables.conf        &lt;span class="c"&gt;# syntax check only&lt;/span&gt;
cp /etc/nftables.conf /root/nft.good
nft -f /etc/nftables.new &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; sleep 60 &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; nft -f /root/nft.good
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;If the new rules keep your SSH session alive, interrupt the command with Ctrl-C before the timer fires and keep them. If they lock you out, the old ruleset returns after sixty seconds. Start every file with &lt;code&gt;flush ruleset&lt;/code&gt; so a reload replaces everything instead of appending duplicates.&lt;/p&gt;

&lt;h2&gt;
  
  
  Watch what you drop
&lt;/h2&gt;

&lt;p&gt;Add a counter and a rate-limited log before the default drop while you settle in:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight vcl"&gt;&lt;code&gt;counter log prefix &lt;span class="s2"&gt;"edge-in-drop "&lt;/span&gt; limit rate 5/second
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then use &lt;code&gt;nft list chain inet edge input&lt;/code&gt; to read counters and &lt;code&gt;journalctl -k | grep edge-in-drop&lt;/code&gt; to see what is being dropped. After a week, remove the log rule or lower the rate; the counters are enough.&lt;/p&gt;

&lt;h2&gt;
  
  
  Connection tracking limits
&lt;/h2&gt;

&lt;p&gt;A busy proxy or DNS server can fill the conntrack table. Check usage with &lt;code&gt;conntrack -C&lt;/code&gt; against &lt;code&gt;sysctl net.netfilter.nf_conntrack_max&lt;/code&gt;. For high-rate UDP services, consider &lt;code&gt;notrack&lt;/code&gt; rules in a &lt;code&gt;raw&lt;/code&gt; priority chain for that traffic so it bypasses connection tracking entirely, and write matching stateless accept rules.&lt;/p&gt;

&lt;h2&gt;
  
  
  After the session comes up
&lt;/h2&gt;

&lt;p&gt;When the upstream accepts your LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; and the prefix is live, scan yourself from outside with &lt;code&gt;nmap -6 -Pn&lt;/code&gt; against the published addresses. You should see exactly the ports in the &lt;code&gt;web6&lt;/code&gt; rule and nothing else. Rerun that scan after every firewall change.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;Write and test the ruleset on a lab VM, then &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;claim a free /48 from IPYE&lt;/a&gt; and announce it only once the firewall is in place. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>security</category>
    </item>
    <item>
      <title>Announce Your Free /48 with BIRD 2: A Small Production-Oriented Config</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 01:42:41 +0000</pubDate>
      <link>https://dev.to/ipye_com/announce-your-free-48-with-bird-2-a-small-production-oriented-config-2dic</link>
      <guid>https://dev.to/ipye_com/announce-your-free-48-with-bird-2-a-small-production-oriented-config-2dic</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;BIRD 2 is a small, fast routing daemon with a filter language that makes export policy explicit. That suits a single-prefix edge: you want to announce one aggregate and accept a controlled set of routes. This article builds a production-oriented BIRD 2 config for a free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt;, using ASN &lt;code&gt;64496&lt;/code&gt;, an upstream &lt;code&gt;64497&lt;/code&gt; at &lt;code&gt;2001:db8:0:1::1&lt;/code&gt;, and the prefix &lt;code&gt;2001:db8:100::/48&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Install and keep the config in Git
&lt;/h2&gt;

&lt;p&gt;On Debian or Ubuntu, &lt;code&gt;apt install bird2&lt;/code&gt; installs the daemon with its config at &lt;code&gt;/etc/bird/bird.conf&lt;/code&gt;. Put that file under version control from the first day and deploy it with a check:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;bird -p -c /etc/bird/bird.conf &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; birdc configure
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;bird -p&lt;/code&gt; parses the file and exits, so a syntax error never reaches the running daemon. &lt;code&gt;birdc configure&lt;/code&gt; reloads gracefully and keeps sessions up when only filters change.&lt;/p&gt;

&lt;h2&gt;
  
  
  Base configuration
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight conf"&gt;&lt;code&gt;&lt;span class="n"&gt;router&lt;/span&gt; &lt;span class="n"&gt;id&lt;/span&gt; &lt;span class="mf"&gt;192.0&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="mf"&gt;2.1&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="n"&gt;log&lt;/span&gt; &lt;span class="n"&gt;syslog&lt;/span&gt; &lt;span class="n"&gt;all&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;

&lt;span class="na"&gt;define OUR_ASN&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;64496&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="na"&gt;define OUR_V6&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="o"&gt;::/&lt;/span&gt;&lt;span class="mi"&gt;48&lt;/span&gt; &lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;

&lt;span class="n"&gt;protocol&lt;/span&gt; &lt;span class="n"&gt;device&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;

&lt;span class="n"&gt;protocol&lt;/span&gt; &lt;span class="n"&gt;kernel&lt;/span&gt; &lt;span class="n"&gt;kernel6&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="na"&gt;ipv6 { import none; export filter { if source&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="no"&gt;RTS_BGP&lt;/span&gt; &lt;span class="n"&gt;then&lt;/span&gt; &lt;span class="n"&gt;accept&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="n"&gt;reject&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;merge&lt;/span&gt; &lt;span class="n"&gt;paths&lt;/span&gt; &lt;span class="n"&gt;on&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;

&lt;span class="n"&gt;protocol&lt;/span&gt; &lt;span class="n"&gt;static&lt;/span&gt; &lt;span class="n"&gt;static6&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="n"&gt;ipv6&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;route&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;100&lt;/span&gt;&lt;span class="o"&gt;::/&lt;/span&gt;&lt;span class="mi"&gt;48&lt;/span&gt; &lt;span class="n"&gt;blackhole&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The static blackhole gives BIRD a route to originate and keeps unused parts of the prefix from bouncing back upstream. Specific addresses on interfaces win over the blackhole because they are longer matches.&lt;/p&gt;

&lt;h2&gt;
  
  
  Filters before protocols
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight vcl"&gt;&lt;code&gt;filter export_upstream &lt;span class="o"&gt;{&lt;/span&gt;
  &lt;span class="k"&gt;if&lt;/span&gt; net ~ OUR_V6 &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; &lt;span class="nb"&gt;source&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; RTS_STATIC &lt;span class="k"&gt;then&lt;/span&gt; accept&lt;span class="o"&gt;;&lt;/span&gt;
  reject&lt;span class="o"&gt;;&lt;/span&gt;
&lt;span class="o"&gt;}&lt;/span&gt;

filter import_upstream &lt;span class="o"&gt;{&lt;/span&gt;
  &lt;span class="k"&gt;if&lt;/span&gt; net.len &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; 48 &lt;span class="k"&gt;then&lt;/span&gt; reject&lt;span class="o"&gt;;&lt;/span&gt;
  &lt;span class="k"&gt;if&lt;/span&gt; net ~ &lt;span class="o"&gt;[&lt;/span&gt; 2001:db8::/32+, fc00::/7+, fe80::/10+, ::/8+ &lt;span class="o"&gt;]&lt;/span&gt; &lt;span class="k"&gt;then&lt;/span&gt; reject&lt;span class="o"&gt;;&lt;/span&gt;
  &lt;span class="k"&gt;if&lt;/span&gt; bgp_path.len &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; 64 &lt;span class="k"&gt;then&lt;/span&gt; reject&lt;span class="o"&gt;;&lt;/span&gt;
  accept&lt;span class="o"&gt;;&lt;/span&gt;
&lt;span class="o"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The export filter is the safety net that matters most: it accepts only the exact aggregate from the static protocol and rejects everything else, including routes learned from other peers. Note the &lt;code&gt;+&lt;/code&gt; operator, which matches the prefix and all longer ones.&lt;/p&gt;

&lt;p&gt;In production, drop &lt;code&gt;2001:db8::/32&lt;/code&gt; from your own bogon list only if you are actually using documentation space in a lab; on the Internet it should always be rejected.&lt;/p&gt;

&lt;h2&gt;
  
  
  The BGP protocol
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight conf"&gt;&lt;code&gt;&lt;span class="n"&gt;protocol&lt;/span&gt; &lt;span class="n"&gt;bgp&lt;/span&gt; &lt;span class="n"&gt;upstream1&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="n"&gt;description&lt;/span&gt; &lt;span class="s2"&gt;"VPS upstream"&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;local&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt; &lt;span class="n"&gt;as&lt;/span&gt; &lt;span class="no"&gt;OUR_ASN&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;neighbor&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt; &lt;span class="n"&gt;as&lt;/span&gt; &lt;span class="mi"&gt;64497&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;hold&lt;/span&gt; &lt;span class="n"&gt;time&lt;/span&gt; &lt;span class="mi"&gt;30&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;graceful&lt;/span&gt; &lt;span class="n"&gt;restart&lt;/span&gt; &lt;span class="n"&gt;on&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;ipv6&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;limit&lt;/span&gt; &lt;span class="mi"&gt;1000&lt;/span&gt; &lt;span class="n"&gt;action&lt;/span&gt; &lt;span class="n"&gt;restart&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;filter&lt;/span&gt; &lt;span class="n"&gt;import_upstream&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
    &lt;span class="n"&gt;export&lt;/span&gt; &lt;span class="n"&gt;filter&lt;/span&gt; &lt;span class="n"&gt;export_upstream&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;For a multihop route server, add &lt;code&gt;multihop 2;&lt;/code&gt; and make sure a route to the server exists. If the provider requires MD5, add a &lt;code&gt;password&lt;/code&gt; line and load the secret from a file that is not in Git, using BIRD's &lt;code&gt;include&lt;/code&gt; statement.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verification
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;birdc show protocols all upstream1
birdc show route export upstream1
birdc show route for 2001:db8:100::1 all
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;show route export&lt;/code&gt; is the line to check before you tell anyone the prefix is live. It prints exactly what BIRD is sending after the filter. If it lists more than your /48, stop and fix the filter.&lt;/p&gt;

&lt;h2&gt;
  
  
  Test filters without a live session
&lt;/h2&gt;

&lt;p&gt;BIRD can evaluate a filter against a route already in its table:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;birdc eval '2001:db8:100::/48 ~ [ 2001:db8:100::/48 ]'
birdc show route filter export_upstream
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The second command lists routes in the master table that would pass the export filter. Running it in CI against a staging daemon is a cheap way to catch a filter edit that suddenly exports more than one prefix.&lt;/p&gt;

&lt;h2&gt;
  
  
  Match the origin to the paperwork
&lt;/h2&gt;

&lt;p&gt;The upstream will usually ask for an LOA naming &lt;code&gt;AS64496&lt;/code&gt;; download it from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt;. The ROA is created automatically, so the last check is a validator: run &lt;code&gt;routinator vrps --noupdate | grep 2001:db8:100&lt;/code&gt; and confirm the origin matches the &lt;code&gt;local ... as&lt;/code&gt; line in BIRD. If you later change ASN in the IPYE dashboard, update &lt;code&gt;OUR_ASN&lt;/code&gt;, reload, and recheck the VRP before you expect global reachability.&lt;/p&gt;

&lt;h2&gt;
  
  
  Adding a second peer
&lt;/h2&gt;

&lt;p&gt;Use a template so both sessions share the same policy:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight conf"&gt;&lt;code&gt;&lt;span class="n"&gt;template&lt;/span&gt; &lt;span class="n"&gt;bgp&lt;/span&gt; &lt;span class="n"&gt;upstream_tpl&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="n"&gt;local&lt;/span&gt; &lt;span class="n"&gt;as&lt;/span&gt; &lt;span class="no"&gt;OUR_ASN&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;hold&lt;/span&gt; &lt;span class="n"&gt;time&lt;/span&gt; &lt;span class="mi"&gt;30&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
  &lt;span class="n"&gt;ipv6&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;filter&lt;/span&gt; &lt;span class="n"&gt;import_upstream&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="n"&gt;export&lt;/span&gt; &lt;span class="n"&gt;filter&lt;/span&gt; &lt;span class="n"&gt;export_upstream&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="n"&gt;protocol&lt;/span&gt; &lt;span class="n"&gt;bgp&lt;/span&gt; &lt;span class="n"&gt;upstream2&lt;/span&gt; &lt;span class="n"&gt;from&lt;/span&gt; &lt;span class="n"&gt;upstream_tpl&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="n"&gt;neighbor&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt; &lt;span class="n"&gt;as&lt;/span&gt; &lt;span class="mi"&gt;64498&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  Logs and metrics
&lt;/h2&gt;

&lt;p&gt;Send BIRD logs to syslog and alert on lines that contain &lt;code&gt;BGP session&lt;/code&gt; state changes. For metrics, the bird_exporter project exposes protocol state and route counts to Prometheus; alert when the exported route count for an upstream is anything other than one.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, drop it into &lt;code&gt;OUR_V6&lt;/code&gt;, and confirm &lt;code&gt;show route export&lt;/code&gt; lists only that prefix before the first session comes up. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Announce One Free /48 from Two VPS Providers Without Renumbering</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 01:39:05 +0000</pubDate>
      <link>https://dev.to/ipye_com/announce-one-free-48-from-two-vps-providers-without-renumbering-5879</link>
      <guid>https://dev.to/ipye_com/announce-one-free-48-from-two-vps-providers-without-renumbering-5879</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Announcing one prefix from two unrelated VPS providers gives you provider redundancy without renumbering anything when one of them fails. The prefix stays the same; only the paths change. A free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt; is a practical way to build this, because it is a real allocation you announce from your own ASN. Examples use ASN &lt;code&gt;64496&lt;/code&gt;, upstream A as &lt;code&gt;64497&lt;/code&gt; with neighbour &lt;code&gt;2001:db8:a::1&lt;/code&gt;, upstream B as &lt;code&gt;64498&lt;/code&gt; with neighbour &lt;code&gt;2001:db8:b::1&lt;/code&gt;, and the prefix &lt;code&gt;2001:db8:100::/48&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Decide what multihoming means for you
&lt;/h2&gt;

&lt;p&gt;There are two common shapes:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Active/standby.&lt;/strong&gt; Site A carries traffic and site B is announced with AS-path prepends, so the Internet prefers A until it disappears.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Anycast.&lt;/strong&gt; Both sites announce the same prefix with equal attributes and serve the same stateless service, such as DNS, so users land on whichever site their network reaches first.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Stateful applications such as databases or long-lived TCP sessions are a poor fit for anycast across providers. A session that starts at A and flips to B mid-stream breaks, because B has no state for it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Split the /48 deliberately
&lt;/h2&gt;

&lt;p&gt;Keep the /48 as the announced aggregate and number each site from its own /56 inside it:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Range&lt;/th&gt;
&lt;th&gt;Use&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100::/56&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;site A services&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100:100::/56&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;site B services&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;2001:db8:100:ff00::/56&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;anycast service addresses&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Export only the /48. Many networks filter IPv6 prefixes longer than /48, so a /56 announced alone would not be reachable widely. The internal plan still matters: it tells each site which addresses it really owns when traffic arrives.&lt;/p&gt;

&lt;h2&gt;
  
  
  Getting traffic to the right site
&lt;/h2&gt;

&lt;p&gt;Both sites announce the same /48, so a packet for site B's /56 may arrive at site A. You have three options:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Run a tunnel between the sites (WireGuard or GRE) and route the other site's /56 across it.&lt;/li&gt;
&lt;li&gt;Make every service anycast-capable so it does not matter where the packet lands.&lt;/li&gt;
&lt;li&gt;Accept that site-specific addresses only work while their site is the preferred path, which is fine for active/standby.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The tunnel is the most common choice. Keep it on the provider-assigned addresses, never on addresses inside the /48, or the tunnel will try to route through itself.&lt;/p&gt;

&lt;h2&gt;
  
  
  Policy sketch for active/standby in FRR
&lt;/h2&gt;

&lt;p&gt;Site A exports the aggregate plainly. Site B prepends:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;route-map OUT-B permit 10
 match ipv6 address prefix-list OUR-48
 &lt;span class="nb"&gt;set&lt;/span&gt; as-path prepend 64496 64496
!
ipv6 prefix-list OUR-48 seq 10 permit 2001:db8:100::/48
!
router bgp 64496
 address-family ipv6 unicast
  neighbor 2001:db8:b::1 route-map OUT-B out
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Prepending is a hint, not a command. Some networks prefer customer routes or use local preference that overrides path length, so a share of traffic may still reach B. Measure before you assume the standby is idle.&lt;/p&gt;

&lt;h2&gt;
  
  
  Paperwork for two hosts
&lt;/h2&gt;

&lt;p&gt;Each provider will usually want its own LOA naming your ASN. Generate both from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; and file them in each ticket. The ROA covers the origin ASN rather than a specific upstream, so one ROA for &lt;code&gt;AS64496&lt;/code&gt; serves both sites. Before you go live, confirm that both providers have updated their export filters; a common mistake is one site announcing correctly while the other provider silently drops the route.&lt;/p&gt;

&lt;h2&gt;
  
  
  Firewall symmetry
&lt;/h2&gt;

&lt;p&gt;Asymmetric paths are normal in multihoming: a packet can arrive through A and the reply can leave through B. A stateful firewall at each site only sees half of that conversation and may drop it. Either keep replies on the arrival site, using policy routing keyed on the source /56, or loosen connection tracking for that traffic. Reverse-path filtering causes the same symptom on Linux, so set &lt;code&gt;net.ipv4.conf.all.rp_filter=2&lt;/code&gt; (loose) for IPv4 and check any &lt;code&gt;fib saddr&lt;/code&gt; rules in nftables for IPv6.&lt;/p&gt;

&lt;h2&gt;
  
  
  Failover drill
&lt;/h2&gt;

&lt;p&gt;Run this in a maintenance window:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Record the current best paths from a few public route collectors.&lt;/li&gt;
&lt;li&gt;Shut the session at site A with &lt;code&gt;neighbor 2001:db8:a::1 shutdown&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Watch how long collectors take to show the prepended path via B, and run &lt;code&gt;mtr -6&lt;/code&gt; from an outside host.&lt;/li&gt;
&lt;li&gt;Restore A and confirm traffic moves back.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Write the observed convergence time in your runbook. Your numbers are the only ones that matter for your design.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;Claim a free /48 from IPYE&lt;/a&gt;, bring up one site, and add the second only after the first is visible in public collectors. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>Your First eBGP Session on a VPS with a Free /48 from IPYE</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 01:35:10 +0000</pubDate>
      <link>https://dev.to/ipye_com/your-first-ebgp-session-on-a-vps-with-a-free-48-from-ipye-1mf5</link>
      <guid>https://dev.to/ipye_com/your-first-ebgp-session-on-a-vps-with-a-free-48-from-ipye-1mf5</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48, LOA, or ROA helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The first eBGP session on a rented VPS usually stalls on paperwork and small kernel details, not on BGP syntax. A free IPv6 /48 from &lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;IPYE&lt;/a&gt; removes most of the paperwork: setup is instant, the allocation is real and announceable from your own ASN, the LOA comes from the portal, and the ROA is created automatically. This walkthrough brings up FRR on a BGP-capable VPS using documentation values: your ASN &lt;code&gt;64496&lt;/code&gt;, the provider's ASN &lt;code&gt;64497&lt;/code&gt;, the provider's neighbour &lt;code&gt;2001:db8:0:1::1&lt;/code&gt;, and the prefix &lt;code&gt;2001:db8:100::/48&lt;/code&gt;. Substitute your real values when you go live.&lt;/p&gt;

&lt;h2&gt;
  
  
  Preconditions
&lt;/h2&gt;

&lt;p&gt;Ask the host three things before you touch the config. Does the product support BGP sessions with customer ASNs, or only static routing of a bring-your-own prefix? Is the session single-hop to the gateway, or multihop to a route server on a loopback? Do they require an MD5 password, and do they want the LOA attached to the ticket? Keep SSH on the address the provider assigned, never on the /48, so a withdrawn announcement cannot lock you out.&lt;/p&gt;

&lt;h2&gt;
  
  
  Give the route a home
&lt;/h2&gt;

&lt;p&gt;FRR's &lt;code&gt;network&lt;/code&gt; statement only advertises a prefix that already exists in the routing table, unless you disable the import check. The clean fix is a blackhole route for the aggregate plus real addresses on a loopback or dummy interface:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;ip link add dummy0 &lt;span class="nb"&gt;type&lt;/span&gt; dummy
ip link &lt;span class="nb"&gt;set&lt;/span&gt; dummy0 up
ip -6 addr add 2001:db8:100::1/128 dev dummy0
ip -6 route add blackhole 2001:db8:100::/48
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The blackhole keeps unused parts of the /48 from looping back to the upstream, while the longer /128 still delivers traffic to the service address. Persist all of this with systemd-networkd or netplan; a reboot that loses the dummy interface leaves you announcing space the host cannot answer.&lt;/p&gt;

&lt;h2&gt;
  
  
  FRR configuration
&lt;/h2&gt;

&lt;p&gt;Enable &lt;code&gt;bgpd&lt;/code&gt; in &lt;code&gt;/etc/frr/daemons&lt;/code&gt;, then:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;router bgp 64496
 bgp router-id 192.0.2.1
 neighbor 2001:db8:0:1::1 remote-as 64497
 neighbor 2001:db8:0:1::1 description vps-uplink
 neighbor 2001:db8:0:1::1 timers 10 30
 !
 address-family ipv6 unicast
  network 2001:db8:100::/48
  neighbor 2001:db8:0:1::1 activate
  neighbor 2001:db8:0:1::1 prefix-list OUT-V6 out
  neighbor 2001:db8:0:1::1 prefix-list IN-V6 &lt;span class="k"&gt;in&lt;/span&gt;
  neighbor 2001:db8:0:1::1 maximum-prefix 10
 exit-address-family
!
ipv6 prefix-list OUT-V6 seq 10 permit 2001:db8:100::/48
ipv6 prefix-list IN-V6 seq 10 permit ::/0
ipv6 prefix-list IN-V6 seq 99 deny any
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Recent FRR releases enforce RFC 8212 by default (&lt;code&gt;bgp ebgp-requires-policy&lt;/code&gt;), so an eBGP neighbour with no inbound or outbound policy sends and accepts nothing. Applying prefix-lists in both directions satisfies that check without disabling it. If the host only sends you a default route, the inbound list above is all you need; if they send a full table, see the memory notes before you raise &lt;code&gt;maximum-prefix&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;For a multihop route server, add &lt;code&gt;neighbor 2001:db8:0:1::1 ebgp-multihop 2&lt;/code&gt; and make sure a static route to the server's address exists. For MD5, add &lt;code&gt;neighbor 2001:db8:0:1::1 password&lt;/code&gt; followed by the shared secret from your ticket, and keep that secret out of Git.&lt;/p&gt;

&lt;h2&gt;
  
  
  Paperwork in parallel
&lt;/h2&gt;

&lt;p&gt;While the host reviews the ticket, download the LOA from the &lt;a href="https://my.ipye.com" rel="noopener noreferrer"&gt;IPYE portal&lt;/a&gt; and attach it with your ASN and the peer addresses. Then check the ROA in a validator before you announce, for example with Routinator: &lt;code&gt;routinator vrps --noupdate | grep 2001:db8:100&lt;/code&gt;. The VRP should show origin &lt;code&gt;AS64496&lt;/code&gt; and a maxLength that covers exactly what you plan to send.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verification beyond Established
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast summary'&lt;/span&gt;
vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast neighbors 2001:db8:0:1::1 advertised-routes'&lt;/span&gt;
vtysh -c &lt;span class="s1"&gt;'show bgp ipv6 unicast 2001:db8:100::/48'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;advertised-routes&lt;/code&gt; is the important one: it shows what actually left your box after policy. From a host on another network, run &lt;code&gt;ping -6 2001:db8:100::1&lt;/code&gt; and &lt;code&gt;mtr -6&lt;/code&gt; toward the loopback. Then look the prefix up in RIPEstat or a public route collector and confirm origin &lt;code&gt;AS64496&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Common first-session failures
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Session stuck in Active:&lt;/strong&gt; TCP 179 filtered, wrong neighbour address, or a missing multihop setting. &lt;code&gt;tcpdump -ni eth0 port 179&lt;/code&gt; shows whether SYNs leave and whether anything answers.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Established, zero prefixes sent:&lt;/strong&gt; the aggregate is missing from the RIB, or the outbound prefix-list does not match.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Visible at the host, invisible elsewhere:&lt;/strong&gt; the host has not yet updated its export filter, or the ROA does not match the origin.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Session flaps after a reboot:&lt;/strong&gt; FRR started before the dummy interface existed. Order the units so networking comes up first.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;To roll back, run &lt;code&gt;neighbor 2001:db8:0:1::1 shutdown&lt;/code&gt; under &lt;code&gt;router bgp&lt;/code&gt;. The session stays configured but announces nothing.&lt;/p&gt;

&lt;h2&gt;
  
  
  Do this next
&lt;/h2&gt;

&lt;p&gt;When the host confirms BGP support, &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;claim a free /48 from IPYE&lt;/a&gt; and run through this checklist with your own ASN. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>ipv6</category>
      <category>linux</category>
      <category>devops</category>
    </item>
    <item>
      <title>WireGuard Site-to-Site with BGP—Route Your Free /48 Through the Tunnel</title>
      <dc:creator>IPYE</dc:creator>
      <pubDate>Tue, 10 Nov 2026 00:22:58 +0000</pubDate>
      <link>https://dev.to/ipye_com/wireguard-site-to-site-with-bgp-route-your-free-48-through-the-tunnel-49kp</link>
      <guid>https://dev.to/ipye_com/wireguard-site-to-site-with-bgp-route-your-free-48-through-the-tunnel-49kp</guid>
      <description>&lt;p&gt;&lt;em&gt;Disclosure up front: I run IPYE, a company that leases IP space. Most of this article is practical how-to; IPYE appears where a free /48 helps you try the work, with a short call to action at the end.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;WireGuard encrypts; BGP decides which prefixes use the tunnel. Stuffing every subnet into &lt;code&gt;AllowedIPs&lt;/code&gt; alone does not scale. This design uses WireGuard for the outer path and FRR for site prefixes. A natural practice payload is IPYE's free IPv6 /48 (first year free, instant setup, real allocation on your ASN, portal LOA, automatic ROA)—announce it at site A and learn it at site B over &lt;code&gt;wg0&lt;/code&gt;. Lab stand-ins: &lt;code&gt;192.0.2.0/24&lt;/code&gt;, &lt;code&gt;198.51.100.0/24&lt;/code&gt;, tunnel &lt;code&gt;2001:db8:ffff::/64&lt;/code&gt;, ASNs &lt;code&gt;64496&lt;/code&gt;/&lt;code&gt;64497&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Design rules
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Each WireGuard interface gets a small point-to-point /127 or /126 (IPv6) or /30 (IPv4) used only for BGP and probes.&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;AllowedIPs&lt;/code&gt; on each peer includes that tunnel subnet &lt;strong&gt;plus&lt;/strong&gt; the remote site's loopback or a broad encapsulation selector you intentionally manage—not every customer /48 you might ever learn.&lt;/li&gt;
&lt;li&gt;BGP (or OSPF) runs over the tunnel and advertises the real site prefixes.&lt;/li&gt;
&lt;li&gt;Firewall policy allows UDP 51820 (or your chosen port) between public endpoints and BGP between tunnel IPs.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you put the entire Internet into &lt;code&gt;AllowedIPs&lt;/code&gt;, WireGuard becomes your default route cryptokey router and you lose the ability to learn backup paths sensibly.&lt;/p&gt;

&lt;h2&gt;
  
  
  WireGuard skeleton (site A)
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight ini"&gt;&lt;code&gt;&lt;span class="k"&gt;[Interface]&lt;/span&gt;
&lt;span class="na"&gt;PrivateKey&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="n"&gt;siteA&lt;/span&gt; &lt;span class="n"&gt;private&lt;/span&gt;&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;
&lt;span class="na"&gt;Address&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;ffff&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="o"&gt;/&lt;/span&gt;&lt;span class="mi"&gt;127&lt;/span&gt;
&lt;span class="na"&gt;ListenPort&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;51820&lt;/span&gt;

&lt;span class="k"&gt;[Peer]&lt;/span&gt;
&lt;span class="na"&gt;PublicKey&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;&lt;/span&gt;&lt;span class="n"&gt;siteB&lt;/span&gt; &lt;span class="n"&gt;public&lt;/span&gt;&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;
&lt;span class="na"&gt;Endpoint&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mf"&gt;203.0&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="mf"&gt;113.10&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="mi"&gt;51820&lt;/span&gt;
&lt;span class="na"&gt;AllowedIPs&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;2001&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;db8&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;ffff&lt;/span&gt;&lt;span class="o"&gt;::&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="o"&gt;/&lt;/span&gt;&lt;span class="mi"&gt;127&lt;/span&gt;
&lt;span class="na"&gt;PersistentKeepalive&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;25&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Site B mirrors with &lt;code&gt;::0/127&lt;/code&gt; addressing swapped. Bring it up with &lt;code&gt;wg-quick up wg0&lt;/code&gt; and verify &lt;code&gt;ping6 2001:db8:ffff::0&lt;/code&gt;. Only after the tunnel pings should you enable BGP.&lt;/p&gt;

&lt;h2&gt;
  
  
  FRR over the tunnel
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight cisco_ios"&gt;&lt;code&gt;router bgp 64496
 bgp router-id 192.0.2.1
 neighbor 2001:db8:ffff::0 remote-as 64497
 !
 address-family ipv4 unicast
  neighbor 2001:db8:ffff::0 activate
  network 192.0.2.0/24
  neighbor 2001:db8:ffff::0 prefix-list TO-B out
  neighbor 2001:db8:ffff::0 prefix-list FROM-B &lt;span class="k"&gt;in&lt;/span&gt;
 exit-address-family
!
ip prefix-list TO-B permit 192.0.2.0/24
ip prefix-list FROM-B permit 198.51.100.0/24
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Site B advertises &lt;code&gt;198.51.100.0/24&lt;/code&gt; symmetrically. Use eBGP multihop only if you peer to loopbacks; with point-to-point tunnel IPs, TTL one is enough.&lt;/p&gt;

&lt;h2&gt;
  
  
  AllowedIPs versus BGP routes
&lt;/h2&gt;

&lt;p&gt;WireGuard still drops packets whose inner destination is outside &lt;code&gt;AllowedIPs&lt;/code&gt;. So when BGP learns &lt;code&gt;198.51.100.0/24&lt;/code&gt;, you must either:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;add &lt;code&gt;198.51.100.0/24&lt;/code&gt; to &lt;code&gt;AllowedIPs&lt;/code&gt; on site A (static, simple, two-site), or&lt;/li&gt;
&lt;li&gt;use a broader AllowedIPs (for example the aggregate you own) carefully, or&lt;/li&gt;
&lt;li&gt;run a helper that updates AllowedIPs from the BGP table (advanced; automate with care).&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For two sites, listing the remote site aggregate in AllowedIPs and letting BGP select paths among multiple tunnels is usually enough. For many sites, design hierarchical aggregates so AllowedIPs stays small.&lt;/p&gt;

&lt;h2&gt;
  
  
  Multiple tunnels and failover
&lt;/h2&gt;

&lt;p&gt;Build two WireGuard sessions (primary and backup paths). Run BGP over both with different local preference or AS-path length so one tunnel is preferred. When the primary UDP path dies, BGP times out (prefer BFD if available) and the backup path takes over without rewriting AllowedIPs. Keepalive on WireGuard helps NAT middleboxes but does not replace routing timers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Verification
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;wg show
ping6 2001:db8:ffff::0
vtysh -c &lt;span class="s1"&gt;'show bgp summary'&lt;/span&gt;
vtysh -c &lt;span class="s1"&gt;'show ip route 198.51.100.0'&lt;/span&gt;
traceroute 198.51.100.10
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Confirm the traceroute enters &lt;code&gt;wg0&lt;/code&gt; and that killing the primary endpoint shifts the BGP best path. Site-to-site VPN done this way stays understandable months later: crypto on the outside, routing on the inside.&lt;/p&gt;

&lt;h2&gt;
  
  
  Firewall and policy notes
&lt;/h2&gt;

&lt;p&gt;On each site, allow UDP to the WireGuard port only from the peer's public IP when peers are fixed. If a peer has a dynamic home endpoint, Prefer keepalive and authenticate solely by key, but still rate-limit. Allow TCP 179 only between tunnel addresses, not on the public WAN. Separate management SSH from tunnel traffic so a WireGuard outage does not lock you out of the box (out-of-band or provider console access remains mandatory).&lt;/p&gt;

&lt;h2&gt;
  
  
  IPv6 and dual-stack
&lt;/h2&gt;

&lt;p&gt;Prefer IPv6 transport for the outer WireGuard endpoints when both sites have stable AAAA reachability; it avoids CGNAT pain on the outer path. Inside, run both address families in BGP if you carry both. Do not assume that an IPv6 tunnel automatically carries IPv4 payloads without either IPv4 tunnel addresses or a deliberate encapsulation choice—WireGuard carries layer 3 packets for the families you configure in &lt;code&gt;Address&lt;/code&gt;/&lt;code&gt;AllowedIPs&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  When static AllowedIPs is enough
&lt;/h2&gt;

&lt;p&gt;Two sites, stable subnets, no failover engineering: a pure WireGuard config without BGP is fine. Add BGP when you grow a third site, need traffic engineering between two tunnels, or want customer prefixes to appear and disappear without rewriting VPN configs. The goal is boring failure domains, not maximum protocol count.&lt;/p&gt;

&lt;h2&gt;
  
  
  Key rotation and peer lifecycle
&lt;/h2&gt;

&lt;p&gt;Rotate WireGuard keys on a planned cadence and when staff leave. Because routing is separate, you can bring up a parallel tunnel with new keys, shift BGP preference, then decommission the old peer without renumbering LAN subnets. Record public keys in a password manager or vault, not in a wiki plain page. Treat &lt;code&gt;wg0&lt;/code&gt; conf like a router config: peer-reviewed and stored in git with secrets filtered.&lt;/p&gt;

&lt;h2&gt;
  
  
  Carry the free /48 over WG
&lt;/h2&gt;

&lt;p&gt;Put the free /48 from my.ipye.com/products/ipv6-rental in BGP and in &lt;code&gt;AllowedIPs&lt;/code&gt; (or a covering aggregate). If you change region or ASN in the IPYE dashboard, those changes apply automatically—re-check tunnel endpoints and ROA after moves between VPS providers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Lab first
&lt;/h2&gt;

&lt;p&gt;Before touching production routers, reproduce the pattern with two namespaces or two cheap VMs: WireGuard up, BGP established, kill &lt;code&gt;-9&lt;/code&gt; the primary &lt;code&gt;wg&lt;/code&gt; process, and watch the secondary path take traffic. Only promote the design when that failover is boring. Boring is the goal.&lt;/p&gt;

&lt;h2&gt;
  
  
  Tunnel a free /48 this weekend
&lt;/h2&gt;

&lt;p&gt;Claim space at the IPYE IPv6 rental page (see About). Site ipye.com (see About) · &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  About IPYE
&lt;/h2&gt;

&lt;p&gt;IPYE (&lt;a href="https://ipye.com" rel="noopener noreferrer"&gt;https://ipye.com&lt;/a&gt;) leases IPv4, IPv6 and ASN space with instant setup. Your first IPv6 /48 is free for the first year, a real allocation you can announce on your own ASN, with a LOA from the portal and the ROA created automatically. ASN and region changes apply automatically. IPv4 /24 to /22 leases include LOA, ROA and per-/24 rDNS; WHOIS is editable on IPv6 /44 and larger. Locations: Los Angeles, New York, Ashburn, Salt Lake City, London, Amsterdam, Frankfurt. Claim your free /48: &lt;a href="https://my.ipye.com/products/ipv6-rental" rel="noopener noreferrer"&gt;https://my.ipye.com/products/ipv6-rental&lt;/a&gt;. Questions: &lt;a href="mailto:ipye@turnel.com"&gt;ipye@turnel.com&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: I run IPYE&lt;/em&gt;&lt;/p&gt;

</description>
      <category>networking</category>
      <category>linux</category>
      <category>tutorial</category>
    </item>
  </channel>
</rss>
