Optimal Configuration of a Routing Node
Get tips on how to configure LND to get the most out of your routing node
Note: Some of the configuration settings have changed with LND 0.16. You can find the latest settings below.
LND offers a wide range of configuration options that allow deployment in a large variety of platforms, environments and purposes, making use of either btcd, bitcoind or neutrino as a source of blockchain data.
In this article we will discuss various configuration options for LND in the context of a routing node.
The obligations of a good routing node fall into two broad categories. First and in this article, we will need to install and configure LND. Later we will need to open channels, get inbound liquidity and manage the liquidity of our node in a smart way. Have a look at the liquidity and routing sections of the Builder’s Guide.
Your node should be available as much as possible with little latency. This allows others to open channels with you and lets your node respond quickly to payments flowing to you. Your network, server and whether you deploy LND behind Tor or a proxy all play a role in this metric.
There are many reasons why a channel might be marked as ‘disabled’ by one peer or another. Keeping your channel available is an important metric to assess the quality of a node. Keep your node up to date and regularly observe the quality of your channels. Configurations such as tor stream isolation can also affect the quality of channels.
For a high-performance routing node you will need btcd or bitcoind running without pruning. Ideally both the Bitcoin backend and LND will run on the same machine, but it is also possible to connect them via ssh on separate servers, as long as latency is low enough.
For performance reasons you may also specifically configure your Bitcoin node. For instance, when building bitcoind from source you can use the command
./configure CXXFLAGS="--param ggc-min-expand=1 --param ggc-min-heapsize=32768" --enable-cxx --with-zmq --without-gui --disable-shared --with-pic --disable-tests --disable-bench --enable-upnp-default --disable-wallet
make -j "$(($(nproc)+1))"
sudo make installThe parameters in this command are explained below:
CXXFLAGS="--param ggc-min-expand=1 --param ggc-min-heapsize=32768"This allows us to conserve memory.
--with-zmqZMQ is used to stream data from bitcoind to LND.
--without-guiWe will not need the graphical interface for our setup.
--enable-upnp-defaultUPnP allows for automatic port mapping.
--disable-walletAs we will not be using the bitcoind wallet we can disable it.
make -j "$(($(nproc)+1))"might help to speed up the build on some machines compared to make.
Your routing node will need a static IP with open ports or a version 3 onion address if you want others to be able to connect and open channels with you.
If you do not already have a configuration file, you will need to create one and place it in your LND directory. On Linux, that is typically
/Users/[username]/Library/Application Support/Lnd/lnd.confand in Windows
You may also use the lnd.conf sample file and activate the relevant lines by removing the semi-colons at the beginning of the relevant lines.
Make your node visible and more easily discoverable with a unique alias. This might be a url to where peers can find out more information about you and your node, the name or your business or something memorable. Anybody can set any name and color without verification, but don’t abuse this!
Make sure to set the external IP of your LND node here. This IP address should be static. You may remove this line if you are using Tor.
CNCT, CRTR and HSWV provide channel-related logs, while NTFN provides chain-related logs. RPCS will provide you with RPC-related logs.
Activate your Lightning node for Bitcoin payments on mainnet. We specify the Bitcoin node of your choice. Other choices are Beutrino and btcd
You may freely decide on a RPC password and username, as long as it is long and unique. This also needs to be set in your bitcoin.conf file.
We will need to enable ZMQ for our LND node to receive information about the latest blocks from bitcoind.
Set the smallest HTLC you would like to forward in milli-satoshi.
Our routing node will not need historical gossip data, so we can ignore it with this flag.
This is our base fee in milli-satoshi. Meaning for each payment we forward we expect to be paid at least 1 satoshi. The feerate is the fee we charge per 1 million forwarded satoshi.
This setting will compact our
channel.dbdatabase at every startup, which will improve performance of your node.
We can set the maximum amount of fees in a channel here as a percentage of individual channel capacity. The setting allows for one decimal place and defaults to 0.5.
We can set the maximum pending channels with this configuration.
The number of confirmations we expect before a channel is considered active.
This allows our node to accept and create channels larger than 0.16777215 BTC.
Similarly the minimum channel size, denominated in satoshis, allows your routing node to reject small channels. This might mean fewer, but more qualitative channels.
A high CLTV expiration value makes it less likely that we have to settle a forwarded payment on chain, but also requires us to maintain a high uptime. In this example, the maximum number of blocks our funds are timelocked is about one month.
To make use of the following configurations, make sure you are using the binary release or built lnd from source with the routerrpc tag. The following parameters are important if you are using LND to make payments in the Lightning Network, rather than just routing the payments of others.
This sets the default chance of a hop being successful at 50%.
If a node returns many failures LND will begin to ignore them more and more. To turn off this feature set the value to 1.
LND may try paths less likely to succeed only if they allow for significant savings. You can set the minimum of these desired savings here.
LND will keep historical routing records to evaluate future routing success. You can set how many records should be kept.
When setting this parameter to a very small number, LND will try paths even when they have a very low chance of success.
This setting allows you to define after how long LND should forget about past routing failures.
You may want to manage and monitor your node with remote tools requiring RPC. These configurations help you do that.
Defending your node against (distributed) denial of service attacks is not an easy feat. Depending on your node setup, you may opt for one of multiple tools to keep satoshis flowing under all circumstances.
As a precaution and for reasons beyond DDoS protection, do not make multiple services, such as bitcoind and lnd available on the same IP addresses or onion service.
With iptables you can configure how packets are filtered before they reach LND. While executing these rules can also be computationally intensive, it generally allows your machine to sustain a far higher load.
We suggest the following iptable rules for network flood protection:
sudo iptables -N syn_flood
sudo iptables -A INPUT -p tcp --syn -j syn_flood
sudo iptables -A syn_flood -m limit --limit 1/s --limit-burst 3 -j RETURN
sudo iptables -A syn_flood -j DROP
sudo iptables -A INPUT -p icmp -m limit --limit 1/s --limit-burst 1 -j ACCEPT
sudo iptables -A INPUT -p icmp -m limit --limit 1/s --limit-burst 1 -j LOG --log-prefix PING-DROP:
sudo iptables -A INPUT -p icmp -j DROP
sudo iptables -A OUTPUT -p icmp -j ACCEPT
While the Tor network is not immune to DDoS attacks, it may be able to help you stay available to your peers in ways a clearnet IP address may not.Botnets and other tools deployed in DDoS attacks may not work over Tor, and although the network has its own bandwidth constraints, it also has its own ways to mitigate attacks.
Content Delivery Networks (CDNs) have paid infrastructure in place that can detect and mitigate even the most powerful DDoS attacks. To configure your node with a CDN, you will need to configure it with a domain name instead of an IP address.
You will also need to add your domain name to the TLS certificate and instruct LND to advertise a domain name instead of an IP address. If you have previously advertised your IP address, it might be necessary to change that or otherwise restrict traffic.
Many cloud service providers include some basic DDoS protection in their products, or offer them at an extra cost. Similar to CDNs this can be a worthwhile tactic of protecting your node. Do read the fineprint, as protection levels and policies differ. Some providers for example might disable access to your instance completely once they detect a DDoS attack.
While experiencing a sudden large amount of traffic to your node, you may close your ports or even change IP address. You may only remain reachable via the Tor network, or not at all. Unless the entire network or your peers are under attack, you will still be able to make outgoing connections and keep your channels active.