Bolt-agent is a tool for workflow based node management automation through bolt.observer platform.
utility to monitor and manage lightning node
The usage should be pretty self explanatory:
NAME:
bolt-agent - Utility to monitor and manage lightning node
USAGE:
bolt-agent [global options] command [command options] [arguments...]
VERSION:
v0.2.2
COMMANDS:
help, h Shows a list of commands or help for one command
GLOBAL OPTIONS:
--allowedentropy value allowed entropy in bits for channel balances (default: 64)
--apikey value api key
--channel-whitelist value path to file containing a whitelist of channels
--interval value interval to poll - 10s, 1m, 10m or 1h (default: "10s")
--lnddir value path to lnd's base directory (default: "/home/user/.lnd")
--macaroonpath value path to macaroon file
--private report private data as well (default: false)
--preferipv4 if you have the choice between IPv6 and IPv4 prefer IPv4 (default: false)
--rpcserver value host:port of ln daemon (default: "localhost:10009")
--tlscertpath value path to TLS certificate (default: "/home/user/.lnd/tls.cert")
--channel-whitelist value path to file containing a whitelist of channels
--verbosity value log level for V logs (default: 0)
--help, -h show help
--version, -v print the version
ADVANCED OPTIONS:
--datastore-url datastore URL (default: agent-api.bolt.observer:443)
--ignorecln ignore CLN socket (default: false)
--keepalive value keepalive interval (if nothing changed after this time an empty message will be sent) (default: "60s")
--public value used together with --channel-whitelist (default: false)
--url value report url (default: "https://ingress.bolt.observer/api/node-data-report/")
--userest use REST API when true instead of gRPC (default: false)
--fetch-invoices fetch invoices (default: 0)
--fetch-forwards fetch forwards (default: 0)
--fetch-payments fetch payments (default: 0)
--noplugins disable plugins (default: false)
GLOG OPTIONS (described in https://github.com/google/glog#setting-flags):
--verbosity (instead of -v)
--logtostderr
--stderrthreshold
--alsologtostderr
--vmodule
--log_dir
--log_backtrace_at
Agent works with LND and CoreLightning and tries the best to have sane defaults so you can just start it up on your node without further hassle.
Example:
bolt-agent --apikey key
Only thing that is mandatory is the apikey which can also be provided through API_KEY environment variable:
export API_KEY=changeme
bolt-agent
Agent is using glog library for logging which needs writable /tmp (use --logtostderr to prevent writing to filesystem).
Flag --private means whether you want to report data for private ("unannounced") channels too (when true). The default is false.
There is an advanced option of filtering what channels exactly are reported documented in "Filtering on agent side" section.
For more information about what exactly is reported read DATA.md.
If you have a custom LND directory (/storage in the example) you might need:
bolt-agent --lnddir /storage/lnd/ --tlscertpath /storage/lnd/data/secrets/lnd.cert
but that should be it.
By default it will try to communicate with LND using local gRPC connection and readonly.macaroon.
You can use --userest to start using REST API - in that case --rpcserver can be complete URL to the endpoint.
bolt-agent --userest --rpcserver https://bolt.observer/endpoint
For CoreLightning it will use UNIX domain socket. By default agent will try to use ~/.lightning/bitcoin/lightning-rpc.
If that file exists program will default to CoreLightning so if have LND installed on the node too and do not want to use CLN there is a flag
--ignorecln. You can manually override name of UNIX domain socket using --rpcserver flag.
Example:
bolt-agent --rpcserver /storage/cln/socket
Make sure agent is running with correct permissions to access the socket.
There is support for remote connections via commando plugin but you cannot use that option via command-line flags at the moment. We suggest to use
socat utility if you wish to monitor a remote CLN node.
We provide a standalone statically compiled binary. The targets are linux (amd64 build for GNU/Linux), rasp (aarch64 build for GNU/Linux) suitable for installation on your Raspberry PI and darwin (amd64 build for MacOS).
If you need some different architecture that is no problem except that you will have to build the binaries yourself (so you will need to install Golang). You can also open an issue / pull request to extend the defaults (or provide better naming).
You can run:
curl https://bolt.observer/agent | sh
to install the latest version of agent.
The executed script is present here so you can first check what it does. Or you can also do a manual installation.
If you have Nix with flake support you can do:
nix profile install 'github:bolt-observer/agent'
Script will execute same steps that you can invoke manually:
- fetch latest revision from https://github.com/bolt-observer/agent/releases
wget https://github.com/bolt-observer/agent/releases/download/v0.2.2/bolt-agent-v0.2.2-linux.zip https://github.com/bolt-observer/agent/releases/download/v0.2.2/manifest-v0.2.2.txt.asc https://github.com/bolt-observer/agent/releases/download/v0.2.2/manifest-v0.2.2.txt
- verify integrity
wget -qO- https://raw.githubusercontent.com/bolt-observer/agent/main/scripts/keys/developers.asc | gpg --import
gpg --verify manifest-v0.2.2.txt.asc manifest-v0.2.2.txt
and you should see:
gpg: using RSA key F4B8B3B59C1E5AA39A1B9636E897355718E1DBF4
gpg: Good signature from "Gregor Pogacnik <[email protected]>" [ultimate]
or
gpg: using RSA key 04AC8A162C036B650A271A6360886AC6567069C5
gpg: Good signature from "bolt <[email protected]>" [ultimate]
- unpack the compressed binary
unzip bolt-agent-v0.2.2-linux.zip
- copy the binary to a common place
cp bolt-agent-v0.2.2-linux /usr/local/bin/bolt-agent
- start the binary
bolt-agent -apikey changeme
- you may want to run the agent as a systemd service (we have a simple systemd template here)
You need to do that only once:
wget https://raw.githubusercontent.com/bolt-observer/agent/main/bolt-agent.service
cp bolt-agent.service /etc/systemd/system/bolt-agent.service
$EDITOR /etc/systemd/system/bolt-agent.service # to change API_KEY
systemctl daemon-reload
systemctl enable bolt-agent
systemctl start bolt-agent
You can use the docker image from GitHub:
Usage:
docker run -v /tmp:/tmp -e API_KEY=changeme ghcr.io/bolt-observer/agent:v0.2.2
You can limit what channnels are reported using --channel-whitelist option. It specifies a local file path to be used as a whitelist of what channels to report.
Channel IDs are short channel identifiers in LND format (a valid id is 792080480214450176 but not 720393x2228x0)
When adding --private to --channel-whitelist this means every private channel AND whatever is listed in the file. There is also --public to allow all public channels.
Using the options without --channel-whitelist makes no sense since by default all public channels are reported however it has to be explicit with --channel-whitelist in order
to automatically allow all public channels (beside what is allowed through the file).
If you set --private and --public then no matter what you add to the --channel-whitelist file everything will be reported.
The file looks like this:
# Comments start with a # character
# You can list pubkeys for example:
0288037d3f0bdcfb240402b43b80cdc32e41528b3e2ebe05884aff507d71fca71a # bolt.observer
# which means any channel where peer pubkey is this
# or you can specify a specific short channel id e.g.,
759930760125546497
# too, invalid lines like
whatever
# will be ignored (and logged as a warning, aliases also don't work!)
# Validity for channel id is not checked (it just has to be numeric), thus:
1337
# is perfectly valid (altho it won't match and thus allow the reporting of
# any additional channel).
# Empty files means nothing - in whitelist context: do not report anything.
Internally we use:
- nodedata: an abstraction for running the periodic checks and reporting balance related changes
- filter: this is used to filter specific channels on the agent side
- checkermonitoring: is used for reporting metrics via Graphite (not used directly in bolt-agent here)
- lightning: an abstraction around lightning node API (that furthermore heavily depends on common code from lnd)
- lnsocket: a way to establish a commando connection with CLN nodes
- agent: is the GRPC client for reporting payment data
- actions: is the GRPC client for getting what actios to invoke
- raw: is the connection between agent API and lightning API
- plugins: contains the plugins
You might want to read about boltz plugin.
This code depends on some common code.
Read complete documentation on: https://docs.bolt.observer.
Do:
go get -v github.com/lightningnetwork/[email protected]
or else you get an ancient version and everything breaks (should be good due to go.mod/sums now).