Node informant

Node informant actually consists of two tools. neighbour-discovery and gluon-collector.

This is a little utility to continuously request data from announced enabled nodes. The experimental go vendor feature is used to manage dependecies, so you need to set GO15VENDOREXPERIMENT=1 to use the vendored dependencies. To manage dependencies govendor is recommended (and used in this project).

How to build

go get github.com/ffdo/node-informant/...

neighbour-discovery

This tool can act as a replacement for the gluon-neighbour-discovery tool. It has the following command line switches:

Switch	Description	Default	Mandatory
-iface	The interface to use to send packets from and receive packets on	none	Yes
-query	The query to send i.e. "GET nodeinfos"	none	Yes
-deflate	Whether to decompress received data via deflate	false	No
-port	The port to bind to, to receive packets on	12444	No
-timeout	After how many seconds the program should terminate. -1 to keep it running indefinitely.	-1	No
-target	If a target IPv6 address is specified, the query is send via unicast to this target	none	No

gluon-collector

gluon-collector should run in the background. It queries in regular intervals all nodes listening on the default announced multicast group and stores the received information. The received information is available via a REST API or data prepared for meshviewer.

Command line switches

Switch	Description	Default	Mandatory
-config	The path to a valid yaml or json config	/etc/node-collector.yaml	No
-import	Import data from this path. The type of data depends on the import type	none	No
-importType	Specify the type of data to import. Currently only ffmap-backend is supported	ffmap-backend	No

Please not that it is not advised to add the import flags to the default startup config, since this would import the legacy data on every startup, effectively overwriting previously collected data.

Example config

receiver:                 # List of receiver receiving informantion from nodes.     
- type: announced         # Type of the receiver. Currently only announced is supported
  interface: "bat0"       # The interface to use for announced
  port: 21444             # The port to use as a source port announced requests and to listen for responses on

interval:
  statistics: 300         # The interval in seconds to fetch fast changing data like statistics and neighbours
  nodeinfo: 1800          # The interval in seconds to request more static data and discover new nodes
  expire: 3               # This is a multiplicator for the statistics interval. A node is considered offline if
                          # statistics interval multiplied by expire seconds have passed since the last response

logger:     
  level: "warn"           # The log level, see logrus for valid values
  file: /var/log/gluon-collector.log  # If the log file is specified the log is written there. If not everything is send to stdout.

store:
  type: "bolt"            # The type of data store to use. Currently bolt (persistend) and memory (non persistend) are supported
  path: "/opt/gluon-collector/collector.db" # The path is only relevant for bolt store. Where to store the database?
  expireNodesAfterDays: 365 # After this amount of days, a node is considered gone and is deleted from the database

http:             
  port: 8079              # The port where the http server will listen on.
  address: "[::]"         # Optional listen address if you want the server to listen only on a specific interface

prometheus:
  namelabel: true         # Label prometheus node statistics with the host name
  sitecodelabel: true     # Label prometheus node statistics with the received site code

HTTP API

The following rest endpoints are available. All endpoints return JSON (or JSON arrays) and CORS headers. The only valid method is GET.

Endpoint	Description
/nodes.json	Generates a valid nodes.json for meshviewer
/graph.json	Generates valid graph data for meshviewer
/nodeinfos/{nodeid}	Retrieves general node information about the node with nodeid
/nodeinfos	Retrieve all available general node information
/statistics/{nodeid}	Retrieve statistics for node
/statistics	Retrieve all available statistics
/neighbours/{nodeid}	Retrieve mesh neighbour information about node
/neighbours	Retrieve all available neighbour information
/nodestatus/{nodeid}	Retrieve status information like Lastseen, Online status etc. for node
/nodestatus	Retrieve all available status information

Prometheus

Under the /metrics endpoint prometheus metrics are exposed. Currently we expose the following metrics

Metric	Description
total_clients	Total count of clients connected to all nodes
meshnodes_total	Count of all known mesh nodes online and offline
total_traffic_rx	Accumulated on mesh nodes received traffic
total_traffic_tx	Accumulated from mesh nodes transmitted traffic
total_traffic_mgmt_rx	Accumulated on mesh nodes received management traffic
total_traffic_mgmt_tx	Accumulated from mesh nodes transmitted management traffic
meshnodes_online_total	Count of currently available mesh nodes
meshnode_traffic_rx	Received traffic on every node labeled with the nodeid and traffic type
meshnode_traffic_tx	Transmitted traffic from every mesh node labeled with the nodeid and traffic type
meshnode_uptime	Uptime of single mesh nodes labeled with the nodeid
meshnode_clients	Client count on mesh nodes labeled with the nodeid

ffdo / node-informant

readme