A simple tool for network debugging.
Each running instance contains an HTTP client and server running on a given port. The client will periodically connect to other running server instances and log whether the attempt was successful and how long it took. The server simply echos whatever it receives.
To use this start a copy up on every host that will be the destination of a test, and specify the remote hosts that you would like it to connect to for testing.
$ networktest host1 host2 host3
There are several ways to record the results of these tests:
- Logging the results to stderr, either a summary or all results (with
-v) - Exporting the results as prometheus metrics (
--prometheus) - Writing the results to a file (
-o)
For more details, see the help.
$ networktest --help
Poll servers on other instances to test the network status.
Usage: main [--interval INTERVAL] [--port PORT] [--resolve] [--verbose] [--log-server] [--log-interval LOG-INTERVAL] [--output OUTPUT] [--prometheus] [--prom-port PROM-PORT] [HOSTNAME [HOSTNAME ...]]
Positional arguments:
HOSTNAME name of each host to test
Options:
--interval INTERVAL, -t INTERVAL
the interval between polling hosts. [default: 5s]
--port PORT, -p PORT the port to operate on. [default: 8090]
--resolve, -r resolve the hostnames to IP addresses at the start.
--verbose, -v log every connection to attempt.
--log-server log connections to the server
--log-interval LOG-INTERVAL
the interval between writing summaries to the log. [default: 30s]
--output OUTPUT, -o OUTPUT
write the test results to the given file
--prometheus export the test results over prometheus
--prom-port PROM-PORT
the port on which to export prometheus metrics. [default: 8091]
--help, -h display this help and exit