Skip to content

Latest commit

 

History

History
111 lines (80 loc) · 3.8 KB

README.md

File metadata and controls

111 lines (80 loc) · 3.8 KB
Raw

go-tcp-proxy

A simple TCP proxy that exposes network telemetry metrics via Prometheus instrumentation.

See exposed telemetry metrics for a list of metrics exposed by the proxy.

Installation

go get github.com/austingebauer/go-tcp-metrics-proxy
cd go-tcp-metrics-proxy
make

An executable named proxy will be placed into the bin directory after the make target completes.

Usage

go-tcp-metrics-proxy is configured using command line arguments.

The example below shows the usage of go-tcp-metrics-proxy using netcat as a client and server.

1. Start listening for TCP connections

nc -k -l 127.0.0.1 3001

2. Start the TCP proxy

./bin/proxy -listen="127.0.0.1:3000" -target="127.0.0.1:3001" -metrics="127.0.0.1:3002"

3. Connect to the TCP proxy

nc 127.0.0.1 3000

4. View Prometheus Telemetry Metrics

curl http://localhost:3002/metrics

5. Send bytes from client to server

In the window running the netcat TCP client program, type the following followed by a carriage return:

Hello, server!

Observe that the message has been arrived on the netcat TCP server as terminal output.

Sending bytes from server to client works the same way, so give that a try as well.

6. View Prometheus Telemetry Metrics.. Again!

curl http://localhost:3000/metrics

Observe that metrics related to TCP connections handled by the proxy have been updated.

7. Send SIGTERM to the proxy

Observe that the proxy will gracefully terminate when a SIGTERM has been received by draining all existing connections until they're finished.

8. Send SIGTERM to TCP client or server

Observe that the proxy will finally exit after either the client or server closes its end of the connection.

Telemetry Metrics Exposed

The following is a list of telemetry metrics exposed by the proxy in prometheus text-based format

Metrics related to Go have been omitted from the /metrics results below in order to showcase the TCP proxy related metrics.

# HELP active_inbound_connections The number of currently active inbound connections
# TYPE active_inbound_connections gauge
active_inbound_connections{id="75fc83c4-2109-4757-8660-896c170303c3"} 0
# HELP active_outbound_connections The number of currently active outbound connections
# TYPE active_outbound_connections gauge
active_outbound_connections{id="75fc83c4-2109-4757-8660-896c170303c3"} 0
# HELP inbound_bytes_count The total number of bytes sent and received on inbound connections
# TYPE inbound_bytes_count counter
inbound_bytes_count{id="75fc83c4-2109-4757-8660-896c170303c3"} 15
# HELP inbound_connection_count The total number of inbound connections established
# TYPE inbound_connection_count counter
inbound_connection_count{id="75fc83c4-2109-4757-8660-896c170303c3"} 1
# HELP outbound_bytes_count The total number of bytes sent and received on outbound connections
# TYPE outbound_bytes_count counter
outbound_bytes_count{id="75fc83c4-2109-4757-8660-896c170303c3"} 15
# HELP outbound_connection_count The total number of outbound connections established
# TYPE outbound_connection_count counter
outbound_connection_count{id="75fc83c4-2109-4757-8660-896c170303c3"} 1
# HELP promhttp_metric_handler_requests_in_flight Current number of scrapes being served.
# TYPE promhttp_metric_handler_requests_in_flight gauge
promhttp_metric_handler_requests_in_flight 1
# HELP promhttp_metric_handler_requests_total Total number of scrapes by HTTP status code.
# TYPE promhttp_metric_handler_requests_total counter
promhttp_metric_handler_requests_total{code="200"} 3
promhttp_metric_handler_requests_total{code="500"} 0
promhttp_metric_handler_requests_total{code="503"} 0