Collect the WAN bandwidth metrics for the AT&T U-Verse Pace 5628AC home router.

Publish these metric to an InfluxDB V2 time series database.

Graph these metrics with Grafana to see how much bandwidth you're using on average.

1. Clone this repo into some directory on some server
2. Install the python requirements from requirements.txt
3. Copy example-config.toml to config.toml
4. Configure config.toml with your AT&T router and InfluxDB values
5. Configure cron to call get-att-bw.py every 5 minutes

Cron needs absolute paths for things and I've referred to the config file with a relative path. You can fix this with a cd. The following executes every 5 minutes and writes data to some temp file.

*/5 * * * * cd /your/dir && python3 /your/dir/att-scraper/get-att-bw.py >>
/tmp/att-scraper.log

From: http://192.168.1.254/xslt?PAGE=C_1_0

1. TX Packet Count

2. TX Byte Count

3. TX Packet Error Count

4. TX Packet Error %

5. RX Packet Count

6. RX Byte Count

7. RX Packet Error Count

8. RX Packet Error %

9. Time taken to complete request to router (total_time)

10. Python response.elapsed time (elapsed_time)

From: http://192.168.1.254/xslt?PAGE=C_5_5

10. NAT Connection Count (FUTURE)
11. NAT Connection % Used (FUTURE)

• Automate script run with cron
• Import configuration with TOML (because it's what's cool with kids these days)
• Scrape router HTML page (unathenticated) with requests
• Parse HTML and tables with BeautifulSoup (beautifulsoup4)
• Send data to InfluxDB 2 (Line Protocol because I'm too dumb to make JSON work)
• Graph results with Grafana

• Router IP
• InfluxDB IP
• InfluxDB Port (default 8086)
• InfluxDB Token
• InfluxDB Bucket
• InfluxDB Org
• InfluxDB Measurement Name (suggest net)
• InfluxDB Host Tag
• InfluxDB Region Tag (name of your house or location)

• Does not support modern InfluxDB versions
  • Uses InfluxDB 1.8 client and tested on 1.8 database
  • UPDATE 2023-12-31 - FIXED - Tested with Influx2 client and InfluxDB 2.4.7
• Only tested on Linux systems
• Close to zero error handling

RX Bits Per Second

SELECT non_negative_derivative(mean("rx_bytes"), 1s) *8 FROM "net" WHERE ("host"::tag = 'router') AND $timeFilter GROUP BY time($__interval) fill(null)

TX Bits Per Second

SELECT non_negative_derivative(mean("tx_bytes"), 1s) *8 FROM "net" WHERE ("host"::tag = 'router') AND $timeFilter GROUP BY time($__interval) fill(null)

Will these queries handle counter wrapping? I don't know. They have "non-negative derivative" and that's my attempt to deal with that. I don't know what the math does when the counter wraps around.

I've found it's helpful to also graph the packets per second as an indicator that you might have had a byte counter wrap. It's trivial in grafana to modify the above to tx and rx_pkts and stop multiplying by 8, since you don't need to convert from bytes to bits when you're counting packets.