README.md

Prometheus Gsheet

Remote write to google sheets

Turns out to work pretty well for a homelab setup.

Docker images available for armv6,armv7,arm64 and amd64 as byrnedo/prometheus-gsheet

Config

• LISTEN - the address to listen to, defaults to :4700
• CREDENTIALS - required: base64 credentials.json for google api
• SPREADSHEET_ID - required: the hash id from the google sheet url (i.e. the last bit from https://docs.google.com/spreadsheets/d/1Wm55yV6TUN74Yjw4ButU52AUUUzA6Xkh2rfN2t0YD3Y)

On the prometheus side, you need to add something like the following to prometheus.yml:

remote_write:
- name: ghseets
  url: http://gsheets:4700/

See the prometheus docs for more config options.

Sheet columns

Are, in order from left to right:

• Consumed timestamp
• Metric timestamp
• Metric name
• Metric value
• Metric dimensions

Example Google Sheet

You can copy the following google sheet and use it as a starting point. MetricsTemplate

Prometheus-gsheet will push metrics to the first sheet of the supplied document.

Alerting

The template sheet includes alerting via slack. You'll need to supply a slack webhook url in order for that to work (in the Config sheet)

Also, the buildReport function in the apps scripts needs a trigger (in this case every 5 minutes):

Custom Sheet Functions

In order to be able to graph and alert on things like cpu, the following sheet functions are available via appsscript:

• TS_GROUPBY - allows grouping by a dimension
• TS_RATE - sort of like rate in prometheus, useful for calculating cpu usage
• TS_DATETIME - maps first column in result from unix timestamp to a date time
• TS_AVERAGE - averages over the timeseries

Most of these functions work with or produce results where the first column is a unix timestamp.

Note: These functions are a bit of a hack to say the least. YMMV.

Roadmap

• Rework apps script
• More example graphs
• Alert via arbitrary url
• Fix TS_RATE function, doesn't seem to handle rollover among other things.