search

pelias / logger

The centralized logger package for Pelias.
0 stars 3 forks source link

Add support for logging metrics #29

Open orangejulius opened 7 years ago

orangejulius commented 7 years ago

Right now Pelias logger can only log text. If a user of Pelias wants to log metrics somehow, they have to aggregate those log files and define metrics based on patterns within those log files.

This isn't ideal because the pattern matching is often very tough to get right and debug, it requires more services being run as part of a Pelias setup, and there is no indication in the Pelias code that a seemingly unimportant log line might actually be used to drive a critical metric.

First class support for logging metrics directly, through a configurable backend, should be added. statsd is an obvious backend to support. AWS CloudWatch Metrics might be another one worth supporting.

orangejulius commented 6 years ago

I am considering closing this. After #30, we can now send the logs as JSON, which is quite flexible, and doesn't prescribe any particular metrics/observability platform like statsd, prometheus, etc.

If anyone has any ideas of things we could add that would be generally useful, let us know.

One that might be useful is a detailed status page with metrics. This is technically generic but can be used with Prometheus easily, and other systems with some work.