feat(couch-backlog): multi-couch2pg isntances on 1 db, update fake-cht, add prometheus env vars

[mrjones-plip]

Turns out that couch2pg doesn't need extra config to know what CHT instance it's gathering stats from. With this in mind we can both simplify the config as well support multi-couch2pg instances writing the same database.

This PR:

• updates couch2pg backlog SQL to use new query which also gathers CHT Instance
• updates grafana dashboard to use CHT instance from query instead of from sql_instances.yml
• update fake-cht to use new query so it works with it's internal postgres instance - this includes updating fake compose file to use new service name (postgres -> fake-postgres). On the off chance you're running couch2pg in the same docker network you have two postgres service and, just like in highlander, there can be only one!
• adds two new environment variables to the main compose file PROMETHEUS_BIND and PROMETHEUS_PORT. This should be backwards compatible as it defaults to binding to 127.0.0.1. But with these .env values below, you can easily run a debug setup by exposing all critical services on your bare-metal host. If it your host was IP 192.168.68.26, then you could do this:

  SQL_EXPORTER_IP=192.168.68.26
  DATA_INGEST_IP=192.168.68.26
  PROMETHEUS_BIND=192.168.68.26
  GRAFANA_BIND=192.168.68.26

See also

• update related docs

For test steps, see main ticket.

per #107

[jkuester]

So sorry for the delayed review here! :sweat: I plan to get to this today or tomorrow at the latest.

[mrjones-plip]

Nice! Tested out the fake setup and couch2pg data is getting pulled successfully.

One additional thing I did notice is that the apply/revert-dev-patches scripts are giving an error since they are trying to patch an old postgres config file. I think it should be sufficient to just delete the development/patches/apply/exporters.postgres.config.scrape_config.yml.patch file (and the opposite one in the revert directory. Those patches were just setting the scrape_interval to a lower value. Now it looks like we are just inheriting the default scrape_interval from the prometheus.yml so there is no need to patch anything related to postgres....

Ah - thank you!! I saw errors earlier on when applying the patches and should have remembered to update this. fixed in 05266d1

Also - kudos on building an easy to maintain patch system - that was easy peasy!

[medic-ci]

:tada: This PR is included in version 1.14.0 :tada:

The release is available on GitHub release

Your semantic-release bot :package::rocket: