Lameduck alerts, BigQuery Exporter, and Experiment monitoring

This release includes several significant enhancements to the production prometheus configuration.

1. nodes that are in lame-duck mode are excluded from alerts.
2. the bigquery exporter is now available for production dashboards
3. the blackbox exporter supports health checks used by mlab-ns today
4. the gcp-service-discovery side-car service for prometheus, downloads all the new target files including for the inotify exporter.

Templating and snmp_exporter improvements.

This release includes several refinements and improvements to the new templating system. It also incorporate a number of incremental changes to the snmp_exporter configurations.

Prometheus now monitors the snmp_exporter service itself, and will also trigger an alert when it notices it has been down for at least 10 minutes.

Unify k8s configuration across GCP projects

This release marks the consolidation of per-project k8s configurations into a single k8s config that supports per-project templates variables.

This should make it much easier to track changes from sandbox to staging to production and easier to manage the k8s configurations.

As well this change updates the nagios exporter port from 5000 to the registered port 9267.

Use OAuth2 for Grafana login authentication

This release includes and update to Grafana v4.5.2. With this change we are also using OAuth login exclusively.

This release includes alert changes from #88

This release includes support for snmp-exporter.

Update resource limits for mlab-staging & mlab-oti

Merge pull request #72 from stephen-soltesz/master

Add notes for upgrading resource limits

Alertmanager Github Receiver

Includes the v0.0 github receiver running in mlab-oti.

Initial Release

This is the first version to support autodeploy to mlab-oti from tags.