cluster: Add option to configure ng-monitoring continuous_profiling

[dveeden]

Feature Request

When using TiUP Cluster to deploy a TiDB Cluster one can set the ng_port to get ng-monitoring deployed as part of the cluster. Then one can enable continuous profiling as described here. However to automate deployment and make this persist on restarts of the cluster and cluster components it would be good to configure this in the configuration in the same way other cluster config is handled.

For ng-monitoring:

• There is a continuous_profiling section in the config.
• In this section one can enable continuous profiling and set a retention time.

For tiup:

• The template for ngmonitoring.toml doesn't have a continuous_profiling section.
• There doesn't seem to be a section for continuous profiling in NgMonitoringConfig

The workaround for now is to:

• Manually enable continuous profiling via the dashboard
• Send a HTTP POST to the dashboard or ng-monitoring service
• Possibly do this after cluster and/or component restarts.

$ curl -s -X POST  -d '{"continuous_profiling":{"enable":true, "data_retention_seconds":259200}}' http://127.0.0.1:12020/config | jq .
{
  "status": "ok"
}

[dveeden]

/category monitoring /component cluster

[dveeden]

/component cluster