grafana_mongodb_dashboards

Grafana dashboard templates for use with the prometheus_mongodb_exporter for MongoDB w/Prometheus

Features

• Cluster Summary, Sharding, Replica Set and single-instance dashboards for Percona Server MongoDB / MongoDB
• Grouping of OS metrics with MongoDB instances
• Cluster, replica set and node member discovery via Prometheus data
• WiredTiger storage engine metrics (beta/experimental)

Required

1. A Prometheus server (https://prometheus.io/)
2. A Grafana 2.0+ server (http://grafana.org/)
3. The prometheus_mongodb_exporter for Prometheus on at least one Percona Server MongoDB or MongoDB instance
4. The node_exporter for Prometheus on each node running the prometheus_mongodb_exporter
5. "Target Group" labels in Prometheus server configuration file (See "Prometheus Configuration")
6. Prometheus is added as a Data Source to Grafana with the Data Source name "Prometheus" (otherwise you need to change the "datasource" field in the templates)

Prometheus Configuration

The grouping of metrics of the OS (node_exporter) and MongoDB (prometheus_mongodb_exporter) is achieved by using labels applied to Prometheus Target Groups in the Prometheus server config file (prometheus.yml).

Each MongoDB instance to be monitored must be added as a new group in the prometheus.yml file with both the 'node' and 'mongodb' exporter as 'targets'.

Each MongoDB instance added must also have the following labels added to it's target group:

1. 'cluster' - A human-defined name for the cluster the node is part of. required
2. 'alias' - A field containing the MongoDB hostname/port combination for the instance. required
3. 'nodetype' - A field for grouping node types: 'mongod' for standalone/replset instances, 'config' for shard config servers and 'mongos' for shard mongos instances. required
4. 'replset' - An field for instances that are members of a MongoDB Replication Set. This must match the replset name seen in the MongoDB command 'rs.status()'. optional

Example target section for a replset member of "shard12" (port 9140 = prometheus_mongodb_exporter/port 9100 = node_exporter):

      - targets: ['tyrion.westeros.com:9140','tyrion.westeros.com:9100']
        labels:
          alias: 'tyrion.westeros.com:27017'
          nodetype: 'mongod'
          replset: 'shard12'
          cluster: 'prodCluster'

Example target section for a mongos member (notice no 'replset' label because it is mongos):

      - targets: ['arya.westeros.com:9140','arya.westeros.com:9100']
        labels:
          alias: 'arya.westeros.com:27018'
          nodetype: 'mongos'
          cluster: 'prodCluster'

Installation

1. Install and start Prometheus and Grafana 2.0+ servers with default configuration (not explained here)
2. Configure Grafana to use Prometheus as a Data Source. Link: Grafana Support for Prometheus: Using
3. Install prometheus_mongodb_exporter on all nodes that run MongoDB. Link: Readme
4. Install node_exporter on all nodes that run the prometheus_mongodb_exporter. Link: Readme
5. For each instance to monitor, add 'targets' and 'labels' to the prometheus.yml file as described in 'Prometheus Configuration' section above
6. Reload Prometheus configuration file ("kill -HUP PID") or restart the Prometheus process/service
7. Import each Grafana template file from the 'dashboards' subdir into Grafana's UI and press 'Save' after each import. Link: Import/Export Templates (Grafana Docs)

Roadmap

• Add storage engine metrics for PerconaFT and RocksDB, when they're available in the exporter
• Automate/rethink target 'labels' in prometheus.yml (used for grouping metrics in graphs)
• Version scheme

Contact

• David Murphy - Twitter / Github / Email
• Tim Vaillancourt - Github / Email
• Percona - Twitter / Contact Page