Grafana Dashboard Enhancements
Below are some basic instruction for adding some extra dashboards to the existing Graph dashboards. You might find some of them useful. If you have created more useful dashboard views for Graph metrics, please share them in this document.
The link below is a json for a dashboard with all the panels below. If you prefer to add them separately/manually, you can ignore this and continue through the guide, which contains manual instructions and jsons for each individual panel.
NOTE: You must be running the latest version of the graph-node source code for these metrics to work (as of 11th Sept2020)
These panels present the current blocks behind for each subgraph - one shows the data per subgraph and the other shows the data as a sum of all subgraphs. This can be useful in getting an immediate view on historical indexing performance across your deployment. This data comes from the graph-node metrics endpoint on port 8040.
If you know how to use them you can import these panel jsons (update the datasource to your own), otherwise follow the guides below.
BlocksBehindOverTimeAllSubgraphs
Add a new panel and use your main Prometheus endpoint as the data source.
Set the metric query to deployment_blocks_behind and the visualization type in the panel tab to Graph
In order to create the same panel but as a total for all subgraphs, repeat the above instructions but change the query to sum(deployment_blocks_behind)
Make sure to apply and save your changes.
These panels present the historical performance of blocks processed per minute - one shows the data per subgraph and the other shows the data as a sum of all subgraphs. This can be useful in getting an immediate view on historical indexing performance across your deployment.
If you know how to use them you can import these panel jsons (update the datasource to your own), otherwise follow the guides below.
BlocksProcessedPerMinuteAllSubgraphs
Add a new panel and use your main Prometheus endpoint as the data source.
Set the metric query to increase(deployment_block_processing_duration_sum[1m]) and the visualization type to Graph
If you want some extra summary information in the legend, you can add min, max, avg values in the legend section of the panel tab:
In order to create the same panel but as a total for all subgraphs, repeat the above instruction but change the query to sum(increase(deployment_block_processing_duration_sum[1m]))
Make sure to apply and save your changes.
NOTE: You must be running the latest version of the graph-node source code for these metrics to work (as of 11th Sept2020)
These panels provide some Eth node RPC endpoint performance data from the graph-node metrics endpoint on port 8040.
The panel jsons are below, you will need to update the data sources "datasource": "Prometheus_graph-indexer-0" to your own datasource for your graph nodes.