Tracking Metrics to Assess Retrieval Network Health

[davidd8]

• Overall Boost Retrieval Volume - We already track this weekly in Bedrock, and it did show a significant drop from last week --> track daily
  • Breakdown volume of requests from Gateway and Autoretrieve sources
  • [Nice-to-have] Further Breakdown based on the transport protocol
• Overall Retrieval Success Rate - We track this from autoretrieve weekly in Bedrock, and it also showed a significant drop from last week --> track daily
  • Breakdown volume of requests from Gateway and Autoretrieve sources
  • [Nice-to-have] Further Breakdown based on the transport protocol
• [New] Distinct Number of SPs serving a successful retrieval - A quantitative indicator of SP sentiment in serving retrievals, to go along with slack DMs
  • Breakdown volume of requests from Gateway and Autoretrieve sources
  • [Nice-to-have] Further Breakdown based on the transport protocol
• [New] Percentage of Lassie retrievals served by SPs - Answers the Rhea goal of serving Filecoin content from SPs
• [Nice-To-Have] - Distribution of the volume of Lassie retrieval requests per SP - Helps us understand how evenly distributed the gateway traffic is w.r.t. to SPs

[davidd8]

cc @LaurenSpiegel

[LaurenSpiegel]

@davidd8 , would help to clarify which are rhea v autoretrieve and which are combined metrics. right now we do have visibility into daily volume and success rate but just for autoretrieve. It would be helpful to see volume and success rate from rhea versus these from autoretrieve.

[davidd8]

Updated issue to add those breakdowns.

[davidd8]

@hannahhoward @kylehuntsman PTAL as we add the metrics pipeline to track these metrics.

[davidd8]

@hannahhoward can we mark this as done given the updates to the most recent grafana dashboard?

[hannahhoward]

yes