[tm2] Node Metrics

[zivkovicmilos]

Description

This effort covers the need to introduce extensive metrics tracking capabilities for critical modules in Tendermint2 / Gno as a whole.

Currently, there is no metrics support in the form of Prometheus or other standards which can be used to continuously monitor active Gno nodes.

A good foundation has been set already with #1762

Suggestions for possible metrics collection areas:

• Consensus

  • Block interval (time between current and prev block)
  • Number of transactions
  • Block gas used
  • Block gas limit
  • Number of validators
  • Number of rounds to reach consensus (implementation specific)

• Networking

  • number of inbound peers
  • number of outbound peers
  • number of pending inbound peers
  • number of pending outbound peers
  • bytes gossiped (ex. for consensus)

• Syncer

  • number of blocks synced
  • number of block txs
  • number of block receipts

• JSON-RPC

  • response time for requests
  • error count when handling request

• Mempool

  • number of free mempool slots used (implementation specific)
  • number of initialized mempool accounts (implementation specific)

• VM

  • gas used per execution
  • number of processed transactions

Successful outcomes of this effort:

• A user can tap into Prometheus / whatever standard metrics for their Gno node
• A user can display these metrics on a dashboard like Grafana
• Ongoing test-nets / main-nets have clear metrics dashboards the core team can use