Third Meeting: Monday, November 23⋅3:00 – 4:00pm UTC

Status update on github discoverability progress

• https://github.com/jackzampolin/registry
• https://github.com/jackzampolin/cosmos-registrar

New Questions

Priority: Give feedback!

TODO: Automating the checkpoints (cron)?

What does it look like to build a node?

• Marko is a good person to ask about this
• Pass in bootstrap repository and chain id
  • Fill out config and fetch the genesis file
  • Could be automated
• Could at a minimum be a list of instructions
• Eventivize has something similar (We could tune, https://github.com/apeunit/LaunchPayload and https://github.com/apeunit/LaunchControlD to make it more universally useful)

What does it look like to build a relayer with this information?

• RPC endpoints are needed for statesync
• In order to enable statesync you need to pass tendermint two RPC endpoints, anyone who wants to run a new node needs to run their own two nodes
• Small critical mass of RPC endpoints and create a culture of people building RPC endpoints
  • ICF?
  • AiB?
  • Chainapsis?
  • Some Validators
  • Chainlayer => Quicksync
    • Still takes ~12 hours
  • Forbole
  • Figment => DataHub
  • Social pressure + small grants
    • Bounty program for uptime
    • Fixed amount distributed to nodes that are up
    • Automated polling
  • Statesync specific endpoint
    • Need RPC for light client
    • Why don’t we expose / integrate the light client into the P2P layer, then you don’t need RPC for light client or state sync
    • Lean archival node
      • Only runs fast sync, always in fast sync mode, fetches new blocks and exposes RPC and only exposes data from the block store (commit, blockhashes) everything needed for a light client
      • DDoS vectors are still there
  • Minimal Viable Node
    • NGINX config
    • Cheapest plan possible
      • 4GB minimum
      • 2CPUs 8GB Ram
      • 500GB Disk
      • $80-$100/mo
        • GCP (Google Compute)
    • Enable State Sync
  • Easy to run solutions
    • EVENTEVIZE EZ Deploy
    • Validator package manager
  • Node service
    • Ping each node
      • 10 blocks from chain tip
      • 500ms latency
      • How much data is it holding
        • Pruning enabled?
    • Keep track of uptime
      • Rank and pay out top 10—100
  • Tm-crawler
    • Exposed ~45 endpoints
    • Begin just by showing the ones already open?
    • TODO: integrate tm-crawler into github?
    • On atlas.cosmos.network?
• People need to regularly run these on cron
  • Find someone from each network to run

TODO:

• CODEOWNERS enforcement
• chainlayer/quicksync — potential user

CAIP issue 27

• Is it necessary to have a fully automated list of asset verification?
  • Alternatively you just limit views of assets you’ve personally verified
• Manual or Automated process will need to have access to all relative blockchain RPC endpoints
  • Alternatively it would have to be network by network via communication channels
• Query verification

Use Case

• Resolving a chain id
• Resolving an asset
• Tied to each other
  • Once you’ve got the asset id you need the chain id
• User perspective
  • The dapp needs the RPC endpoints
• Who should be running the RPC
  • Wallet doesn’t have capacity to be infrastructure for various networks
  • Best interest in applications to provide
  • Wallets can bridge these services

Just saw Lunie sunset their infrastructure

• Can’t sustain running servers

JS light client

• Still needs RPC endpoints

Compensation for node operators

• Pay on delivery
• Pay for storage?
  • IPFS
• Eth land
  • Celo has something similar already

Vulcanize

• Generic IPFS blockexplorer
• Convert blocks to IPLD blocks
• Blockchain specific block explorer based on top of that
• Generated IPLD data for Cosmos SDK apps