merge-devnet-3

[parithosh]

• Decide date for genesis + delay + fork epoch
• Decide on number of validators
• Contract
• Setup config
• Generate mnemonics
• Deploy eth1 miner with high TTD (Ensure the contract address is pre-deployed with the code + in the state)
• Collect enode from miner and pass to config variable
• Generate genesis state
• Get ENR
• PR with the information
• Provision nodes
• Deploy other clients with ENR to get them peering
• Check if genesis root + genesis data matches with client API. It should match what the python script (genesis testnet tool repo) outputs for our genesis.ssz
• Publish list of ENR address of the deployed clients, add as file to the PR
• Setup monitoring and dashboards for the clients
• Setup testnet faucet
• Setup forkmon
• Setup beacon explorer
• Setup eth1 explorer
• Setup public RPC (HAproxy route)
• Document making deposits

Aim of devnet:

• Test transition again, single miner to avoid issues
• Test v3 specs
• Have a new devnet for optimistic sync testing
• Test @skylenet's kubernetes based setup for tooling + minority nodes
• Dry run for public Kintsugi testnet

[parithosh]

All tooling has been setup with @skylenet 's system, it can be found here: https://devnet3.themerge.dev/

Genesis has occurred, TTD is still some time away to give us a chance to test pre-merge. Merge epoch is epoch 20 and hasn't been hit yet.

[skylenet]

PR with the whole code for the k8s infra is in here: ethpandaops/ethereum-k8s-testnets#4

[jgresham]

Doesn't TTD need to be set to 5,000,000,000? It's currently 1,000,000,000. https://github.com/parithosh/consensus-deployment-ansible/blob/master/scripts/quick-run/merge-devnets/merge-devnet-3.vars#L4

[parithosh]

thank you for noticing that, i've merged your PR fixing the issue.

[parithosh]

We had an issue post merge with non finality. The issue was due to non-communication of strict requirements to be a part of the testnet. Due to the non-precise requirements, we had some clients implement api changes that others hadn't merged or had in PRs. This led to an unhealthy network that was fixed by merging the api changes and deploying it.

Currently almost all client combinations work. Prysm has fixed issues with block processing, but we are waiting on some api compatibility changes. Nimbus has interesting issues with geth and the team is investigating it.

Additionally some tooling fixes were applied to beaconchain that fixes missing slots being incorrectly reported. Rafael's k8s based tooling infra seems to be stable and running, we will go ahead with using it for Kintsugi.

Lessons for future tesntets:

• Have defined feature requirements for clients to join devnets

[parithosh]

This devnet has been deprecated and turned off in favor of the public Kintsugi testnet.