EIP-6110 Stop Eth1 data polling

[StefanBratanov]

PR Description

If we have already transitioned to the new deposit mechanism:

• Don't log missing deposit data
• Don't do actual Eth1Data vote, just return the one from the state
• Don't update Eth1Data metrics
• Don't initialize and start PowchainService services
• Implement stop for AsyncRunLoop because TimeBasedEth1HeadTracker was running in loop and stop was not implemented

Also attached FinalizedCheckpointChannel to PowchainService to stop Eth1 polling when transitioned on the fly. I am not 100 % sure of the cleanest way to pass a state to PowchainService because RecentChainData is initialized after starting of the services, so passing it before that will be in null. That's why chose to pass a Supplier (along with some comments) but happy for suggestions.

Update:
Refactored PowchainService to initialize when starting. This has couple of benefits. No initialization needed when we have already finalized the transition to the new deposit mechanism. This allows thing like using the websockets web3j client after Electra instead of relying on http, so no more of this error:

if (executionWeb3jClientProvider.getWeb3JClient().isWebsocketsClient()) {
   throw new InvalidConfigurationException("Eth1 endpoint fallback is not compatible with Websockets execution engine endpoint");
 }

It also allows to log the eth1 polling disabling only in case it has been enabled before.

Fixed Issue(s)

will solve one of the tasks under #7898

Documentation

• I thought about documentation and added the doc-change-required label to this PR if updates are required.

Changelog

• I thought about adding a changelog entry, and added one if I deemed necessary.

[zilm13]

Do I understand it correctly that since EIP-6110 we don't need DepositTree and snapshots any more?
Though I think we need to approve it I still have some concerns on transition moment, that we are not breaking something too early when we are in Electra but former mechanism is not yet disabled. Anyway, we will run testnets for this.

[zilm13]

I'd test something new here. At least that we start preElectra and don't start postElectra

[StefanBratanov]

Good point actually. Added a couple of relevant tests.

[zilm13]

I'd test also that getDeposits() is empty since former mechanism is disabled

[StefanBratanov]

This was part of a previous PR, can have a look at noDepositsIncludedIfFormerDepositMechanismHasBeenDisabled and getsRemainingEth1PendingDepositsIfElectraIsEnabled

[zilm13]

ahhh the one I wanted comment for
All good, all done already

[StefanBratanov]

Do I understand it correctly that since EIP-6110 we don't need DepositTree and snapshots any more? Though I think we need to approve it I still have some concerns on transition moment, that we are not breaking something too early when we are in Electra but former mechanism is not yet disabled. Anyway, we will run testnets for this.

Yes, once we have finalized state where eth1_deposit_index == deposit_receipts_start_index we no longer need the DepositTree and snapshots since the getDeposits method would always be empty list from that point onwards, so no merkle proofs needed. So the idea here is to stop Eth1 polling once we finalize such a state and also don't even initialize if the anchor point on startup is such a state.

Also when syncing we don't need the deposit tree, so historical sync will still work. Essentially we only need the proofs when proposing in order to include it in the block.

[zilm13]

LGTM, just final nit

[zilm13]

I'd add assertThat(powchainService.isRunning()).isTrue(); so it's clear that it's stopped only after finalized checkpoint. I know that it's clear from Service code, just the test will be more clear I think

[StefanBratanov]

Makes sense. It's more clear I agree.