Ethers.js: Issues: "failed to meet quorum" on homestead network only.

[kaizvn]

@ricmoo I'm still having this issue after fresh install version 5.0.12.
It's no issue happens if I add an event listener on rinkeby network but homestead got this error.

filters :

[homestead] register listenner: [ partnerowner1 : 0xBb993EBccB72F523f068220Ed557aDAE3CdA2cd4 ]

filters {
  address: '0xdac17f958d2ee523a2206206994597c13d831ec7',
  topics: [
    '0xddf252ad1be2c89b69c2b068fc378daa952ba7f163c4a11628f55a4df523b3ef',
    null,
    '0x000000000000000000000000bb993ebccb72f523f068220ed557adae3cda2cd4'
  ]
}

and this is what I got:

Error: failed to meet quorum (method="getBlockNumber", params={}, results=[{"weight":1,"start":1603061155831,"result":11078005},{"weight":1,"start":1603061155831,"result":11082697},{"weight":1,"start":1603061156585,"result":11082697},{"weight":1,"start":1603061156585,"result":11082697}], provider="[object Object]", code=SERVER_ERROR, version=providers/5.0.12)

additional log

 {
  reason: 'failed to meet quorum',
  code: 'SERVER_ERROR',
  method: 'getBlockNumber',
  params: {},
  results: [
    { weight: 1, start: 1603061697719, result: 11078005 },
    { weight: 1, start: 1603061697719, result: 11082737 },
    { weight: 1, start: 1603061698474, result: 11082737 },
    { weight: 1, start: 1603061698474, result: 11082737 }
  ],
  provider: FallbackProvider {
    _isProvider: true,
    _events: [ [Event] ],
    _emitted: { block: 11082736 },
    formatter: Formatter { formats: [Object] },
    anyNetwork: false,
    _network: {
      name: 'homestead',
      chainId: 1,
      ensAddress: '0x00000000000C2E074eC69A0dFb2997BA6C7d2e1e',
      _defaultProvider: [Function]
    },
    _maxInternalBlockNumber: 11082736,
    _lastBlockNumber: 11082736,
    _pollingInterval: 4000,
    _fastQueryDate: 1603061693718,
    providerConfigs: [ [Object], [Object], [Object], [Object] ],
    quorum: 2,
    _highestBlockNumber: 11082736,
    _internalBlockNumber: Promise { <rejected> [Circular] },
    _poller: Timeout {
      _idleTimeout: 4000,
      _idlePrev: [TimersList],
      _idleNext: [Timeout],
      _idleStart: 28586,
      _onTimeout: [Function: bound ],
      _timerArgs: undefined,
      _repeat: 4000,
      _destroyed: false,
      [Symbol(refed)]: true,
      [Symbol(kHasPrimitive)]: false,
      [Symbol(asyncId)]: 438,
      [Symbol(triggerId)]: 0
    },
    _bootstrapPoll: null,
    _fastBlockNumber: 11082736,
    _fastBlockNumberPromise: Promise { 11082736 }
  }

[ricmoo]

That looks like quorum should certainly have been met. I’ll look into this shortly.

[kaizvn]

@ricmoo I found this block of code that may cause this error.
I read your code and workaround by add { quorum: 1 } into option and the error it's gone.

[ricmoo]

That should be fine. You want to use a quorum of at least 2 in production.

I think there must be some logic issue on the custom blockNumber validation rules. It’s been a while since I have dug into that though, so I’ll set up some reproduction scripts later.

[kaizvn]

at least it fixed my issue, the error hurt my logs.
I'm so sorry that I don't have much experience on concept of quorum, so it's not easy for me to help about this issue.
Hope you can find the root cause and fix it soon!

[ricmoo]

No worries, I’ll look into it. I am planing to re-factor the FallbackProvider overall.

Reducing it to 1 will eliminate the error in you code, but reduces the security and resilience. If any backend is out of sync or has been compromised, your app may behave incorrectly...

[kaizvn]

No worries, I’ll look into it. I am planing to re-factor the FallbackProvider overall.

Reducing it to 1 will eliminate the error in you code, but reduces the security and resilience. If any backend is out of sync or has been compromised, your app may behave incorrectly...

I have no choice in this case until have your fix. Actually may I increase it more than 3 or something else? Honestly I don't understand what quorum does there. If you have time please lighten me up, thanks!

[ricmoo]

Increasing it beyond the sum of providers / 2 could start to impact its quorum error rate.

Basically, the goal of the FallbackProvider is to create a provider with many backends.

The defaultProvider creates a FallbackProvider backed by multiple third-party backends (such as INFURA and Alchemy). But since using a third-party service means you must trust that service, multiple backends are queried at random, to ensure a quorum of them agree. That way if a backend is lying (through malice or being out-of-sync) it will be “out-voted” by honest, synced nodes. If a quorum is not reached initially, additional nodes are queries until a quorum is reached.

So, if you use eth_call to lookup an ENS name and the first node says address A and the second address B, then a third will be queried. If it answers with address B, the default quorum has been reached and address B is returned.

So, but specifying a quorum of 1, you are saying you will accept any answer that comes back and that you trust any backend. If you do this, I recommend pulling the CloudflareProvider out of the list of providers as it is regularly out-of-sync. I might make this the default behaviour; if a person is selecting a quorum of 1, remove Cloudflare. I have had relatively issue with any of the other third-party services.

Hope that helps a bit. :)

[kaizvn]

thanks for your explaination @ricmoo , very clear!
Currently i'm starting getDefaultProvider with 3 configs of INFURA, ALCHEMY and ETHERSCAN. I'll recheck if CloudflareProvider was created regardless the option didn't mention it, I'll remove it manually.
Btw, if I have time I gonna help you a little bit by adding function to query provider, now if I wanted to get etherscanProvider, I have to look up like this :

esProvider  = ProviderInstance.Providersconfigs.find(provider => provider.provider.apiKey === ETHERSCAN_API_KEY);

It was broken my code once since I ran version 4 and apiKey was a randomKey instead of mine, then the function getHistory stop working (esProvider === null)