JsonRpcProvider failed to startup; retry in 1s

[szanuje]

Hello,

I have a strange issue with Ethers in my AWS Lambda setup. Basically, I use lambda to get some Price Oracle data and return it.

I Instantiate the provider in such way:

export const provider = new ethers.JsonRpcProvider(
  `https://eth-mainnet.g.alchemy.com/v2/${ALCHEMY_KEY}`,
  ETHEREUM_MAINNET_CHAIN_ID
);

Sometimes when I call a function GET request multiple times at the same time, some of the executions last long and I'm aborting the request and the last log that I see from the Lambda is:

JsonRpcProvider failed to startup; retry in 1s

I don't know why it fails and how to solve that.

EDIT: Adding some more info

Lambda is timing out after 15 min. I see no more logs after the (single) one that I shared.

[mktcode]

Having the same issue with GitHub Actions.

[vladikopl01]

Has the same issue while connecting to Infura ETH Mainet public node

[artemmartyhin]

Same issue here with Filecoin

[ricmoo]

That seems like a strange error to come out of ethers. I don’t think that is something it should log anywhere.

Is there any more details as to what is not working? Does AWS lambda have a custom fetch? Is it using the lib.esm or lib.commonjs? Or is it using one of the dist files?

[artemmartyhin]

I found a reason. My problem is reffered to Filecoin EVM. The JsonRpcProvider is set up after first request and i were sending a get_logs request with 'fromBlock' 0. Ethers v6 did not provide me an error respnse, only code 500. But ethers v5 told me that "bad tipset height: lookbacks of more than 24h0m0s are disallowed". So the request was failed and provider has not been setup. I made a shorter interval for get_logs so now everything works.

So what i suggest is to check the first request that you make from JsonRpcProvider

[ricmoo]

Can you provide me with a URL and the request? That way I can reproduce it and figure out what reflex to add to detect this type of error.

Every network and backend has their own custom error strings, so it’s a constant arms race to add them all. :)

[kongtaoxing]

I got this error, too. I found this error is after this error

TypeError: "listener" argument must be a function
    at at (<anonymous>:25:22615)
    at rt.addListener (<anonymous>:25:25178)
    at http.request (<anonymous>:581:15)
    at https.request (<anonymous>:30:29)
    at getUrl (<anonymous>:1181:88)
    at #send (<anonymous>:1626:26)
    at FetchRequest.send (<anonymous>:1669:24)
    at JsonRpcProvider._send (<anonymous>:13381:38)
    at <anonymous>:12884:41
    at <anonymous>:12902:13

Maybe there are some relationship between them.

[ricmoo]

Can you provide basic reproduction steps?

[Admirated]

I have the same error, it's occured after i try to request with non-valid provider, and ethers want to reconnect this provider -> error in loop

[thisispalash]

I had something similar actually, where my env vars were not being read and hence the rpc was undefined which caused JsonRpcProvider failed to startup; retry in 1s

[algtm]

In my case ,i use tenderly fork rpc and get this error. I found that JsonRpcProvider will try to send eth_chainId and eth_getBlockByNumber together in one request ,which is not supported by tenderly rpc(only single method batches are supported). So I fixed this by disabling batch requests as follows.

    let option: JsonRpcApiProviderOptions = {
        batchMaxCount: 1
    };
    provider = new ethers.JsonRpcProvider(rpc, undefined, option);

[ricmoo]

That’s really helpful to know. Thanks. I may add some logic to the provider then to automatically prevent that for tenderly.

[algtm]

I suggest directly displaying the error message returned by the server to the user instead of showing JsonRpcProvider failed to startup; retry in 1s. I had to use many techniques to successfully intercept the requests sent by Ethers , and then figure out what happened. Moreover, this was a specific case that I encountered, and your fix only solves the issue with Tenderly.

[zakrad]

I had same issue with a private network this solved my issue for ethers v6, thank you so much.

[bogdoslavik]

This helped me. Thanks!

[Inkvisto]

Obvious behavior, but i changed "any" in v6 provider const connector = new ethers.BrowserProvider(provider, "any") to current extension that i use and it's helped(or disable all not used wallet extensions).
If u have many wallet extensions enabled and forget to change network from "any" to one of them you will end up with this error.

[kuzdogan]

I was also having excessive logs of JsonRpcProvider failed to startup; retry in 1s because my code was trying to fetch the contract code without first checking if the Provider is working. When the Provider fails it indeed sends a new request every second. Instead I first. Now I check before with provider._detectNetwork().

Before:

  for (const rpcURL of chain.rpc) {
    try {
      const provider = rpcURL.startsWith('http')
        ? new JsonRpcProvider(rpcURL)
        : new WebSocketProvider(rpcURL);

      // Race the RPC call with a timeout
      const bytecode = await Promise.race([
        provider.getCode(address),
        rejectInMs(RPC_TIMEOUT, rpcURL),
      ]);

      return bytecode;
    } catch (err) {
      console.log(err);
      continue;
   }
}

After:

  for (const rpcURL of chain.rpc) {
    const provider = rpcURL.startsWith('http')
      ?new JsonRpcProvider(rpcURL)
      : new WebSocketProvider(rpcURL);

    // If the provider is not working, skip it
    try {
      await provider._detectNetwork();
    } catch (err) {
      console.log(err)
      continue;
    }
  }

[yarikMyroniuk]

Hi @kuzdogan
I see you solution but the library doesn't stop to spam this log, isn't it?

[kuzdogan]

For me it had stopped