FR: facilitate debugging of "could not detect network (event="noNetwork", code=NETWORK_ERROR..."

Errors of the form Error: could not detect network (event="noNetwork", code=NETWORK_ERROR, version=providers/x.y.z) seem to crop up quite a bit, e.g.:

https://github.com/ethers-io/ethers.js/issues?q=is%3Aissue+event%3D"noNetwork"++code%3DNETWORK_ERROR

There appears to be no indication in the error of what went wrong, or in the documentation about how to debug this. One cause appears to be non-functional RPC endpoints, but this is certainly not the only one. For example, I am seeing this from hre.ethers.provider provided by hardhat-ethers with an RPC endpoint URL which works perfectly fine via new ethers.providers.JsonRpcProvider(url).

So I would suggest any/all of the following, in order to facilitate root cause analysis:

• If possible, add more detail to the error message.
• If not, provide some kind of tracing of network activity (e.g. export ETHERS_TRACE=network or similar).
• Add some suggested trouble-shooting procedures to the docs.

[ricmoo]

The will help if the problem is the node is erroneously returning invalid responses. But for most other cases you won’t see much, unfortunately. 😦

[ricmoo]

This is usually a link error, which means there is not really any meaningful way to get more info. If it is more than a link error, the provider.on("debug", console.log) can probably help.

Otherwise, link errors can occur because the node isn’t up and running yet, there is a firewall in the way (I.E mobile platforms like iOS and android block all http traffic), the geth instance hasn’t been configured to allow a connection over the default interface, etc.

There is not generally much that can be done at an application level to debug a bad network link as much of that is at the OS level, server configuration or network topology. 😒