Logs Index RPC API

[shemnon]

Add a RPC API to create the logs index.

Example usage:

curl -H "Content-Type: application/json" \
  -X POST \
  --data '{"jsonrpc":"2.0","method":"admin_generateLogBloomCache", "params":["0x0", "0x10000"], "id":1}' \
  <node RPC port>

Signed-off-by: Danno Ferrin <danno.ferrin@gmail.com>

PR description

Fixed Issue(s)

[shemnon]

Some more examples
query the indexing state, no params

$ curl -H "Content-Type: application/json" -X POST --data '{"jsonrpc":"2.0","method":"admin_generateLogBloomCache", "params":[], "id":1}' localhost:8555
{
  "jsonrpc" : "2.0",
  "id" : 1,
  "result" : {
    "startBlock" : "0x0",
    "endBlock" : "0x0",
    "currentBlock" : "0x0",
    "indexing" : false,
    "requestAccepted" : false
  }
}

Start indexing from 0 to chain head

$ curl -H "Content-Type: application/json" -X POST --data '{"jsonrpc":"2.0","method":"admin_generateLogBloomCache", "params":["0x0"], "id":1}' localhost:8555
{
  "jsonrpc" : "2.0",
  "id" : 1,
  "result" : {
    "startBlock" : "0x0",
    "endBlock" : "latest",
    "currentBlock" : "0x0",
    "indexing" : true,
    "requestAccepted" : true
  }
}

request is ignored if indexing is already occurring

$ curl -H "Content-Type: application/json" -X POST --data '{"jsonrpc":"2.0","method":"admin_generateLogBloomCache", "params":["0x0"], "id":1}' localhost:8555
{
  "jsonrpc" : "2.0",
  "id" : 1,
  "result" : {
    "startBlock" : "0x0",
    "endBlock" : "latest",
    "currentBlock" : "0x738c",
    "indexing" : true,
    "requestAccepted" : false
  }
}

[CjHare]

The changes look good (although spotless is currently complaining) ...but I notice tthe absence of tests, is the logic tested elsewhere (already covered / follow up PR)?

[shemnon]

That's why it's a draft PR. ;)

[shemnon]

Tests for the cache generation and querying were in the initial checkin last week. This focuses on the RPC exposure of the GenerateLogBloomCache and refactors the CLI, so the new test validates we get what we want out of the RPC call into the other systems.

[CjHare]

nit; might a DEBUG or TRACE level log message be useful in this InterruptedException case, given it is around indexing?

[shemnon]

The only place it would be seen is when attempting to acquire the lock, and we wouldn't re-try if we got interrupted waiting for the lock, we would fail just like a timeout occured. So I don't see logging as adding any value.