4.0.0-rc2: gelf logging is busted when using dns pointing to both IPv4 and IPv6 addresses

[matthewdarwin]

Upgrading a nodeos 3.2.2 node to 4.0.0-rc2 and enabling the prometheus plugin, the gelf logging shows now output with

$ tcpdump udp port 12201

Probably prometheus has no impact, but that was the only config change made.

logging.json was unchanged, which in part has this:

    {
      "name": "net",
      "type": "gelf",
      "args": {
        "endpoint": "prod-server:12201",
        "host": "jungle4-bp203",
        "_operator": "eosnationftw",
        "_network": "jungle4"
      },
      "enabled": true
    },
    {
      "name": "test",
      "type": "gelf",
      "args": {
        "endpoint": "test-server:12201",
        "host": "jungle4-bp203",
        "_operator": "eosnationftw",
        "_network": "jungle4"
      },
      "enabled": true
    }

...

    {
      "name": "transaction_failure_tracing",
      "level": "debug",
      "enabled": true,
      "additivity": false,
      "appenders": [
        "net",
        "test"
      ]
    },
    {
      "name": "transaction_success_tracing",
      "level": "debug",
      "enabled": true,
      "additivity": false,
      "appenders": [
        "net",
        "test"
      ]
    }

[matthewdarwin]

Works with IP address, but not DNS which resolves to both IPv4 and IPv6 address.

[heifner]

Need to remove the use of libfc udp_socket which is hard coded to use IPv4 only. gelf_appender should use boost directly. That should have been done as part of moving gelf to another thread: #650 (comment)

[spoonincode]

yes that clean up should have been done, but worst case to resolve this issue quickly in 4.0 can simply change the resolver to v4 only.