Consider health information

[EnricoMi]

Each partition could fetch latest health information to identify alphas that should not be used. This avoids connecting to un-healthy alpha nodes while reading partition data. Health information could be fetched once by the driver (when creating the DataFrame) or for each partition when it starts reading.

The health information also provides the Dgraph version of the alpha nodes, which could be used to make version-dependent features and code paths in the connector more explicit.

[
  {
    "instance": "alpha",
    "address": "localhost:7080",
    "status": "healthy",
    "group": "1",
    "version": "v20.03.0",
    "uptime": 635,
    "lastEcho": 1638524298
  }
]

https://dgraph.io/docs/deploy/dgraph-alpha/