Health queries (index-node) are failing for an xDai subgraph

[crisog]

Do you want to request a feature or report a bug?
A bug.

What is the current behavior?
Our subgraph (xdai) is synced, but when hitting to check if our subgraph is healthy, it says the opposite (https://api.thegraph.com/index-node/graphql).

If the current behavior is a bug, please provide the steps to reproduce and if possible a minimal demo of the problem.

1. Check our subgraph, confirm it's synced. Subgraph: https://thegraph.com/explorer/subgraph/1hive/uniswap-v2.
2. Send a query to https://api.thegraph.com/index-node/graphql with the following body:

query MyQuery {
  indexingStatusForCurrentVersion(subgraphName: "1hive/uniswap-v2") {
    synced
    health
    chains {
      chainHeadBlock {
        number
      }
      latestBlock {
        number
      }
    }
  }
}

3. You will notice that it shows up as health = failed and numbers on chainHeadBlock and latestBlock are ~300k blocks away.

Also, if you query the subgraph field it says that the current version is QmdUKfpQ7bHGQi4bJU5wLT5s3tviotmRSmiwDtrJ5b8oZ8, when the current version showing at the explorer is QmRETLERrB4Z6AuK3YgHmmfjLSwXX6AWsAARAuz2hnwjEp.

What is the expected behavior?
We should get health = healthy and similar numbers on chainHeadBlock and latestBlock.

[leoyvens]

From inspecting the hosted service DB, this seems to be due to an inconsistency between the explorer and the graph node DB, so possibly not an issue in graph node itself.