Better diagnostics for failed GraphQL requests

[lutter]

This code logs information about failed requests. It can happen quite a bit that the failure reason is logged as unknown Restructure the code to provide more detail, in particular, to provide a subgraph id when queries fail in more cases.

[fordN]

The main goal I have for updating this code is to be able to identify issues and resolve them on my indexer.

In the case I'm looking at right now the query_execution_time_count metric is showing many queries with status='failed' and deployment='unknown'. There is a steady stream of these unknown, failing queries and I'd like to figure out where they are coming from to find a solution.

The query body could help dig in deeper, but that's probably overkill. Really I'm most concerned about origin and subgraph API endpoint, so I'd love to be able to figure out what deployment the query was for, and what the origin of the query was.
To that end: is it feasible to reduce the instances where deployment = unknown? Is it really coming in without a deployment id, subgraph id, or subgraph name? Or include some logging earlier on in the query process that includes deployment/subgraph and use some common query id to link them together?

[github-actions[bot]]

Looks like this issue has been open for 6 months with no activity. Is it still relevant? If not, please remember to close it.