The “telemetry” error is common and can be ignored. There is a startup flag to send anonymous usage statistics to Dgraph (which your system has on), but the server that accepts that info has not been running for some time, so the requests time out. “Deadline exceeded” is the message for a timeout, per the GRPC spec, and it just means the connection timed out.
I recommend turning on request logging: Logging - Deploy to see what errors correspond to the failed query. If that does not clear it up, use Jaeger to view internal trace/span data per Tracing - Deploy .