When running Loki components deployed using loki-operator some of them are currently reporting an error like this every few seconds:

level=info ts=2022-02-24T16:29:56.763760598Z caller=dns_resolver.go:209 msg="failed DNS SRV record lookup" err="lookup _grpclb._tcp.loki-query-frontend-grpc-lokistack-dev.default.svc.cluster.local on 10.96.0.10:53: no such host"

This is because the SRV lookup is looking for a "grpclb" service but our Kubernetes Services define a "grpc" service instead (see here).

From what I can see the lookup uses the A record available on the k8s Service DNS address as a fallback.