Fix meaningless service mesh metrics in none mesh environment #13615
Conversation
| service_sidecar_internal_req_latency_nanos = from(Service.sideCar.internalRequestLatencyNanos).filter(sideCar != null).longAvg(); | ||
| service_sidecar_internal_resp_latency_nanos = from(Service.sideCar.internalResponseLatencyNanos).filter(sideCar != null).longAvg(); |
There was a problem hiding this comment.
There are mesh OAL only, why do we filter this?
There was a problem hiding this comment.
If you don't need mesh, you don't have to activate these OALs.
There was a problem hiding this comment.
I think the out-of-the-box state is better: if users don't use mesh, mesh metrics are not generated; if users use mesh, mesh metrics are generated. Changing configuration files may introduce more works for user.
There was a problem hiding this comment.
Generally, users are not encouraged to activate all analyses, as, it actually has a price(extra and unnecessary resources) to pay.
Please follow this best practice. The extra resource costs(CPU, memory) are much more expensive in the product env, compared with a config to be set.
2 participants
Fix meaningless service mesh metrics in none mesh environment
CHANGESlog.In no sidecar environment, service mesh metrics should not be generated.