Per-operation and client error metric

[bonnici]

This adds a new metric that tracks errors per operation, client, and error code. The metric in incremented whenever an error occurs and includes the following attributes:

• client name
• client version
• apollo operation ID
• error code
• operation name
• operation type (query/mutation/subscription)

This new metric is marked as experimental and disabled by default, since using it in a high-cardinality graph will likely mean that we'll hit OTel cardinality issues.

---

Checklist

Complete the checklist (and note appropriate exceptions) before the PR is marked ready-for-review.

• Changes are compatible¹
• Documentation² completed
• Performance impact assessed and acceptable
• Tests added and passing³
  • Unit Tests
  • Integration Tests
  • Manual Tests

Exceptions

Note any exceptions here

Notes

Footnotes

1. It may be appropriate to bring upcoming changes to the attention of other (impacted) groups. Please endeavour to do this before seeking PR approval. The mechanism for doing this will vary considerably, so use your judgement as to how and when to do this. ↩

2. Configuration is an important part of many changes. Where applicable please try to document configuration examples. ↩

3. Tick whichever testing boxes are applicable. If you are adding Manual Tests, please document the manual testing (extensively) in the Exceptions. ↩

[svc-apollo-docs]

✅ Docs preview ready

The preview is ready to be viewed. View the preview

File Changes

0 new, 2 changed, 0 removed

* graphos/reference/migration/from-router-v1.mdx
* graphos/reference/router/telemetry/instrumentation/standard-instruments.mdx

Build ID: d6eafb54039dc8c1ca88d695

URL: https://www.apollographql.com/docs/deploy-preview/d6eafb54039dc8c1ca88d695

[router-perf]

CI performance tests

• connectors-const - Connectors stress test that runs with a constant number of users
• const - Basic stress test that runs with a constant number of users
• demand-control-instrumented - A copy of the step test, but with demand control monitoring and metrics enabled
• demand-control-uninstrumented - A copy of the step test, but with demand control monitoring enabled
• enhanced-signature - Enhanced signature enabled
• events - Stress test for events with a lot of users and deduplication ENABLED
• events_big_cap_high_rate - Stress test for events with a lot of users, deduplication enabled and high rate event with a big queue capacity
• events_big_cap_high_rate_callback - Stress test for events with a lot of users, deduplication enabled and high rate event with a big queue capacity using callback mode
• events_callback - Stress test for events with a lot of users and deduplication ENABLED in callback mode
• events_without_dedup - Stress test for events with a lot of users and deduplication DISABLED
• events_without_dedup_callback - Stress test for events with a lot of users and deduplication DISABLED using callback mode
• extended-reference-mode - Extended reference mode enabled
• large-request - Stress test with a 1 MB request payload
• no-tracing - Basic stress test, no tracing
• reload - Reload test over a long period of time at a constant rate of users
• step-jemalloc-tuning - Clone of the basic stress test for jemalloc tuning
• step-local-metrics - Field stats that are generated from the router rather than FTV1
• step-with-prometheus - A copy of the step test with the Prometheus metrics exporter enabled
• step - Basic stress test that steps up the number of users over time
• xlarge-request - Stress test with 10 MB request payload
• xxlarge-request - Stress test with 100 MB request payload

[timbotnik]

😂 Cool to see a PoC in under 50 lines.

[timbotnik]

IIUC for errors without a code, it would count against the "empty" code e.g. code="". I think we could consider defining an explicit UNKNOWN code but it seems fine to support this with an empty code.

[bonnici]

Yeah this is an empty string. I can easily add a fallback code but I think an empty string is just as good.

[timbotnik]

I know that the Router team has asked for this to be a private metric - would that mean we would need to call it apollo_private.router.operations.error? Ultimately we don't want it to be, so I'm fine with it not being private at all.

[bonnici]

This should pass the "private" filter regex and not pass the "public" filter regex, so it should be private. I don't really know how to test that thought because it is still coming through in prometheus.

    pub(crate) fn private<T: Into<MeterProvider>>(delegate: T) -> Self {
        FilterMeterProvider::builder()
            .delegate(delegate)
            .allow(
                Regex::new(
                    r"apollo\.(graphos\.cloud|router\.(operations?|lifecycle|config|schema|query|query_planning|telemetry|instance))(\..*|$)|apollo_router_uplink_fetch_count_total|apollo_router_uplink_fetch_duration_seconds",
                )
                .expect("regex should have been valid"),
            )
            .build()
    }

    pub(crate) fn public<T: Into<MeterProvider>>(delegate: T) -> Self {
        FilterMeterProvider::builder()
            .delegate(delegate)
            .deny(
                Regex::new(r"apollo\.router\.(config|entities|instance|operations\.(connectors|fetch|request_size|response_size)|schema\.connectors)(\..*|$)")
                    .expect("regex should have been valid"),
            )
            .build()
    }

[timbotnik]

This is looking pretty good to me.

[timbotnik]

Good from my side.

[SimonSapin]

error[E0432]: unresolved import `opentelemetry_api`
  --> apollo-router/src/services/router/tests.rs:12:5
   |
12 | use opentelemetry_api::KeyValue;
   |     ^^^^^^^^^^^^^^^^^ use of undeclared crate or module `opentelemetry_api`

The dependency on opentelemetry_api was removed in #6493. In dev I see other files with use opentelemetry::KeyValue;, maybe that's what you need?