v1.55.1-rc.1

1.55.1-rc.1

v1.55.1-rc.0

1.55.1-rc.0

v1.52.2-rc.0

1.52.2-rc.0

v1.55.1-alpha.2

1.55.1-alpha.2

v1.55.1-alpha.1

1.55.1-alpha.1

v1.55.1-alpha.0

1.55.1-alpha.0

v1.55.0

Important

If you have enabled Distributed query plan caching, this release changes the hashing algorithm used for the cache keys. On account of this, you should anticipate additional cache regeneration cost when updating between these versions while the new hashing algorithm comes into service.

🚀 Features

Support aliasing standard attributes for telemetry (Issue #5930)

The router now supports creating aliases for standard attributes for telemetry.

This fixes issues where standard attribute names collide with reserved attribute names. For example, the standard attribute name entity.type is a reserved attribute name for New Relic, so it won't work properly. Moreover entity.type is inconsistent with our other GraphQL attributes prefixed with graphql.

The example configuration below renames entity.type to graphql.type.name:

telemetry:
  instrumentation:
    spans:
      mode: spec_compliant # Docs state this significantly improves performance: https://www.apollographql.com/docs/router/configuration/telemetry/instrumentation/spans#spec_compliant
    instruments:
      cache: # Cache instruments configuration
        apollo.router.operations.entity.cache: # A counter which counts the number of cache hit and miss for subgraph requests
          attributes:
            graphql.type.name: # renames entity.type
              alias: entity_type # ENABLED and aliased to entity_type

By @bnjjj in #5957

Enable router customizations to access demand control info (PR #5972)

Rhai scripts and coprocessors can now access demand control information via the context. For more information on Rhai constants to access demand control info, see available Rhai API constants.

By @tninesling in #5972

Support Redis connection pooling (PR #5942)

The router now supports Redis connection pooling for APQs, query planners and entity caches. This can improve performance when there is contention on Redis connections or latency in Redis calls.

By @Geal in #5942

🐛 Fixes

Remove unused fragments and input arguments when filtering operations (PR #5952)

This release fixes the authorization plugin's query filtering to remove unused fragments and input arguments if the related parts of the query are removed. Previously the plugin's query filtering generated validation errors when planning certain queries.

By @Geal in #5952

Hot-reloads will no longer interrupt certain gauges (PR #5996, PR #5999, PR #5999)

Previously when the router hot-reloaded a schema or a configuration file, the following gauges stopped working:

• apollo.router.cache.storage.estimated_size
• apollo_router_cache_size
• apollo.router.v8.heap.used
• apollo.router.v8.heap.total
• apollo.router.query_planning.queued

This issue has been fixed in this release, and the gauges now continue to function after a router hot-reloads.

By @BrynCooke in #5996 and #5999 and #6012

Datadog sample propagation will respect previous sampling decisions (PR #6005)

PR #5788 introduced a regression where sampling was set on propagated headers regardless of the sampling decision in the router or upstream.

This PR reverts the code in question and adds a test to check that a non-sampled request doesn't result in sampling in the downstream subgraph service.

By @BrynCooke in #6005

Include request variables when scoring for demand control (PR #5995)

Demand control scoring in the router now accounts for variables in queries.

By @tninesling in #5995

📃 Configuration

Enable new and old schema introspection implementations by default (PR #6014)

Starting with this release, if schema introspection is enabled, the router runs both the old Javascript implementation and a new Rust implementation of its introspection logic by default.

The more performant Rust implementation will eventually replace the Javascript implementation. For now, both implementations are run by default so we can definitively assess the reliability and stability of the Rust implementation before removing the Javascript one.

You can still toggle between implementations using the experimental_introspection_mode configuration key. Its valid values:

• new runs only Rust-based validation
• legacy runs only Javascript-based validation
• both (default) runs both in comparison and logs errors if differences arise

Having both as the default causes no client-facing impact. It will record and output the metrics of its comparison as a apollo.router.operations.introspection.both counter. (Note: if this counter in your metrics has rust_error = true or is_matched = false, please open an issue with Apollo.)

Note: schema introspection itself is disabled by default, so its implementation(s) are run only if it's enabled in your configuration:

supergraph:
  introspection: true

By @SimonSapin in #6014

🧪 Experimental

Allow disabling persisted-queries-based query plan cache prewarm on schema reload

The router supports the new persisted_queries.experimental_prewarm_query_plan_cache.on_reload configuration option. It toggles whether a query plan cache that's prewarmed upon loading a new schema includes operations from persisted query lists. Its default is true. Setting it false precludes operations from persisted query lists from being added to the prewarmed query plan cache.

Some background about the development of this option:

• In router v1.31.0, we started including operations from persisted query lists when the router prewarms the query plan cache when loading a new schema.

• Then in router v1.49.0, we let you also prewarm the query plan cache from the persisted query list during router startup by setting persisted_queries.experimental_prewarm_query_plan_cache to true.

• In this release, we now allow you to disable the original feature so that the router can prewarm only recent operations from the query planning cache (and not operations from persisted query lists) when loading a new schema.

Note: the option added in v1.49.0 has been renamed from persisted_queries.experimental_prewarm_query_plan_cache to persisted_queries.experimental_prewarm_query_plan_cache.on_startup. Existing configuration files will keep working as before, but with a warning that can be resolved by updating your config file:

 persisted_queries:
   enabled: true
-  experimental_prewarm_query_plan_cache: true
+  experimental_prewarm_query_plan_cache:
+    on_startup: true

By @glasser in #5990

v1.55.0-rc.2

1.55.0-rc.2

v2.0.0-alpha.7

2.0.0-alpha.7

v1.55.0-rc.1

1.55.0-rc.1