Enum ParameterScope 

pub enum ParameterScope {
    Environment,
    Cluster,
    Replica,
}

The scope at which a synced parameter’s value may be overridden.

Every synced parameter declares its scope class as part of its definition. The declaration is the single source of truth for which contexts the LaunchDarkly sync loop evaluates and where the resolved value may be overridden. See doc/developer/design/20260609_scoped_feature_flags.md.

Environment is the safe declaration, and the right one when in doubt. It preserves the unscoped behavior of a single value everywhere. A finer scope enables divergence, so declaring one asserts that divergence is both safe and useful for this config. Getting that wrong introduces a way to break an environment that did not previously exist, while erring toward Environment only forgoes a capability. Prefer forgoing the capability.

Where a config is realized is therefore a necessary condition for a finer scope, not a sufficient one. A config qualifies for Replica only if it is realized per replica, and should be declared Replica only if it also tunes that replica process’s own resource usage (memory, CPU, I/O, concurrency, timing) and cannot change what a dataflow produces, what reaches durable state, or anything externally visible. lgalloc, the memory limiter, the spill and pager knobs, and the timely zero-copy settings are the shape to match.

In particular, declare Environment for a flag selecting a different implementation or code path whose output-equivalence is an assumption rather than a guarantee. Where the assumption holds, Environment costs nothing. Where it does not, a per-replica rollout turns one bug into query results that differ by which replica served them.

Which contexts a config can be realized in:

• A config realized inside a clusterd process is a Replica candidate. That covers the compute and storage worker config sets and mz_metrics, which the per-replica dyncfg push reaches. environmentd may read such a config too, for its own process. That read legitimately sees the environment-wide value.
• A config that environmentd resolves for one specific replica, whether it ships the value there or acts on it itself, is also a candidate. The read site has to resolve that replica’s override, either with Config::get_with_overrides or from a per-replica config set. Without that, the override never takes effect and the declaration is a silent no-op.
• A config realized in environmentd with no single replica in scope is Environment. So is every balancerd config. balancerd syncs LaunchDarkly itself, against a balancer context keyed by cloud provider, region and build version, and has no environment, cluster or replica beneath it to target.
• A config consumed at plan time, once per cluster, is Cluster.

NOTE: a config whose value must agree across the replicas of one cluster (because they render the same dataflow and their outputs are compared) is Environment even when it is read on clusterd. Per-replica divergence in how a dataflow is rendered is fine, in what it produces is not.

NOTE: persist configs are Environment as a class, even though the persist client runs on clusterd and the per-replica push reaches its config set. The same client code also runs in environmentd, and every copy of it acts on shared durable state. A replica-scoped persist config could never reach the environmentd client, so a rollout targeting replicas would leave a shard’s other writer on the old value indefinitely. Environment is the only scope covering every persist client in an environment.

Variants

Environment

Environment-wide only; no cluster/replica overrides. The default, so all existing synced parameters are unchanged.

NOTE: this names the coarsest targeting granularity, not the environmentd process. A config a process other than environmentd resolves for itself, with nothing finer beneath it, is Environment.

Cluster

Cluster-coherent: env-wide base plus per-cluster overrides. Evaluated with the cluster context (replica-free) and resolved at plan time via OptimizerFeatureOverrides. e.g. optimizer features.

Replica

Replica-local: env-wide base plus per-replica / per-size-family overrides. Evaluated with the replica context and resolved at the controller’s per-replica dyncfg push. e.g. lgalloc, persist pager, LZ4.

Implementations

impl ParameterScope

pub const DEFAULT: ParameterScope = ParameterScope::Environment

The scope applied to a parameter that does not declare one: environment- wide, i.e. no cluster/replica overrides. A const so it can be used in the const contexts (system-var constructors, the feature_flags! macro) where Default::default is unavailable.

pub const fn as_str(&self) -> &'static str

Returns the lowercase string name of this scope, as surfaced in documentation and introspection.

Trait Implementations

impl Clone for ParameterScope

fn clone(&self) -> ParameterScope

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for ParameterScope

impl Debug for ParameterScope

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ParameterScope

fn default() -> Self

Returns the “default value” for a type.

impl Eq for ParameterScope

impl PartialEq for ParameterScope

fn eq(&self, other: &ParameterScope) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for ParameterScope