Kotlin Monitoring Public Interface

Interface table of contents · Java Monitoring

Kotlin uses Java's RouteMesh, ClientServer, automatic fanout, and host runtime status types. The Flow projection only connects the Java publisher to coroutine cancellation — it doesn't define a separate status or event value. Each item is a complete status after a change, not an event holding only some fields.

Loss information also passes through as Java ZLinkObservationLoss. Runtime monitoring §7.2 owns the semantics, range, and saturation of its two long counters.

The host's capacity state also uses Java's ZLinkHostCapacityStatus unchanged. A Kotlin-only data class or Flow aggregation isn't added. Core HWM runtime snapshot, application job queue configured pause/resume percentages, computed permit counts, RUNNING|PAUSED pressure state, Duration current pause duration, and resetCapacityMetrics() therefore use the Java projection unchanged. The inherited applicationAccountedBytes, outstandingApplicationLeaseCount, retiredQueueCount, and deferredOriginCreditBytes fields are ABI-reserved and always 0 since 0.13.1. Cumulative pause duration, transition counts, and configuration-failure counts are metric-only values, not fields of the public status projection. Reset preserves the current pressure state and current pause duration while clearing those metric-only cumulative/count values.

Endpoint, lifecycle generation, and descriptor source are kept only for the framework to judge a stale descriptor and connection. Other than aggregate queue reservation, per-owner admission/claim, payload, and connection intent aren't added to the Kotlin projection.

A RouteMesh peer uses Java's ZLinkPeerState unchanged. NOT_CONNECTED is a state where a connection is needed but there's no ready connection, and NOT_REQUIRED is a normal state where neither Object Client has RouteMesh Channel Server membership so a connection isn't needed. The same applies when only Channel Client membership is registered. If either side has Channel Server membership, including weight 0, absence of connection is marked NOT_CONNECTED. The two states aren't merged into a Kotlin-only boolean or string. NOT_REQUIRED is excluded from ready peer count and liveness/health failure aggregation.

The topology runtime uses Java ZLinkFrameworkRuntime's routeMeshRuntime(), clientServerRuntime(), and fanoutRuntime() unchanged. A Kotlin wrapper accessor isn't added, and the topology bean injected in Spring has the same reference identity as the return value of that Java accessor. As with the Java monitoring contract, MeshNode status has no Logical Multicast statistics, publish target count, or per-target admission/failure fields. A Kotlin-only projection doesn't add these.

The ClientServer target and fanout publisher use Java's ZLinkPeerState and ZLinkTopologyReason unchanged. A Kotlin-only connection status enum isn't created. The local role of a snapshot that registered Client and Server together on the same ChannelName is represented by Java's ZLinkClientServerRole.CLIENT_AND_SERVER. This is only an aggregate projection of two separate role registrations, not a builder role or registration key. A Kotlin-only enum or conversion value isn't created.

Fanout ready semantics also use the Java contract unchanged. The publisher-dedicated SUB socket's native-ready alone doesn't become ready — the first valid application record or liveness beacon must also be received on the same socket. The 15-second inbound timeout changes that publisher's peer state to NOT_CONNECTED.

An error occurring in an internal runtime callback or provider is recorded by the framework as a structured log. A message-flow observer, error sink, and raw event DTO the Kotlin application implements or registers aren't the public contract. Kotlin uses Java's four diagnostic levels OFF, ERRORS, NORMAL, and DETAILED unchanged. A failure in a standard logger/trace/metric provider configured by the application does not change the original message operation's terminal result and is isolated as separate diagnostics.

RouteMesh placement status only provides whether new objects can be accepted and the current process's active Actor/Spot count. Node-wide placement weight, per-stable-type capacity, pending activation, and reservation failure are internal placement judgment values, so they aren't made public. isAvailable is only true when the host is SERVING and Object Server, placement weight is positive, and both Actor/Spot capacity and activation concurrency have room. Activation's current value and limit also aren't added to the Kotlin projection.

Framework Error Values

Kotlin uses Java's ZLinkFrameworkErrorKind unchanged. The enum name and number are part of the public exception classification and fix the following values.

NOT_FOUND = 0
ALREADY_EXISTS = 1
TYPE_MISMATCH = 2
NOT_CONFIGURED = 3
REJECTED = 4
UNAVAILABLE = 5
CAPACITY_EXCEEDED = 6
DEADLINE_EXCEEDED = 7
SHUTTING_DOWN = 8
PROTOCOL_ERROR = 9
INVALID_OPERATION = 10
DATA_LOST = 11
INTERNAL_FAILURE = 12

A remote framework error is delivered as ZLinkFrameworkException. Public argument validation uses the JVM standard IllegalArgumentException, and a startup configuration conflict uses ZLinkConfigurationException. The public exception doesn't provide whether it's retryable.

When reading the Java Publisher status stream as a Kotlin Flow, the common asFlow() bridge owned by Location And Maintenance is used. This bridge's cancellation only releases that subscriber registration. It doesn't cancel the shared runtime, monitoring publisher, or an already-started host operation.