Spring Boot
This page lists every occurrent.* property the starters read. Your IDE can also code-complete them, and the core ones are defined in OccurrentProperties. A property marked blocking only is read by the blocking starter and ignored by the reactive starter.
| Property | Default | What it does |
|---|
occurrent.event-store.enabled | true | Whether the starter creates the EventStore bean. false also turns off the ApplicationService and DomainEventQueries beans. |
occurrent.event-store.capabilities | [STREAM] | Which capabilities the event store has, STREAM, DCB or both. Adding DCB turns on the Dynamic Consistency Boundary beans. The set can’t be empty. |
occurrent.event-store.mongodb.collection | events | The MongoDB collection events are stored in. |
occurrent.event-store.mongodb.time-representation | DATE | How time is stored in the CloudEvents, DATE or RFC_3339_STRING. |
occurrent.event-store.stream.position | on | Whether events written to a stream carry a global position. Unset means on, unless the store finds an existing collection without positions and turns it off. true keeps it on even then, and false turns it off for a store without DCB. A store with DCB always writes it. |
| Property | Default | What it does |
|---|
occurrent.application-service.enabled | true | Whether the starter creates the ApplicationService beans. A missing key counts as true. |
occurrent.application-service.enable-default-retry-strategy | true | Whether the application service retries when a write condition is not fulfilled. It backs off exponentially from 100 ms up to 2 s, and rethrows the original exception after 5 retries. |
occurrent.cloud-event-converter.cloud-event-source | none | The CloudEvent source the converter writes, a URN such as urn:mycompany:mygame:game. Think of it as the stream type or category. |
occurrent.cloud-event-converter.time-precision | MILLIS with DATE | A ChronoUnit the converter truncates the event time to. When it’s unset and the time representation is DATE, which can’t store more than milliseconds, the converter truncates to MILLIS. Otherwise it leaves the time alone. See CloudEvent conversion. |
| Property | Default | What it does |
|---|
occurrent.subscription.mode | AUTO | How much of the subscription machinery the starter creates. AUTO creates subscriptions and starts them, MANUAL creates them and lets you start them yourself, and DISABLED creates none. |
occurrent.subscription.mongodb.collection | subscriptions | The MongoDB collection checkpoints are stored in. |
occurrent.subscription.mongodb.restart-on-change-stream-history-lost | true | Whether a subscription restarts from the current time when the MongoDB oplog no longer has the history it needs to resume. It is not completely safe, see Restarting when the oplog is lost. |
occurrent.subscription.start-when-no-start-position-can-be-recorded | false | Whether a durable subscription with no checkpoint starts anyway when the subscription model can’t tell it where the event store is right now, for example on a shared MongoDB Atlas cluster. Without it, the subscription is refused when you create it. See Starting when no start position can be recorded. |
occurrent.subscription.competing-consumer.fence-checkpoints | true | Blocking only. Whether a checkpoint write carries the competing-consumer lease version, so that a node that has lost its lease can’t move the checkpoint backward. It needs a CheckpointStorage that evaluates write conditions. See Checkpoint fencing. |
occurrent.subscription.catchup-then-live.dedup-cache-size | 10000 | How many recently delivered event ids a @Projection(source = PUSH) remembers, to remove the duplicates where the catch-up and the live feed overlap. |
occurrent.subscription.catchup-then-live.max-buffered-events | 100000 | How many events from the push feed may wait to be delivered before the projection fails. It is a ceiling, not a throttle. |
These configure the store that records applied appends for @Projection(recordAppliedAppends = true). That page has the details.
| Property | Default | What it does |
|---|
occurrent.projection.applied-append.collection | appliedAppends | The MongoDB collection recorded appends are stored in. |
occurrent.projection.applied-append.retention | 7d | How long a recorded append is kept before a MongoDB TTL index deletes it. |
occurrent.projection.applied-append.max-attempts | 10 | How many times the store calls MongoDB for one read or write before it fails, at most 1000. |
occurrent.projection.applied-append.wait-backoff.initial | 25ms | How long waitUntilApplied waits before it checks again the first time. |
occurrent.projection.applied-append.wait-backoff.max | 250ms | The longest wait between two checks. |
occurrent.projection.applied-append.wait-backoff.multiplier | 2.0 | What the wait is multiplied by after each check that found the append not yet applied. |
occurrent.projection.applied-append.replay-poll.initial | 200ms | How often the starter checks a projection that just registered, or that still has a clear or a catch-up pending. |
occurrent.projection.applied-append.replay-poll.max | 5s | The longest interval between two checks. |
occurrent.projection.applied-append.replay-poll.multiplier | 2.0 | What the interval is multiplied by after each check that found nothing to do. |
All of these are blocking only.
| Property | Default | What it does |
|---|
occurrent.saga.timer-poll-interval | 15s | How often a saga’s timer poller looks for timeouts that are due. See Running across multiple instances. |
occurrent.saga.competing-consumer.enabled | true | Whether only one instance of the application polls for due timers, using the shared competing-consumer lease. See Several instances of a @Saga. |
occurrent.saga.quarantine-after | none | How long one saga instance may keep failing before it can be quarantined. Unset means quarantine is off. Zero or a negative duration stops the application from starting when it has a @Saga. See Observing @Saga instances. |
occurrent-broker-kafka-spring-boot-starter reads these. See Brokers with Spring Boot.
| Property | Default | What it does |
|---|
occurrent.broker.kafka.bootstrap-servers | none | Kafka’s bootstrap.servers, as a comma-separated string or a list. The sink and source factories are only created when it is set. |
occurrent.broker.kafka.topic | none | The topic the auto-configured KafkaSharedTopicDestinationResolver publishes to and subscribes to. You don’t need it when you declare your own DestinationResolver<KafkaDestination>. |
occurrent.broker.kafka.producer.additional-properties.<name> | none | Any other Kafka producer setting, TLS and SASL included. The serializers are always overwritten, acks defaults to all, and any acks other than all or -1 is refused. |
occurrent.broker.kafka.consumer.additional-properties.<name> | none | Any other Kafka consumer setting. enable.auto.commit defaults to false and any other value is refused. A group.id set here is ignored. |
occurrent.broker.kafka.sink.acknowledgement-timeout | 5s | How long the sink waits for Kafka to acknowledge a write. |
occurrent.broker.kafka.sink.retry.initial | 100ms | The first delay when the sink retries a write. |
occurrent.broker.kafka.sink.retry.max | 2s | The longest delay between two retries. |
occurrent.broker.kafka.sink.retry.multiplier | 2.0 | What the delay is multiplied by after each retry. |
occurrent.broker.kafka.source.poll-timeout | 1s | How long one poll of a source waits for records. |
occurrent.broker.kafka.source.close-timeout | 30s | How long closing a source may take. |
occurrent.broker.kafka.source.on-delivery-failure | REDELIVER | The DeliveryFailurePolicy of a source when a handler fails, REDELIVER or PARK. |
occurrent.broker.kafka.source.parking-destination.topic | none | The topic PARK publishes a failed record to. It’s only needed with PARK. |
occurrent.broker.kafka.source.commit-retry.initial | 100ms | The first delay when a source retries an offset commit. |
occurrent.broker.kafka.source.commit-retry.max | 2s | The longest delay between two retries. |
occurrent.broker.kafka.source.commit-retry.multiplier | 2.0 | What the delay is multiplied by after each retry. |
occurrent-broker-rabbitmq-spring-boot-starter reads these. See Brokers with Spring Boot.
| Property | Default | What it does |
|---|
occurrent.broker.rabbitmq.exchange | none | The exchange the auto-configured RabbitMqTopicExchangeDestinationResolver publishes to and binds against. You don’t need it when you declare your own DestinationResolver<RabbitMqDestination>. |
occurrent.broker.rabbitmq.sink.acknowledgement-timeout | 5s | How long the sink waits for RabbitMQ to confirm a publish. |
occurrent.broker.rabbitmq.sink.retry.initial | 100ms | The first delay when the sink retries a publish. |
occurrent.broker.rabbitmq.sink.retry.max | 2s | The longest delay between two retries. |
occurrent.broker.rabbitmq.sink.retry.multiplier | 2.0 | What the delay is multiplied by after each retry. |
occurrent.broker.rabbitmq.source.poll-interval | 1s | How often a source polls its queue. |
occurrent.broker.rabbitmq.source.prefetch-count | 1 | How many unacknowledged messages RabbitMQ sends a source at a time. |
occurrent.broker.rabbitmq.source.close-timeout | 30s | How long closing a source may take. |
occurrent.broker.rabbitmq.source.declare-topology | true | Whether a source declares its queue and bindings. Set it to false when your platform team owns them. |
occurrent.broker.rabbitmq.source.on-delivery-failure | REDELIVER | The DeliveryFailurePolicy of a source when a handler fails, REDELIVER or PARK. |
occurrent.broker.rabbitmq.source.parking-destination.exchange | none | The exchange PARK publishes a failed message to. It’s only needed with PARK, together with the routing key. |
occurrent.broker.rabbitmq.source.parking-destination.routing-key | none | The routing key PARK publishes a failed message with. |
occurrent.broker.rabbitmq.source.retry.initial | 100ms | The first delay when building a source retries. |
occurrent.broker.rabbitmq.source.retry.max | 2s | The longest delay between two retries. |
occurrent.broker.rabbitmq.source.retry.multiplier | 2.0 | What the delay is multiplied by after each retry. |
occurrent.broker.rabbitmq.source.retry.max-attempts | 10 | How many times building a source tries before it fails. |
These still work, and will be removed two releases after 0.34.0. You can set a deprecated key and its replacement together as long as they have the same value. The Spring Boot starter page says how to migrate.
| Deprecated property | Use instead |
|---|
occurrent.event-store.collection | occurrent.event-store.mongodb.collection |
occurrent.event-store.time-representation | occurrent.event-store.mongodb.time-representation |
occurrent.subscription.collection | occurrent.subscription.mongodb.collection |
occurrent.subscription.restart-on-change-stream-history-lost | occurrent.subscription.mongodb.restart-on-change-stream-history-lost |
occurrent.subscription.enabled | occurrent.subscription.mode, where false is DISABLED and true is AUTO |
