Package org.apache.ignite.table

Class ContinuousQueryOptions

java.lang.Object
org.apache.ignite.table.ContinuousQueryOptions
public class ContinuousQueryOptions extends Object
Continuous query options.

  • Field Details

    • DFLT_PAGE_SIZE

      public static final int DFLT_PAGE_SIZE
      Default page size.
      See Also:

    • DFLT_POLL_INTERVAL_MS

      public static final int DFLT_POLL_INTERVAL_MS
      Default poll interval in milliseconds.
      See Also:

    • DFLT_LONG_POLLING_WAIT_TIME_MS

      public static final int DFLT_LONG_POLLING_WAIT_TIME_MS
      Default long polling wait time in milliseconds.
      See Also:

    • DEFAULT

      public static final ContinuousQueryOptions DEFAULT

  • Method Details

    • builder

      public static ContinuousQueryOptions.Builder builder()
      Creates a new builder.
      Returns:
      Builder.

    • pageSize

      public int pageSize()
      Gets the per-partition page size.

      Continuous Query polls every partition in a loop. This parameter controls the number of entries that will be requested from a single partition in one network call. Therefore, the maximum number of entries that the query may hold in memory at any given time is pageSize * partitions.

      Default value is DFLT_PAGE_SIZE.

      Returns:
      Page size.

    • pollIntervalMs

      public int pollIntervalMs()
      Gets the poll interval in milliseconds. Must be non-negative. Determines the pause between partition polling cycles.

      Default value is DFLT_POLL_INTERVAL_MS.

      Returns:
      Poll interval in milliseconds.

    • eventTypes

      public EnumSet<TableRowEventType> eventTypes()
      Gets the included event types.

      By default, all event types are included.

      Returns:
      Included event types.

    • columnNames

      @Nullable public @Nullable Set<String> columnNames()
      Gets the column names to include in the result.

      By default, all columns are included. Excluding columns will reduce the amount of data transferred over the network.

      Excluded columns are still present in the resulting entries, but are set to null.

      Returns:
      Included column names, or null to load all columns.

    • partitions

      @Nullable public @Nullable Set<Partition> partitions()
      Gets the partitions to subscribe to.

      By default (null), all partitions are included.

      Returns:
      Included partitions, or null to include all partitions.

    • watermark

      @Nullable public @Nullable ContinuousQueryWatermark watermark()
      Gets the starting watermark. When null, the query will start from the current time.

      Watermark can be obtained with ContinuousQueryWatermark.ofInstant(Instant), or from an event with TableRowEvent.watermark(). The latter allows resuming a query from a specific event (excluding said event, providing exactly-once semantics).

      Default value is null, meaning the query will start from the current time.

      Returns:
      Starting watermark.

    • enableEmptyBatches

      public boolean enableEmptyBatches()
      Gets the flag that indicates whether empty batches are enabled.

      When enabled, empty batches will be sent to the subscriber when there are no new events. This is useful for watermark updates (see TableRowEventBatch.watermark()).

      Default value is false, meaning empty batches are not sent to the subscriber.

      Returns:
      True if empty batches are enabled, false otherwise.

    • executor

      public Executor executor()
      Returns the executor that is going to be used for async delivery and execution of subscriber methods.

      By default common pool is used - ForkJoinPool.commonPool().

      Returns:
      Executor to use for async delivery.

    • skipOldEntries

      public boolean skipOldEntries()
      Gets the flag that indicates whether old entries should be skipped for TableRowEventType.UPDATED events. When true, TableRowEvent.oldEntry() will return null for TableRowEventType.UPDATED events. This reduces the amount of data transferred over the network.

      Default value is false, meaning old entries are not skipped.

      Returns:
      True if old entries should be skipped, false otherwise.

    • longPollingWaitTimeMs

      public int longPollingWaitTimeMs()
      Gets the long polling wait time in milliseconds.

      Default value is DFLT_LONG_POLLING_WAIT_TIME_MS. A zero or negative value disables long polling.

      Long polling is an optimization that reduces latency and network round trips. When the client polls a partition and no events are immediately available, instead of returning an empty response, the server holds the request open for up to longPollingWaitTimeMs milliseconds, waiting for new events to arrive. If events arrive during this wait period, they are returned immediately. If the timeout expires with no events, an empty response is returned.

      Works together with pollIntervalMs(). After receiving a response (with or without data), the client waits pollIntervalMs. For minimal latency, reduce pollIntervalMs and increase longPollingWaitTimeMs. For better throughput and batching, increase pollIntervalMs and pageSize.

      Returns:
      Long polling wait time in milliseconds. A zero or negative value means no long polling.