The connector option 'flush_interval' has been removed and…

Error message

The connector option 'flush_interval' has been removed and is ignored. Use the engine-level 'sink.flush.interval' in the job 'env' block instead. Engine-level timer flush is supported only by Zeta; on Spark and Flink there is no periodic flush, so tune 'batch_size' instead.

What it means

The Prometheus sink's connector-level 'flush_interval' option was removed in favor of the engine-level 'sink.flush.interval' setting in the job 'env' block. A leftover key in an upgraded job config is silently ignored on direct job runs (only --check/--dry-run reject unknown keys), so the writer logs this warning once per writer instance to signal that periodic flushing at the connector level is no longer happening.

Solutions

Remove 'flush_interval' from the sink config block
Add 'sink.flush.interval' to the job 'env' block (Zeta only)
On Spark/Flink, tune 'batch_size' instead since there is no periodic timer flush
Run the job with --check or --dry-run to catch leftover unknown keys before production runs

Example fix

// before
sink {
  Prometheus {
    flush_interval = 5000
  }
}
// after
env {
  sink.flush.interval = 5000
}
sink {
  Prometheus {
  }
}

Defensive patterns

Strategy: validation

Validate before calling

if (sinkConfig.containsKey("flush_interval")) { throw new IllegalArgumentException("flush_interval removed; use env-level sink.flush.interval"); }

Prevention

Grep job configs for 'flush_interval' before upgrading
Use --check/--dry-run to surface removed keys
Move flush settings to the env block on Zeta
Tune batch_size on Spark/Flink instead of expecting timer flush

When it happens

Trigger: A Prometheus sink job config still contains the 'flush_interval' key under the sink options; the key is detected in pluginConfig.getSourceMap() when PrometheusWriter is opened.

Common situations: Upgrading SeaTunnel from a version where connector-level flush_interval existed without cleaning the job config; copying old batch config templates; expecting periodic flush on Spark/Flink where only batch_size applies.

Understand the failure class

Background: "is deprecated and will be removed" — deprecation warnings for old API names, keywords, and options, and how to migrate before the removal release — this error's family across 29 libraries.

Related errors

AI-assisted analysis of apache/seatunnel@cf67b549a7 (2026-09-10). Data as JSON: /api/errors/4bc6d32ec2600e1d. Report an issue: GitHub.

Appendix: source

Thrown at seatunnel-connectors-v2/connector-prometheus/src/main/java/org/apache/seatunnel/connectors/seatunnel/prometheus/sink/PrometheusWriter.java:84

        super(seaTunnelRowType, httpParameter);
        this.batchList = new ArrayList<>();
        this.sinkConfig = PrometheusSinkConfig.loadConfig(pluginConfig);
        this.batchSize = sinkConfig.getBatchSize();
        this.serializer =
                new PrometheusSerializer(
                        seaTunnelRowType,
                        sinkConfig.getKeyTimestamp(),
                        sinkConfig.getKeyLabel(),
                        sinkConfig.getKeyValue());
        this.httpClient = new HttpClientProvider(httpParameter);
        // The connector-level `flush_interval` option was removed in favor of the engine-level
        // `sink.flush.interval`. A leftover key in an upgraded job config is silently ignored on a
        // direct job run (only `--check`/`--dry-run` reject unknown keys), so warn here (once per
        // writer instance) to give operators a signal instead of silently dropping periodic
        // flushing.
        if (pluginConfig.getSourceMap().containsKey(REMOVED_FLUSH_INTERVAL_KEY)) {
            log.warn(
                    "The connector option 'flush_interval' has been removed and is ignored. Use the "
                            + "engine-level 'sink.flush.interval' in the job 'env' block instead. "
                            + "Engine-level timer flush is supported only by Zeta; on Spark and "
                            + "Flink there is no periodic flush, so tune 'batch_size' instead.");
        }
        // Opt in to engine-level timer flush. On Zeta the engine invokes this action on the normal
        // Sink input-processing path when a FlushSignal arrives, so there is no connector-owned
        // scheduler thread and no concurrency with write/checkpoint/close. On Spark and Flink the
        // Context does not implement registerFlushAction (it keeps the interface's no-op default),
        // so there is no periodic timer flush there; the buffer is flushed on batch_size and on
        // close(). The null-check is defensive for non-standard/test call sites that may not supply
        // a context.
        if (context != null) {
            context.registerFlushAction(this::flush);
        }
    }

    @Override

View on GitHub (pinned to cf67b549a7)