[Chronosphere docs]: if the Collector cannot scrape a job or component, push metrics to it manually. For Kubernetes: if the Collector cannot scrape kube-state-metrics, the cluster itself is likely unhealthy, so resolve the cluster issue first. A DaemonSet Collector can OOM as the cluster scales; run the Collector as a sidecar to get staleness markers. Staleness markers require Collector 0.86.0 or later, are disabled by default, and cost about 10 percent more memory at 500 series per second; enable with scrape.enableStalenessMarker set to true. Verify with kubectl logs on the Collector pod and confirm data streaming in Observability Platform.

Context: Missing metrics on Chronosphere usually mean a scrape problem, a scaling problem, or stale series in query results. Each has a documented check.

## Matched source
Source: Source: https://docs.chronosphere.io/ingest/metrics-traces/collector
Original query: "Chronosphere Collector: verify streaming, scraping, and stale targets"
Key terms: chronosphere, collector, scraping, stale, streaming, targets, verify