PAI collectors currently use the batch processor but do not have a memory limiter.
During traffic spikes, or when Parseable is slow/down, data can build up inside the collector. This may increase memory usage until the collector gets OOM-killed.
Add support for the OpenTelemetry memory_limiter processor for logs, metrics, traces, and events pipelines.
The processor order will be:
receiver → memory_limiter → batch → exporter
Keep it disabled by default so existing users.
PAI collectors currently use the batch processor but do not have a memory limiter.
During traffic spikes, or when Parseable is slow/down, data can build up inside the collector. This may increase memory usage until the collector gets OOM-killed.
Add support for the OpenTelemetry memory_limiter processor for logs, metrics, traces, and events pipelines.
The processor order will be:
receiver → memory_limiter → batch → exporter
Keep it disabled by default so existing users.