Add information about limits of opened perf event files.

Signed-off-by: Paweł Szulik <[email protected]>

docs/runtime_options.md

@@ -152,6 +152,9 @@ Core perf events can be exposed on Prometheus endpoint per CPU or aggregated by
 - `--disable_metrics="percpu"` - core perf events are aggregated
 - `--disable_metrics=""` - core perf events are exposed per CPU.
 
+It's possible to get "too many opened files" error when a lof of perf events are exposed per CPU. This happens because of passing system limits.
+Try to increase it with `ulimit -n <value>`.
+
 Aggregated form of core perf events significantly decrease volume of data. For aggregated form of core perf events scaling ratio (`container_perf_metric_scaling ratio`) indicates the lowest value of scaling ratio for specific event to show the worst precision.
 
 ### Perf subsystem introduction