Short description

After upgrading to OPA 1.9.0, 90% of our decision logs are dropped.

level: error
msg: Dropped 1 chunks from buffer. Reduce reporting interval or increase buffer size.
plugin: decision_logs

Our PDP service and OPA each have 0,25 CPU and 0,25 Gb memory. Memory usage is <50%

Examples:

This is with 80k req/min, but we also see this in lower environments where the throughput is barely nothing.

OPA 1.5.1, pushed 15k logs per min.
OPA 1.9.0, pushed 1k logs per min, rest was dropped

config;

package rules.discovery

config = {
	"bundles": {"pdp-service": {"polling": {
		"min_delay_seconds": 10,
		"max_delay_seconds": 30,
	}}},
	"decision_logs": {"reporting": {
		"min_delay_seconds": 10,
		"max_delay_seconds": 30,
		"buffer_size_limit_bytes": 100000000,
	}},
	"status": {},
	"default_decision": "rules/main/allow",
}

Expected behavior

I recall there is new event based logger, that's more performant, but I would expect the size based to continue work as is.