OpenRCA 2.0: From Outcome Labels to Causal Process Supervision
For researchers and developers of LLM-based RCA agents, this work provides a more rigorous evaluation benchmark that exposes a critical failure mode (ungrounded diagnosis) hidden by outcome-only labels.
The paper identifies a gap in root cause analysis (RCA) datasets that only label root causes without propagation paths, simplifying the task to pattern matching. It introduces PAVE, a protocol to reconstruct causal propagation paths, and creates OpenRCA 2.0, a benchmark with step-wise causal annotations. Evaluation of 11 LLMs shows only 20.7% success in recovering exact root-cause sets, and reveals that agents often identify correct root causes but fail to ground them in verified causal paths (61.5% grounding rate).
Root cause analysis (RCA) poses a holistic test of LLM agentic capabilities, such as long-context understanding, multi-step reasoning, and tool use. However, existing datasets suffer from a fundamental gap: they label only the root cause, not the propagation path connecting it to the observed symptom, which largely simplifies the task to naive pattern matching. To support rigorous evaluation, we introduce PAVE, a step-wise labeling protocol that leverages known interventions from fault injection to reconstruct causal propagation paths. The mechanism is forward verification: reasoning from cause to effect rather than inferring backward from symptoms. Applying PAVE yields OpenRCA 2.0 (500 instances), the first cross-system RCA benchmark with step-wise causal annotations for LLM agents. Across 11 frontier LLMs, recovering the exact root-cause set succeeds in only 20.7% of cases on average. To locate where this difficulty lies, we relax the criterion and find what we call the ungrounded diagnosis: agents identify at least one correct root-cause service in 76.0% of cases, but ground that service in a verified causal propagation path to the observed symptom in only 61.5%. Outcome-only evaluation hides this failure mode; step-wise causal ground truth is the missing piece for trustworthy LLM-based RCA agents.