# atlassian automated rca

Atlassian's automated root cause analysis system correlates metrics, logs, traces, time, and service topology to generate ranked fault hypotheses.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## CLEW: Log-Anchored Backward Tracing for Root-Cause Code Localization

DevFeed: [CLEW: Log-Anchored Backward Tracing for Root-Cause Code Localization](<https://devfeed.tech/articles/clew-log-anchored-backward-tracing-for-root-cause-code-localization-137449.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19448>)

Author: Wenji Bai, Muhammad Waseem, Jaakko Peltonen, Pekka Abrahamsson

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [reliability-engineering](<https://devfeed.tech/topics/reliability-engineering.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [approach](<https://devfeed.tech/tags/approach.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [cause](<https://devfeed.tech/tags/cause.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [defect](<https://devfeed.tech/tags/defect.md>), [diagnostic](<https://devfeed.tech/tags/diagnostic.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [events](<https://devfeed.tech/tags/events.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [failure](<https://devfeed.tech/tags/failure.md>), [fault-injection](<https://devfeed.tech/tags/fault-injection.md>), [function](<https://devfeed.tech/tags/function.md>), [functions](<https://devfeed.tech/tags/functions.md>), [localization](<https://devfeed.tech/tags/localization.md>), [log](<https://devfeed.tech/tags/log.md>), [root](<https://devfeed.tech/tags/root.md>), [tracing](<https://devfeed.tech/tags/tracing.md>)

### AI overview

CLEW anchors runtime log observations to symptom functions, then uses LLM-guided backward tracing to rank candidate root-cause functions and provide a proposed mechanism path linked to log events. On the 192-case LogLoc benchmark, it achieved the highest reported Acc@1 (77.3%) and MRR (81.3%) on LogLoc-Issues, and matched leading baselines at 97.6% Acc@1 on LogLoc-Airflow. Across the cases, its wall-clock time was about 73% lower than LocAgent's at similar average inference cost. The abstract reports no aggregate degradation under the tested perturbations, with modest metric changes across evaluated seeds and base models; these findings are limited to the reported evaluation conditions.

### Source excerpt

arXiv:2611.19448v1 Announce Type: cross Abstract: Failures in industrial software systems may involve multiple components and be difficult to reproduce outside the production environment. Root-cause code localization therefore often relies on issue descriptions and runtime logs. However, log evidence may be noisy or incomplete, and the code reporting a symptom may differ from the code containing the underlying defect. These conditions make it challenging to identify relevant log evidence and determine which functions could have caused the reported failure. To address this gap, we introduce CLEW, an agentic approach that combines rule-based log-to-code anchoring with LLM-guided backward code tracing. CLEW organizes selected observations into diagnostic states, guides source exploration from anchored symptom functions, and returns ranked candidate functions, an affected module, and an explanation comprising a proposed mechanism path and associated log events. For evaluation, we construct LogLoc, a benchmark comprising 192 real-world issue and fault-injection cases. Against three baselines, CLEW records the highest observed Acc@1 (77.3%) and MRR (81.3%) on LogLoc-Issues and matches the leading baselines at 97.6% Acc@1 on LogLoc-Airflow. Averaged over all 192 cases, CLEW's wall-clock time is approximately 73% lower than LocAgent's at a similar average inference cost. No aggregate degradation is observed under the tested perturbations, with modest changes in Acc@1 and MRR across the evaluated seeds and base models. These results suggest that CLEW can combine function-level localization with evidence links and runtime efficiency under the evaluated conditions.

## Fault-Free Profiles and Decision-Level Training for Small-Model Root Cause Analysis

DevFeed: [Fault-Free Profiles and Decision-Level Training for Small-Model Root Cause Analysis](<https://devfeed.tech/articles/fault-free-profiles-and-decision-level-training-for-small-model-root-cause-analysis-113526.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.02416>)

Author: Kyoohyun Lee, Woohyeok Park, Taeyoon Kim, Sungyong Baik, Kyungyong Lee

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [post-training](<https://devfeed.tech/topics/post-training.md>)

Tags: [cs-se](<https://devfeed.tech/tags/cs-se.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [profiling](<https://devfeed.tech/tags/profiling.md>), [root-cause-analysis](<https://devfeed.tech/tags/root-cause-analysis.md>), [small-language-models](<https://devfeed.tech/tags/small-language-models.md>), [systems](<https://devfeed.tech/tags/systems.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

The proposed system improves small-model root cause analysis by adding context about how often metric and log signals appear anomalous during fault-free operation, then correcting decision inconsistencies through preference optimization. The study finds that models can favor large anomaly scores even when their propagation analysis points elsewhere. A 4B model using these methods achieved the highest or tied-highest accuracy across four datasets among the reported baselines, including 14B OpsAgent and RCA-Agent with commercial models; on Bank, it reached 48.7% versus OpsAgent's 18.5%.

### Source excerpt

arXiv:2611.02416v1 Announce Type: new Abstract: Root cause analysis (RCA) in microservice systems requires distinguishing failure sources from routine fluctuations and propagated symptoms. Small language models (SLMs) offer a practical solution under cost and privacy constraints, and recent approaches improve their capabilities through structured diagnostic procedures. However, performing diagnostic analysis does not ensure correct root-cause identification. Our empirical study shows that final decisions often favor large anomaly scores despite the model's own propagation analysis suggesting otherwise. Moreover, the input does not reveal whether these apparently anomalous signals are common during normal operation. These findings motivate our system, which improves how SLMs assess evidence and select root causes through Historical Frequency Profiling and Biased-Decision Correction. Historical Frequency Profiling quantifies how often metric and log signals are flagged as anomalous in the fault-free period, providing context for distinguishing routine deviations from potential failure evidence. Biased-Decision Correction identifies points where the model reaches different conclusions from the same initial analysis and uses preference optimization to favor more accurate diagnoses. Against baselines including the 14B OpsAgent and RCA-Agent with commercial models, our system with a 4B model achieves the highest or tied-highest accuracy on all four datasets, reaching 48.7% on Bank compared with 18.5% for OpsAgent.

## How Far Can Root Cause Analysis Go on Real-World Telemetry Data?

DevFeed: [How Far Can Root Cause Analysis Go on Real-World Telemetry Data?](<https://devfeed.tech/articles/how-far-can-root-cause-analysis-go-on-real-world-telemetry-data-97953.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.13548>)

Author: Athira Gopal, Ashwanth Krishnan

Published: 2026-10-23T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [observability](<https://devfeed.tech/topics/observability.md>), [benchmark overfitting machine learning](<https://devfeed.tech/topics/benchmark-overfitting-machine-learning.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [root-cause-analysis](<https://devfeed.tech/tags/root-cause-analysis.md>), [telemetry](<https://devfeed.tech/tags/telemetry.md>)

### AI overview

For OpenRCA microservice failure analysis, telemetry-specific rules mined from audit reports improved Full Score on a held-out cloudbed from 14.10 to 35.90, outperforming manual rules (24.36) and a label-and-schema control (19.23); the result survived multiple-testing correction. The rules added 17 solved cases alongside a five-step procedure and 11 without it, with overlapping repairs. Removing the procedure raised scores in all three tested conditions, but no contrast survived correction; deleting its earliest-onset criterion yielded a descriptive net gain of 8 cases versus 11 for the full procedure. The study audits whether agents followed grounded evidence chains, not whether causes are identifiable from telemetry alone. Rule mining is label-supervised, and findings are specific to OpenRCA.

### Source excerpt

arXiv:2607.13548v2 Announce Type: replace Abstract: Root cause analysis (RCA) of microservice failures requires reasoning over telemetry far larger than a model's context, so a language agent reasons over extracted evidence. We start from an error analysis of such an agent on OpenRCA. A reverse reasoning agent reconstructs an evidence chain to each labelled cause and compares it against the agent's own trace, locating failures where a grounded chain existed but was not followed; its labels agree with two site reliability engineers. Being label-conditioned, it audits evidence rather than testing forward identifiability, leaving open what the agent lacked: procedure or telemetry-specific knowledge. Existing systems cannot separate them, extracting evidence inside the reasoning loop. We compute the evidence once, offline, and supply it identically to every arm. Rules mined from the audit reports of one Market cloudbed raise Full Score on a held-out cloudbed from 14.10 to 35.90, surviving correction for multiple testing, against 24.36 for manual rules and 19.23 for a label-and-schema control. The rules add 17 solved cases on top of the five-step procedure but 11 without it, the repairs overlapping. Removing the procedure raises Full Score in all three conditions tested, though no contrast survives correction. Deleting only its earliest-onset criterion yields a net gain of 8 cases against 11 for the whole procedure, which we report descriptively. Mining is label-supervised, and all results are specific to OpenRCA.

## Every Alarm Has a Cause: An Open Generator of Telecom Alarm Storms with Exact Root Causes, Bound to NORIA-O, RFC 8632 and TMF642

DevFeed: [Every Alarm Has a Cause: An Open Generator of Telecom Alarm Storms with Exact Root Causes, Bound to NORIA-O, RFC 8632 and TMF642](<https://devfeed.tech/articles/every-alarm-has-a-cause-an-open-generator-of-telecom-alarm-storms-with-exact-root-causes-bound-to-noria-o-rfc-8632-and-tmf642-95192.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.30004>)

Author: Madhulatha Mandarapu, Sandeep Kunkunuru

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [graph theory](<https://devfeed.tech/topics/graph-theory.md>), [communications](<https://devfeed.tech/topics/communications.md>)

Tags: [alarm](<https://devfeed.tech/tags/alarm.md>), [cause](<https://devfeed.tech/tags/cause.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [generator](<https://devfeed.tech/tags/generator.md>), [open](<https://devfeed.tech/tags/open.md>), [rfc](<https://devfeed.tech/tags/rfc.md>), [root](<https://devfeed.tech/tags/root.md>)

### AI overview

The paper presents an alarm-storm generator on real backbone topologies that records each alarm's exact root cause and propagation path during generation. Emitted fields must bind to terms in NORIA-O, RFC 8632, RFC 8345 or TMF642; unknown terms fail the run. It includes fixed configurations, splits, a scorer and nine baselines. In reported results, ranking the device with the earliest critical alarm achieved top-1 accuracy of 0.650, versus 0.575 for the authors' temporal method; an LLM scored 0.600 with the same view and 0.625 when allowed to query the graph. For site power loss, the PoP is silent: methods ranking alarmed devices scored 0.000 or near it, while two methods that traverse the topology to find a silent candidate scored 1.000 and 0.833. The supplied excerpt does not identify those two methods.

### Source excerpt

arXiv:2610.30004v1 Announce Type: new Abstract: Alarm root-cause analysis is studied on data that cannot be shared. Operator logs are confidential, so what the field publishes is a causal graph between alarm types rather than per-incident causes, a fixed scenario set, gated, or never released. Two recent benchmarks carry per-incident root causes at scale; neither states a download location or a licence. We give a generator rather than a dataset. It runs on real backbone topologies from the Internet Topology Zoo and records the exact root cause and the propagation path of every alarm -- a by-product of generation, not a labelling pass. Every emitted field binds to a term in NORIA-O, RFC 8632, RFC 8345 or TMF642, and an unknown term fails the run. We ship fixed configurations, splits, a scorer and nine baselines. A one-line rule -- rank the device that raised the earliest critical alarm -- takes top-1 at 0.650 against 0.575 for our own temporal method, and wins 13 of 15 runs over three backbones and five seeds. An LLM given the same view reaches 0.600, the same model allowed to query the graph 0.625, and the rule wins on real operator alarms too, at 0.773: a property of alarm streams, not of our generator. One fault class separates the table: when a site loses power its PoP reports nothing, so every method that ranks alarmed devices scores 0.000 or near it, while the two that walk the topology to a silent candidate take 1.000 and 0.833.

## CloudIR-Bench: A Benchmark for Root Cause Analysis and Mitigation of Cloud Incidents

DevFeed: [CloudIR-Bench: A Benchmark for Root Cause Analysis and Mitigation of Cloud Incidents](<https://devfeed.tech/articles/cloudir-bench-a-benchmark-for-root-cause-analysis-and-mitigation-of-cloud-incidents-94629.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.29185>)

Author: Hoang Tran, Huong Ha, Tri Dang, Jeffrey Chan, Hongyu Zhang

Published: 2026-10-22T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [reproducibility](<https://devfeed.tech/topics/reproducibility.md>), [reliability-engineering](<https://devfeed.tech/topics/reliability-engineering.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [incident](<https://devfeed.tech/tags/incident.md>), [incident-report](<https://devfeed.tech/tags/incident-report.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llms](<https://devfeed.tech/tags/llms.md>), [models](<https://devfeed.tech/tags/models.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [production](<https://devfeed.tech/tags/production.md>), [real-world](<https://devfeed.tech/tags/real-world.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [reproducibility](<https://devfeed.tech/tags/reproducibility.md>), [root-cause-analysis](<https://devfeed.tech/tags/root-cause-analysis.md>), [tasks](<https://devfeed.tech/tags/tasks.md>), [zero-shot](<https://devfeed.tech/tags/zero-shot.md>)

### AI overview

CloudIR-Bench addresses the shortage of public data for incident report-based root cause analysis and mitigation prediction. It provides a dataset curated from reports across five cloud-based systems, plus an open-source evaluation framework with eight recent LLM-based approaches. The authors evaluate zero-shot and in-context learning and qualitatively examine where and why models fail. The abstract reports that results characterize current LLM capabilities and limitations, but gives no specific performance figures or model comparisons.

### Source excerpt

arXiv:2610.29185v1 Announce Type: new Abstract: Root cause analysis (RCA) and mitigation prediction are essential for incident management, enabling the identification of underlying causes of failures and ensuring the reliability of complex cloud-based systems. With rising incident volume and system complexity, incident reports have become a valuable resource for these tasks. Recent studies have shown promising results using Large Language Models (LLMs) to predict root causes and mitigation actions from textual descriptions in incident reports. Despite these advances, further progress is hindered by the lack of publicly available datasets; existing work relies on private data, which limits transparency, reproducibility, and generalization. To address this gap, we introduce CloudIR-Bench, the first publicly available benchmark providing a dataset and evaluation framework for incident report-based RCA and mitigation prediction. Curated from incident reports across five leading cloud-based systems, CloudIR-Bench captures a diverse range of real-world production scenarios representing authentic failures and engineers' reasoning. We also release an open-source evaluation framework that includes eight recent LLM-based approaches. Our comprehensive evaluation includes zero-shot and in-context learning experiments, along with a qualitative analysis to identify where and why LLMs fail. Benchmarking results reveal the capabilities and limitations of existing LLMs in this domain. The benchmark, including the dataset, framework and supplementary materials, is available at https://github.com/hoangtnm/CloudIR-Bench.

## "Why Did We Miss Target?" -- Ask Snowflake CoWork, Get the Root-Cause PDF

DevFeed: ["Why Did We Miss Target?" -- Ask Snowflake CoWork, Get the Root-Cause PDF](<https://devfeed.tech/articles/why-did-we-miss-target-ask-snowflake-cowork-get-the-root-cause-pdf-88062.md>)

Original publisher: [Read original article](<https://medium.com/snowflake/why-did-we-miss-target-ask-snowflake-cowork-get-the-root-cause-pdf-1b937dbbc2f7?source=rss----34b6daafc07---4>)

Author: Christopher Vimal Selvadurai

Published: 2026-10-14T19:01:02Z

Content type: tutorial

Language: en

Sources: [Snowflake Builders](<https://devfeed.tech/sources/snowflake-builders.md>)

Topics: [Snowflake streaming data architecture](<https://devfeed.tech/topics/snowflake-streaming-data-architecture.md>), [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>)

Tags: [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [cowork](<https://devfeed.tech/tags/cowork.md>), [data-engineering](<https://devfeed.tech/tags/data-engineering.md>), [ddl](<https://devfeed.tech/tags/ddl.md>), [docker](<https://devfeed.tech/tags/docker.md>), [powerpoint](<https://devfeed.tech/tags/powerpoint.md>), [python](<https://devfeed.tech/tags/python.md>), [rca](<https://devfeed.tech/tags/rca.md>), [snowflake](<https://devfeed.tech/tags/snowflake.md>), [snowflake-cortex-agent](<https://devfeed.tech/tags/snowflake-cortex-agent.md>)

### AI overview

This tutorial describes a Snowflake stored procedure that analyzes missed manufacturing production targets, renders a self-contained HTML root-cause report, and converts it to PDF using WeasyPrint from Snowflake's PyPI Artifact Repository. It explains the report calculations, the limitations encountered with other PDF libraries, and how a Snowflake Cortex Agent can return a download link for the report.

### Source excerpt

Analysts used to pull data from several systems, build a PowerPoint and email versions of it. Now one question returns a production-grade RCA PDF, all inside Snowflake, with no SPCS and no Docker. Every shift change in a manufacturing plant starts with the same questions: why did we miss target, what actually caused it, and can we recover it next shift? A 24/7 line short of its daily production target needs more than an "equipment issues" shrug -- it needs a quantified variance bridge: top causes, impact in units, what's recoverable, what to action first. Today, that takes an analyst and one or two reviewers 3-6 hours, pulling from four source systems and assembling a PowerPoint that usually goes through a revision or two before it ships around noon. By then, the next shift has already started blind. We built something different: one stored procedure call that queries the data, computes the root-cause analysis, renders a styled HTML report, converts it to a polished PDF, and hands back a download link -- entirely inside Snowflake. Now, the shift supervisor or plant manager simply asks in CoWork (Snowflake's personal agent for knowledge workers, formerly Snowflake Intelligence) and get a finished PDF back in 5 minutes, in time to brief the incoming shift before it starts, not after. That frees the analyst from manual report-building to focus on higher-value work: strengthening the underlying data and driving adoption of the system across the plant. Here's how we built it, the dead ends we hit along the way, and the surprisingly simple fix that made it work. GitHub repo: The complete code -- engine, sample data generator, and DDL -- is available at repo. Clone it, point it at your own views, and adapt the reason-bucketing to your domain. Why We Started With a Spec, Not Code Before writing a line of code, we defined what the system must do using structured requirements. This isn't ceremony for its own sake -- it's about making technology choices testable instead of hoping t

## HN800: Root Cause Analysis for the Entire Stack

DevFeed: [HN800: Root Cause Analysis for the Entire Stack](<https://devfeed.tech/articles/hn800-root-cause-analysis-for-the-entire-stack-66097.md>)

Original publisher: [Read original article](<https://packetpushers.net/podcasts/heavy-networking/hn800-root-cause-analysis-for-the-entire-stack-sponsored/>)

Author: Heavy Networking Ep ; October

Published: 2025-10-10T12:00:00Z

Content type: article

Language: en

Sources: [Automate Your Network](<https://devfeed.tech/sources/automate-your-network.md>)

Topics: [atlassian automated rca](<https://devfeed.tech/topics/atlassian-automated-rca.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>), [Service fabric](<https://devfeed.tech/topics/service-fabric.md>)

Tags: [apm](<https://devfeed.tech/tags/apm.md>), [appearance](<https://devfeed.tech/tags/appearance.md>), [automation](<https://devfeed.tech/tags/automation.md>), [back-end](<https://devfeed.tech/tags/back-end.md>), [bgp](<https://devfeed.tech/tags/bgp.md>), [root-cause-analysis](<https://devfeed.tech/tags/root-cause-analysis.md>)

### AI overview

This sponsored podcast episode discusses how Selector applies AI to full-stack root cause analysis across network, infrastructure, and application telemetry. The guests describe customer-specific data preparation and contextual metadata, correlated alerts, natural-language queries, and integrations with automation workflows. Examples include Kubernetes resource remediation and consolidating many related network tickets into a single actionable incident.

### Source excerpt

Packet Pushers - Podcast - Network automation