<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Rahul Bansod on Medium]]></title>
        <description><![CDATA[Stories by Rahul Bansod on Medium]]></description>
        <link>https://medium.com/@rahulbansod519?source=rss-162cd635fe74------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*UTjAwH6vLf7s1cEUgc37tg.png</url>
            <title>Stories by Rahul Bansod on Medium</title>
            <link>https://medium.com/@rahulbansod519?source=rss-162cd635fe74------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 12:28:43 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rahulbansod519/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Role of Data Structures and Algorithms in AI: Why Every Aspiring Developer Should Master Them]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">Artificial Intelligence looks glamorous from the outside &#x2014; neural networks, generative models, autonomous systems.</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/the-role-of-data-structures-and-algorithms-in-ai-why-every-aspiring-developer-should-master-them-57214b903aac?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/the-role-of-data-structures-and-algorithms-in-ai-why-every-aspiring-developer-should-master-them-57214b903aac?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/57214b903aac</guid>
            <category><![CDATA[data-structure-algorithm]]></category>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Sun, 01 Mar 2026 04:56:35 GMT</pubDate>
            <atom:updated>2026-03-01T04:56:35.485Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Building a Kubernetes Troubleshooting Copilot using Elasticsearch Vector Search, Logs, and RAG]]></title>
            <link>https://medium.com/@rahulbansod519/building-a-kubernetes-troubleshooting-copilot-using-elasticsearch-vector-search-logs-and-rag-26f58731001b?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/26f58731001b</guid>
            <category><![CDATA[vector-search]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[elasticsearch]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Sat, 28 Feb 2026 03:34:54 GMT</pubDate>
            <atom:updated>2026-02-28T03:44:06.976Z</atom:updated>
            <content:encoded><![CDATA[<p>This blog post was submitted to the Elastic Blogathon Contest and is eligible to win a prize.</p><h3>Introduction</h3><p>Kubernetes troubleshooting is rarely a knowledge problem.<br>It is a <strong>signal discovery problem</strong>.</p><p>When a pod enters CrashLoopBackOff, the information needed to diagnose it already exists, but it is scattered across:</p><ul><li>container logs</li><li>Kubernetes events</li><li>documentation</li><li>previous incident patterns</li></ul><p>Engineers spend significant time switching between kubectl describe, kubectl logs, documentation pages, and search engines to identify the root cause.</p><p>This project explores a different approach:</p><p><strong>What if Elasticsearch could act as the memory layer for an AI SRE assistant?</strong></p><p>I built a Kubernetes Troubleshooting Copilot that uses:</p><ul><li>Elasticsearch vector search (retrieval)</li><li>Kubernetes documentation (knowledge)</li><li>Cluster logs (real operational signals)</li><li>LLM reasoning (explanation)</li></ul><p>Instead of searching keywords, the system searches <em>meaning</em>.</p><h3>The Problem</h3><p>Consider a very common production scenario.</p><p>A deployment succeeds, but the pod repeatedly restarts:</p><p>CrashLoopBackOff</p><p>Back-off restarting the failed container</p><p>Typical debugging workflow:</p><ol><li>kubectl get pods</li><li>kubectl describe pod</li><li>kubectl logs</li><li>Google search error message</li><li>Open multiple Kubernetes documentation pages</li><li>Compare with previous incidents</li></ol><p>This workflow is a manual correlation across multiple data sources.</p><p>The information exists, but the engineer must assemble it.</p><p>This project attempts to automate that reasoning process.</p><h3>System Architecture</h3><p>The copilot follows a Retrieval Augmented Generation (RAG) architecture.</p><p>User Question</p><p>↓</p><p>Embedding</p><p>↓</p><p>Elasticsearch Vector Search</p><p>↓</p><p>Retrieve Logs + Documentation</p><p>↓</p><p>LLM Reasoning</p><p>↓</p><p>Root Cause Explanation</p><h3>Components</h3><p><strong>1. Elasticsearch (Vector Database)</strong></p><ul><li>Stores documentation embeddings</li><li>Stores Kubernetes cluster logs</li><li>Performs semantic similarity search</li></ul><p><strong>2. Documentation Knowledge Base<br></strong>Official Kubernetes documentation was ingested and converted into plain text, chunked, and embedded.</p><p><strong>3. Log Pipeline<br></strong>Fluent Bit collects container logs and sends them to Elasticsearch.</p><p><strong>4. Embedding Model<br></strong>all-MiniLM-L6-v2 (384-dimension sentence embeddings)</p><p><strong>5. LLM Layer<br></strong>Used only to explain results retrieved from Elasticsearch.</p><p>Important:<br>The LLM does not “know Kubernetes”.<br>It only reasons over retrieved operational evidence.</p><h3>Building the Knowledge Base</h3><p>Kubernetes documentation was mirrored locally and converted to raw text.<br>The documents were then chunked into ~900 character segments with overlap to preserve semantic context.</p><p>Each chunk was embedded and indexed in Elasticsearch as a dense_vector field.</p><p>Example mapping:</p><p>{</p><p>“content”: “text”,</p><p>“embedding”: {</p><p>“type”: “dense_vector”,</p><p>“dims”: 384,</p><p>“similarity”: “cosine”</p><p>},</p><p>“source”: “filename”</p><p>}</p><p>This transforms Elasticsearch from a keyword search engine into a semantic retrieval engine.</p><h3>Ingesting Kubernetes Logs</h3><p>To move beyond a documentation chatbot, real cluster signals were required.</p><p>A failing pod was intentionally created:</p><ul><li>container exits with code 1</li><li>restart policy: Always</li></ul><p>This produced a CrashLoopBackOff condition.</p><p>Fluent Bit was deployed as a DaemonSet to ship container logs to Elasticsearch.</p><p>Now, Elasticsearch is contained:</p><ul><li>documentation knowledge</li><li>real operational telemetry</li></ul><h3>Semantic Retrieval</h3><p>Instead of searching for the keyword “CrashLoopBackOff”, the system embeds the user’s question.</p><p>Example query:</p><p>“Why is my pod restarting after deployment?”</p><p>The embedding is compared against stored vectors using cosine similarity.</p><p>The retriever returns:</p><ul><li>relevant documentation sections</li><li>related failure log entries</li></ul><p>This step is critical.</p><p>The LLM is not guessing.<br>It is reasoning over evidence retrieved from Elasticsearch.</p><h3>Retrieval Augmented Reasoning</h3><p>The retrieved context is sent to the LLM with strict instructions:</p><ul><li>Use only provided context</li><li>identify root cause</li><li>Provide verification steps</li><li>Provide remediation steps</li></ul><p>Because logs are included, the system can correlate:</p><p>documentation → theory<br>logs → actual failure</p><p>Example reasoning produced:</p><ul><li>container exits non-zero</li><li>Restart policy triggers a restart</li><li>kubelet back-off behavior</li><li>debugging commands to verify</li></ul><h3>Why Vector Search Matters</h3><p>Keyword search fails in operational debugging.</p><p>A user may ask:</p><p>“My container keeps restarting every 10 seconds.”</p><p>The documentation might only contain:</p><p>“restartPolicy behavior after container failure”</p><p>There is no keyword overlap.</p><p>Vector search solves this by matching semantic meaning rather than text.</p><p>Elasticsearch becomes a troubleshooting memory rather than a log archive.</p><h3>Observations</h3><h3>What worked well</h3><ul><li>Dense vector retrieval consistently located relevant lifecycle documentation</li><li>Combining logs and docs improved explanation accuracy</li><li>Elasticsearch handled both structured logs and unstructured documentation effectively</li></ul><h3>Challenges</h3><ul><li>Raw documentation ingestion contained navigation noise</li><li>Chunk sizing strongly affected retrieval quality</li><li>Logs alone were insufficient without a documentation context</li></ul><h3>Why This Is Useful</h3><p>This system does not replace engineers.</p><p>It reduces the <em>time to understanding</em>.</p><p>Instead of:</p><p>search → read → correlate → infer</p><p>The engineer gets:</p><p>evidence → reasoning → verification steps</p><p>In real environments, this could shorten incident triage significantly, especially for junior engineers or on-call rotations.</p><h3>Future Work</h3><p>Possible extensions:</p><ul><li>integrate Kubernetes events API</li><li>index previous incidents as organizational memory</li><li>Slack bot integration</li><li>Anomaly detection using Elastic Observability</li></ul><h3>Conclusion</h3><p>Elasticsearch is commonly viewed as a log storage or search tool.</p><p>This project demonstrates a different role:</p><p><strong>Elasticsearch as the memory layer for operational AI systems.</strong></p><p>By combining:</p><ul><li>vector search</li><li>cluster telemetry</li><li>documentation knowledge</li></ul><p>We can build systems that do not just store data, but help engineers understand systems.</p><p>The value of GenAI in infrastructure is not chat; it is <em>assisted reasoning over operational data</em>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=26f58731001b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Kubernetes API Server Crash Caused by Excessive CRD Writes (Real Production Incident)]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/kubernetes-api-server-crash-caused-by-excessive-crd-writes-real-production-incident-a53b495dc287?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/2600/1*fMtvp97sWdbS_DohhgAuaA.png" width="2816"></a></p><p class="medium-feed-snippet">Kubernetes API server crashes are rare&#x200A;&#x2014;&#x200A;but when they happen, the cause is often self-inflicted.</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/kubernetes-api-server-crash-caused-by-excessive-crd-writes-real-production-incident-a53b495dc287?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/kubernetes-api-server-crash-caused-by-excessive-crd-writes-real-production-incident-a53b495dc287?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/a53b495dc287</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[solutions]]></category>
            <category><![CDATA[issues]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[crash]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Thu, 18 Dec 2025 03:32:36 GMT</pubDate>
            <atom:updated>2025-12-18T03:32:36.533Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[When Zombie Pods Broke Our Kubernetes Node Upgrade]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/when-zombie-pods-broke-our-kubernetes-node-upgrade-d354dc960291?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/2600/1*Ucm8Cvb8eFjGJp1kKfj3sw.png" width="2816"></a></p><p class="medium-feed-snippet">Problem category: Kubernetes reliability / operational failure
&#xA0;Impact: Node upgrade stalled, autoscaling blocked, manual intervention&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/when-zombie-pods-broke-our-kubernetes-node-upgrade-d354dc960291?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/when-zombie-pods-broke-our-kubernetes-node-upgrade-d354dc960291?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/d354dc960291</guid>
            <category><![CDATA[operations]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[devops-practice]]></category>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Wed, 17 Dec 2025 03:51:27 GMT</pubDate>
            <atom:updated>2025-12-17T03:51:27.225Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Mental Model to Write Any Level of SQL Query — Without Fear]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/mental-model-to-write-any-level-of-sql-query-without-fear-f93194a249ca?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/1024/1*bjS_JAbd8pOvknkd9FiXvg.png" width="1024"></a></p><p class="medium-feed-snippet">If SQL makes you nervous, you&#x2019;re not alone.
&#xA0;Complex queries, endless joins, mysterious errors&#x200A;&#x2014;&#x200A;it&#x2019;s easy to feel overwhelmed.</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/mental-model-to-write-any-level-of-sql-query-without-fear-f93194a249ca?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/mental-model-to-write-any-level-of-sql-query-without-fear-f93194a249ca?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/f93194a249ca</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[database]]></category>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Tue, 28 Oct 2025 07:00:52 GMT</pubDate>
            <atom:updated>2025-10-28T07:00:52.507Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Peeking Under the Hood of oc debug in OpenShift: The Journey Behind a Single Command]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/peeking-under-the-hood-of-oc-debug-in-openshift-the-journey-behind-a-single-command-910a7bf86fb2?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/683/1*DqDuc_uDDsLoYYtFZk0ESQ.png" width="683"></a></p><p class="medium-feed-snippet">When facing a tricky application or node issue in OpenShift, the oc debug command feels like magic&#x2014;launch it, and you&#x2019;re instantly&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/peeking-under-the-hood-of-oc-debug-in-openshift-the-journey-behind-a-single-command-910a7bf86fb2?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/peeking-under-the-hood-of-oc-debug-in-openshift-the-journey-behind-a-single-command-910a7bf86fb2?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/910a7bf86fb2</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[openshift]]></category>
            <category><![CDATA[api]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Fri, 12 Sep 2025 07:20:00 GMT</pubDate>
            <atom:updated>2025-09-12T07:20:00.842Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[ Exploring OpenShift Routes & Ingress in My CRC Lab: A Day of Debugging, Learning, and Wins]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/exploring-openshift-routes-ingress-in-my-crc-lab-a-day-of-debugging-learning-and-wins-bd5d14e2c31f?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/1518/1*qLOYg2BSCUlHFf_z9ynt4w.png" width="1518"></a></p><p class="medium-feed-snippet">Today I rolled up my sleeves and decided to really understand OpenShift Routes &amp; Ingress. Instead of just reading docs, I fired up my CRC&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/exploring-openshift-routes-ingress-in-my-crc-lab-a-day-of-debugging-learning-and-wins-bd5d14e2c31f?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/exploring-openshift-routes-ingress-in-my-crc-lab-a-day-of-debugging-learning-and-wins-bd5d14e2c31f?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/bd5d14e2c31f</guid>
            <category><![CDATA[nginx]]></category>
            <category><![CDATA[red-hat-openshift]]></category>
            <category><![CDATA[openshift]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Thu, 21 Aug 2025 12:07:12 GMT</pubDate>
            <atom:updated>2025-08-21T12:07:12.621Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Implementing Horizontal Pod Autoscaling (HPA) for a Python API in Kubernetes]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">Introduction</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/implementing-horizontal-pod-autoscaling-hpa-for-a-python-api-in-kubernetes-52a42acfca98?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/implementing-horizontal-pod-autoscaling-hpa-for-a-python-api-in-kubernetes-52a42acfca98?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/52a42acfca98</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[api]]></category>
            <category><![CDATA[scalling]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Fri, 14 Mar 2025 16:11:22 GMT</pubDate>
            <atom:updated>2025-03-14T16:19:25.377Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Kubernetes: kubectl Command Cheat Sheet for Beginners]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/kubernetes-kubectl-command-cheat-sheet-for-beginners-3704bac5bfb6?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/799/1*IZ_w71dY6-_L1EwkEGwN0w.png" width="799"></a></p><p class="medium-feed-snippet">If you&#x2019;re new to Kubernetes, one of the first tools you&#x2019;ll encounter is kubectl. As Kubernetes&apos; command-line interface, kubectl is&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/kubernetes-kubectl-command-cheat-sheet-for-beginners-3704bac5bfb6?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/kubernetes-kubectl-command-cheat-sheet-for-beginners-3704bac5bfb6?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/3704bac5bfb6</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[cloud]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Fri, 10 Jan 2025 16:21:16 GMT</pubDate>
            <atom:updated>2025-01-10T16:21:16.632Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Step-by-Step Guide to Deploying Your First Application on Kubernetes]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@rahulbansod519/step-by-step-guide-to-deploying-your-first-application-on-kubernetes-447600866dca?source=rss-162cd635fe74------2"><img src="https://cdn-images-1.medium.com/max/722/1*AQe5WdfypLY1qXTHgpVsXA.png" width="722"></a></p><p class="medium-feed-snippet">Are you ready to take your first steps into the world of Kubernetes? Deploying your first application on Kubernetes can seem daunting, but&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@rahulbansod519/step-by-step-guide-to-deploying-your-first-application-on-kubernetes-447600866dca?source=rss-162cd635fe74------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@rahulbansod519/step-by-step-guide-to-deploying-your-first-application-on-kubernetes-447600866dca?source=rss-162cd635fe74------2</link>
            <guid isPermaLink="false">https://medium.com/p/447600866dca</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[deployment]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[computer-science]]></category>
            <dc:creator><![CDATA[Rahul Bansod]]></dc:creator>
            <pubDate>Wed, 08 Jan 2025 17:16:21 GMT</pubDate>
            <atom:updated>2025-01-10T14:45:58.256Z</atom:updated>
        </item>
    </channel>
</rss>