<?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 Akshay Jain  on Medium]]></title>
        <description><![CDATA[Stories by Akshay Jain  on Medium]]></description>
        <link>https://medium.com/@akshayjn93?source=rss-56cedc6fb484------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*wpAkgCEm-2GTP6Avl_C3eQ.jpeg</url>
            <title>Stories by Akshay Jain  on Medium</title>
            <link>https://medium.com/@akshayjn93?source=rss-56cedc6fb484------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 05:22:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@akshayjn93/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[ Migration from AWS VPC CNI to Cilium CNI on EKS]]></title>
            <link>https://medium.com/@akshayjn93/migration-from-aws-vpc-cni-to-cilium-cni-on-eks-08f0fda47332?source=rss-56cedc6fb484------2</link>
            <guid isPermaLink="false">https://medium.com/p/08f0fda47332</guid>
            <category><![CDATA[service-mesh]]></category>
            <category><![CDATA[cilium]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Akshay Jain ]]></dc:creator>
            <pubDate>Thu, 28 Aug 2025 18:19:35 GMT</pubDate>
            <atom:updated>2025-08-28T18:19:35.038Z</atom:updated>
            <content:encoded><![CDATA[<p>Migrating your EKS cluster from AWS VPC CNI to Cilium CNI can provide advanced networking features, including enhanced observability, security policies, and eBPF-based performance improvements. Below are the steps I took to safely migrate to Cilium while minimizing downtime.</p><h3>Step 1: Label Existing Nodes</h3><p>Label all existing nodes to indicate they are using AWS VPC CNI:</p><pre>kubectl label nodes --all cni=aws-cni</pre><h3>Step 2: Patch AWS VPC CNI and Kube Proxy</h3><p>Ensure aws-node and kube-proxy DaemonSets are scheduled <strong>only on nodes labeled</strong> with cni=aws-cni.</p><p>Patch both DaemonSets with this node affinity:</p><pre>affinity:<br>  nodeAffinity:<br>    requiredDuringSchedulingIgnoredDuringExecution:<br>      nodeSelectorTerms:<br>        - matchExpressions:<br>            - key: cni<br>              operator: In<br>              values:<br>                - aws-cni</pre><h3>Step 3: Scale Up Node Group</h3><p>Add new nodes <strong>without the </strong><strong>cni=aws-cni label</strong>. These will:</p><ul><li>Not run aws-node or kube-proxy.</li><li>Initially be in NotReady state until Cilium is installed.</li></ul><h3>Step 4: Install Cilium</h3><p>Install Cilium on the new nodes, ensuring it’s scheduled <strong>only on nodes that do not have the </strong><strong>cni=aws-cni label</strong>.</p><p>Here’s the values.yaml configuration I used (key section only):</p><pre>USER-SUPPLIED VALUES:<br>affinity:<br>  nodeAffinity:<br>    requiredDuringSchedulingIgnoredDuringExecution:<br>      nodeSelectorTerms:<br>      - matchExpressions:<br>        - key: cni<br>          operator: NotIn<br>          values:<br>          - aws-cni<br>bpf:<br>  hostLegacyRouting: true<br>cluster:<br>  name: &lt;Eks cluster name&gt;<br>egressMasqueradeInterfaces: eth+<br>endpointRoutes:<br>  enabled: true<br>eni:<br>  enabled: true<br>envoy:<br>  affinity:<br>    nodeAffinity:<br>      requiredDuringSchedulingIgnoredDuringExecution:<br>        nodeSelectorTerms:<br>        - matchExpressions:<br>          - key: cni<br>            operator: NotIn<br>            values:<br>            - aws-cni<br>hubble:<br>  relay:<br>    enabled: true<br>    rollOutPods: true<br>  ui:<br>    enabled: true<br>    rollOutPods: true<br>ipam:<br>  mode: eni<br>k8sServiceHost: &lt;API SERVER ENDPOINT&gt;<br>k8sServicePort: 443<br>kubeProxyReplacement: true<br>kubeProxyReplacementHealthzBindAddr: 0.0.0.0:10256<br>operator:<br>  prometheus:<br>    enabled: true<br>    serviceMonitor:<br>      enabled: true<br>policyEnforcementMode: default<br>routingMode: native<br>serviceAccounts:<br>  cilium:<br>    name: cilium<br>  operator:<br>    name: cilium-operator</pre><p>Install Cilium using Helm:</p><pre>helm install cilium cilium/cilium --version 1.17.1 -f values.yaml</pre><h3>Step 5: Migrate Workloads</h3><p>Begin migrating your workloads to the new Cilium nodes.</p><p>Validate the connectivity with:</p><pre>kubectl create ns cilium-test<br>kubectl apply --namespace=cilium-test -f \<br>https://raw.githubusercontent.com/cilium/cilium/1.16.3/examples/kubernetes/connectivity-check/connectivity-check.yaml</pre><h3>Step 6: Clean Up Legacy Components</h3><p>Once everything is migrated and tested:</p><ul><li>Delete aws-node and kube-proxy DaemonSets (via Terraform if managed).</li><li>Optionally, taint or cordon the old AWS CNI nodes if planning to decommission.</li></ul><p><a href="https://hackmd.io/@Echo-Live/106#Migration-Strategies">https://hackmd.io/@Echo-Live/106#Migration-Strategies</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=08f0fda47332" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Issues Faced During Migration from VPC CNI to Cilium CNI on EKS]]></title>
            <link>https://medium.com/@akshayjn93/issues-faced-during-migration-from-vpc-cni-to-cilium-cni-on-eks-6d79c7b8e339?source=rss-56cedc6fb484------2</link>
            <guid isPermaLink="false">https://medium.com/p/6d79c7b8e339</guid>
            <dc:creator><![CDATA[Akshay Jain ]]></dc:creator>
            <pubDate>Sun, 25 May 2025 17:52:47 GMT</pubDate>
            <atom:updated>2025-05-25T17:52:47.849Z</atom:updated>
            <content:encoded><![CDATA[<h3>🛠️ Migration Setup</h3><ul><li><strong>Migrating from</strong>: AWS VPC CNI</li><li><strong>Migrating to</strong>: Cilium CNI on EKS</li><li><strong>IPAM Mode</strong>: cluster-pool</li><li><strong>Per-node Pod CIDR</strong>: 10.245.0.0/16</li><li><strong>Overlay Mode</strong>: VXLAN</li><li><strong>Tunnel Port</strong>: 8473</li><li><strong>BPF Host Routing</strong>: Enabled (hostLegacyRouting: true)</li><li><strong>Policy Enforcement</strong>: Disabled (policyEnforcementMode: never)</li><li><strong>Custom CNI Conf</strong>: True (i.e., Cilium doesn’t overwrite existing CNI configs)</li></ul><p>🧾 Cilium Config Used</p><pre>operator:<br>  unmanagedPodWatcher:<br>    restart: false<br>tunnel: vxlan<br>tunnelPort: 8473<br>cni:<br>  customConf: true<br>  uninstall: false<br>ipam:<br>  mode: &quot;cluster-pool&quot;<br>  operator:<br>    clusterPoolIPv4PodCIDRList: [&quot;10.245.0.0/16&quot;]<br>policyEnforcementMode: &quot;never&quot;<br>bpf:<br>  hostLegacyRouting: true</pre><h3>❗ Issues Observed during the migration</h3><h4>1. Pods Unable to Connect to External Endpoints</h4><ul><li>Only those pods that needed to <strong>communicate with external services</strong> (e.g., third-party APIs, webhooks, public internet) were impacted.</li><li>These pods failed during startup or operation due to the <strong>inability to reach external IPs</strong>.</li></ul><p>Examples:</p><ul><li>Webhook servers hosted outside the cluster</li><li>Applications pulling external configs or dependencies</li><li>Services performing API calls to SaaS platforms</li></ul><h4>2. Webhooks (External) Failing</h4><ul><li>Kubernetes <strong>mutating/validating admission webhooks</strong> pointing to <strong>external URLs</strong> became unreachable.</li><li>This led to pod creation delays or failures due to timeouts during webhook validation.</li></ul><blockquote>The above Cilium configuration is <strong>not suitable for zero-downtime migration</strong> from VPC CNI, especially in clusters using components like <strong>Vault Agent Injector</strong> and <strong>Karpenter</strong>.</blockquote><blockquote>In our case, both Vault Agent Injector and Karpenter deploy <strong>mutating/validating admission webhooks</strong>, and after enabling Cilium with a separate CIDR (10.245.0.0/16 in cluster-pool mode), the <strong>API server was unable to reach these webhook services</strong>. This caused failures in pod creation and other operations that rely on these webhooks.</blockquote><h4>Reference Links:</h4><p><a href="https://isovalent.com/blog/post/tutorial-migrating-to-cilium-part-1/">Tutorial: How to Migrate to Cilium - Isovalent</a></p><p><a href="https://docs.cilium.io/en/stable/installation/k8s-install-migration/#migration-procedure">https://docs.cilium.io/en/stable/installation/k8s-install-migration/#migration-procedure</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6d79c7b8e339" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hubble: A Deep Dive into Cilium’s Observability Dashboard]]></title>
            <link>https://medium.com/@akshayjn93/hubble-a-deep-dive-into-ciliums-observability-dashboard-dea419c34577?source=rss-56cedc6fb484------2</link>
            <guid isPermaLink="false">https://medium.com/p/dea419c34577</guid>
            <category><![CDATA[cilium]]></category>
            <category><![CDATA[hubble]]></category>
            <category><![CDATA[service-mesh]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[cni]]></category>
            <dc:creator><![CDATA[Akshay Jain ]]></dc:creator>
            <pubDate>Sun, 03 Nov 2024 13:10:18 GMT</pubDate>
            <atom:updated>2024-11-03T13:10:18.824Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Hubble</strong> is a fully distributed networking and security observability platform. It is built on top of <strong>Cilium</strong> and <strong>eBPF</strong> to enable deep visibility into the communication and behavior of services as well as the networking infrastructure in a completely transparent manner.</p><p>By building on top of <strong>Cilium</strong>, <strong>Hubble</strong> can leverage <strong>eBPF</strong> for visibility. By relying on <strong>eBPF</strong>, all visibility is programmable and allows for a dynamic approach that minimizes overhead while providing deep and detailed visibility as required by users. <strong>Hubble</strong> has been created and specifically designed to make the best use of these new eBPF powers.</p><p><strong>If you’re looking for answers to the questions below, Hubble can help.</strong></p><ul><li>What services are communicating with each other? How often? What is the appearance of the service dependency graph?</li><li>What HTTP calls are being made? What Kafka topics does a service consume from or produce to?</li><li>Is any network communication failing? Why is communication failing? Is it DNS? Is it an application or network problem? Is the communication broken on layer 4 (TCP) or layer 7 (HTTP)?</li><li>Which services have experienced a DNS resolution problem in the last 5 minutes? Which services have experienced an interrupted TCP connection recently or have seen connections timing out? What is the rate of unanswered TCP SYN requests?</li><li>What is the rate of 5xx or 4xx HTTP response codes for a particular service or across all clusters?</li><li>What is the 95th and 99th percentile latency between HTTP requests and responses in my cluster? Which services are performing the worst? What is the latency between the two services?</li><li>Which services had connections blocked due to network policy? What services have been accessed from outside the cluster? Which services have resolved a particular DNS name?</li></ul><blockquote>In short, Hubble is an observability platform specifically written for Cilium. It provides service dependency maps, operational monitoring and alerting, and application and security visibility based on flow logs.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dea419c34577" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cilium’s Next-Generation Service Mesh]]></title>
            <link>https://medium.com/@akshayjn93/ciliums-next-generation-service-mesh-68d04392a945?source=rss-56cedc6fb484------2</link>
            <guid isPermaLink="false">https://medium.com/p/68d04392a945</guid>
            <category><![CDATA[service-mesh]]></category>
            <category><![CDATA[open-source]]></category>
            <category><![CDATA[cni]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[cilium]]></category>
            <dc:creator><![CDATA[Akshay Jain ]]></dc:creator>
            <pubDate>Thu, 31 Oct 2024 22:02:04 GMT</pubDate>
            <atom:updated>2024-10-31T22:02:04.668Z</atom:updated>
            <content:encoded><![CDATA[<p>Cilium is open-source software for transparently securing the network connectivity between application services deployed using Linux container management platforms like Docker and Kubernetes.</p><p>Traditional service meshes, despite their benefits, can present significant challenges. These include the complexity and error-prone nature of IP and port-based network policies, performance overhead due to their proxy-based architecture, limited granularity in the visibility of service-to-service communication, interoperability issues with existing infrastructure, scalability challenges as the number of services and traffic volumes increase, and operational and resource overhead.</p><p>Cilium Service Mesh redefines traditional service mesh frameworks by integrating the mesh layer directly into the kernel using eBPF, thus eliminating the need for sidecar proxies. It manages connectivity at both the networking and application protocol layers, handling protocols like IP, TCP, UDP, HTTP, Kafka, gRPC, and DNS with greater efficiency.</p><p>At the heart of Cilium is eBPF, a revolutionary technology built into the Linux kernel. With eBPF, Cilium delivers lightning-fast, efficient, and scalable networking. This bypasses the performance drawbacks of traditional proxies, enabling direct and efficient communication between your services.</p><p>With Cilium Service Mesh, users now have the choice to run a service mesh with sidecars or without them, based on their specific requirements and constraints. This flexibility reduces the complexity and overhead impact of sidecars.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=68d04392a945" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>