{"id":925,"date":"2026-11-14T20:25:00","date_gmt":"2026-11-14T20:25:00","guid":{"rendered":"http:\/\/desinri.com\/blog\/?p=925"},"modified":"2026-11-15T15:25:00","modified_gmt":"2026-11-15T15:25:00","slug":"kubernetessecops-travel-desk-story","status":"publish","type":"post","link":"https:\/\/desinri.com\/blog\/kubernetessecops-travel-desk-story\/","title":{"rendered":"A Travel Desk Learns KubernetesSecOps: A Fictional Tale"},"content":{"rendered":"<p>The story below is an invented one, a sketch assembled for illustration, not drawn from any real firm, though anyone who has worked in a busy travel-services office will recognise the rhythm. A small team handles passport scans, itineraries and application paperwork for hundreds of clients. Their client portal runs on a Kubernetes cluster. And one quiet Tuesday they discover that the cluster has been behaving like a border post with every gate propped open. What they did next is a neat illustration of KubernetesSecOps, the day-to-day work of keeping a cluster&#8217;s access, workloads, connections and secrets under control.<\/p>\n<p>I have used the Ops Atlas as the factual spine of the tale. Its detailed <a href=\"https:\/\/www.xopsschool.com\/kubernetessecops\/\">page on KubernetesSecOps<\/a> is where the technical points come from; the characters, the office and the sequence of events are made up for illustration.<\/p>\n<h2>The Tuesday it started<\/h2>\n<p>Our fictional team has five people: two coordinators, a documents specialist, a developer who looks after the portal, and a manager. A client asks a simple question during a call: &#8220;Who in your company can see my passport scan?&#8221; The manager answers with confidence. The developer goes quiet. Later that afternoon he sits down and writes the real answer, and it takes him three hours because he has to trace it through the cluster.<\/p>\n<p>What he finds is typical. The portal, a document storage service and a notification service all share a cluster. Seven people and two automation accounts hold broad administrator rights. The credentials for the storage service sit in the cluster as plain Secret objects, which, as the Atlas explains, are stored unencrypted in the cluster database unless encryption is configured. Any service in the cluster can contact any other. And nobody switched on audit logging, so there is no record of who has looked at what.<\/p>\n<h2>First, a border officer for people<\/h2>\n<p>The developer starts with the question the client asked: who may do what. In cluster terms that is role-based access control. He retires shared logins and connects the cluster to the company&#8217;s sign-in system, so that when a coordinator leaves, her access ends with her employment. He removes anybody from the built-in group that skips every permission check. Then he turns to the automation account used for deployments, which holds full administrator rights, and replaces it with a role limited to the portal&#8217;s own namespace. The Atlas lists cluster-admin for pipelines as a classic anti-pattern, since whoever controls the pipeline controls the whole place.<\/p>\n<p>He also reviews a handful of grants that look harmless. One role lets a support script list secrets across the cluster. Another allows opening a shell inside running services. Neither is needed day to day. He removes the first and moves the second to a break-glass role that only the manager can activate, with a note in the log.<\/p>\n<h2>Then a customs check on what arrives<\/h2>\n<p>Next, the workloads. At a border, travellers are checked before entering; in a cluster, a similar inspection is called admission control. The team picks the built-in Pod Security feature and starts in warning mode, as the Atlas advises. For two weeks developers see messages about containers that run as root or ask for extra privileges. On a date announced in advance, the checkpoint begins refusing the baseline violations. The developer keeps a short list of waivers, every one naming its owner and its expiry, so that a temporary workaround does not become the permanent rule.<\/p>\n<p>He learns two things the hard way. Enforcement only applies to workloads created afterwards, so older services keep running until restarted; he schedules a clean-up week. And a rule written against a Deployment object does nothing, because the check applies to the pods beneath it. Both points appear in the Atlas&#8217;s list of traps.<\/p>\n<h2>Closing the internal doors<\/h2>\n<p>Like an airport terminal where passengers can wander into any corridor, the cluster let each service reach every other. The developer writes default-deny network policies for the documents namespace and then opens only the paths needed: the portal may call the storage service, the storage service may call its database, and nothing else. Before starting he checks that the network plugin really enforces policies. He also notes that connections already open when a rule lands may carry on for a while, so he restarts the busiest services after the change.<\/p>\n<p>Along the way he finds out what namespaces are not. The notification service is in a different namespace from the documents service, and he had assumed that meant isolation. It does not. They can share a machine and a kernel, so namespaces are a way of grouping and scoping permissions rather than a hard barrier.<\/p>\n<h2>Locking the filing cabinet<\/h2>\n<p>For the passport scans&#8217; storage credentials, he turns on encryption at rest through a key-management integration, choosing the current version rather than the one the Atlas describes as deprecated and then disabled by default. He then re-saves the existing Secrets so the old entries are encrypted as well. Finally, he moves the most sensitive keys into an external vault, so the cluster itself keeps the bare minimum. Readers whose own work touches residency or visa paperwork may find the care he takes familiar; our explainer on <a href=\"http:\/\/desinri.com\/blog\/understanding-permanent-residency-eligibility-before-you-apply-abroad\/\">checking residency eligibility before you apply<\/a> stresses the same habit of verifying before committing.<\/p>\n<h2>A visitors&#8217; book that someone actually reads<\/h2>\n<p>Audit logging comes next. Because the cluster is self-managed, nothing is recorded until a policy file exists. He writes one, careful to remember that the first matching rule wins and that secrets and config maps should be logged at metadata level only. Then he installs a runtime detection tool and finds, within a day, that it is raising far too many alerts. Using the Atlas&#8217;s worked example as a guide, he excludes the maintenance sessions that are already audited and gives the remaining rules an owner. The queue becomes small enough that a person reads it every morning.<\/p>\n<h2>A week in the life afterwards<\/h2>\n<p>It is worth describing what the new routine looks like, because it is deliberately dull. On Monday the developer reads the list of refused deployments and the previous week&#8217;s alerts, which takes about twenty minutes. Once a month the manager joins him to go through the waivers and either renew or close each one. Every quarter they review who holds which role and run the benchmark scanner again, saving the report with its date. Twice a year the coordinators join a short drill in which they practise what they would do if a client reported a suspicious message. Nobody enjoys these meetings, and nobody has regretted them either.<\/p>\n<h2>What the office learned<\/h2>\n<ul>\n<li>The answer to &#8220;who can see this?&#8221; should come from evidence, with a date, not from confidence.<\/li>\n<li>Defaults favour convenience. The team had to turn security on, rule by rule.<\/li>\n<li>Exceptions are normal, provided they expire.<\/li>\n<li>A review calendar beats a heroic one-off clean-up: weekly for rejections and alerts, monthly for exceptions, quarterly for permissions.<\/li>\n<\/ul>\n<p>Three months on, the manager gets the same question from another client. This time the developer hands her a one-page summary: named people, narrow roles, encrypted storage, logged access. The client, relieved, asks nothing further. If your own work involves coordinating trips and paperwork on tight timelines, our practical notes on <a href=\"http:\/\/desinri.com\/blog\/easy-ways-to-rent-and-ride-vehicles-in-japan\/\">getting around Japan by rented vehicle<\/a> show the same preference for preparation over improvisation.<\/p>\n<h2>The moral, without the jargon<\/h2>\n<p>None of what the team did required exotic products. It required asking an honest question, writing down the answer, and then closing the gaps in a sensible order. That, more than any tool, is what the discipline asks for.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An invented story of a small travel-services team that discovers its client portal&#8217;s cluster is wide open and fixes access, workloads, networks and secrets in order.<\/p>\n","protected":false},"author":8,"featured_media":926,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[],"class_list":["post-925","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-visa-immigration"],"_links":{"self":[{"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/posts\/925","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/comments?post=925"}],"version-history":[{"count":1,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/posts\/925\/revisions"}],"predecessor-version":[{"id":1027,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/posts\/925\/revisions\/1027"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/media\/926"}],"wp:attachment":[{"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/media?parent=925"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/categories?post=925"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/desinri.com\/blog\/wp-json\/wp\/v2\/tags?post=925"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}