<?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 Piyumi Madurangika on Medium]]></title>
        <description><![CDATA[Stories by Piyumi Madurangika on Medium]]></description>
        <link>https://medium.com/@piyumi.madurangika.au?source=rss-0bcab8e7f433------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*liUtSN5bBTrZQIRLIkGLCg.jpeg</url>
            <title>Stories by Piyumi Madurangika on Medium</title>
            <link>https://medium.com/@piyumi.madurangika.au?source=rss-0bcab8e7f433------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 14:21:26 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@piyumi.madurangika.au/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[5 Production Bugs That Changed How I Think About Software Testing]]></title>
            <link>https://medium.com/@piyumi.madurangika.au/5-production-bugs-that-changed-how-i-think-about-software-testing-5b0832a2594c?source=rss-0bcab8e7f433------2</link>
            <guid isPermaLink="false">https://medium.com/p/5b0832a2594c</guid>
            <category><![CDATA[quality-assurance]]></category>
            <category><![CDATA[qa-engineering]]></category>
            <category><![CDATA[quality-software]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[software-testing]]></category>
            <dc:creator><![CDATA[Piyumi Madurangika]]></dc:creator>
            <pubDate>Thu, 27 Aug 2026 12:13:41 GMT</pubDate>
            <atom:updated>2026-08-27T12:13:41.707Z</atom:updated>
            <content:encoded><![CDATA[<h4><strong>Real-world QA lessons from testing enterprise financial, web, API, and mobile applications</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/601/1*DtqnWZ65w9p3o7vEk693DQ.png" /></figure><p>One lesson became very clear to me early in my career as a Test Analyst:</p><p><strong><em>The most dangerous bugs are not always the biggest ones. Sometimes, they are the small issues that quietly reach production.</em></strong></p><p>During my years working in Quality Assurance and software testing, including enterprise financial technology environments, I had the opportunity to test systems where software quality directly affected transactions, operational processes, business users, and customers.</p><p>In these environments, a defect that looks minor from a technical perspective can have a much larger business impact.</p><p>A few cents of difference in a financial calculation.<br>One incorrect permission.<br>One environment configuration value.<br>One device-specific layout problem.<br>One delayed synchronisation process.</p><p>Each taught me something different about what <strong>quality engineering really means</strong>.</p><p>Here are five types of production defects that have had the greatest influence on how I approach software testing today.</p><h4><strong>1. Currency Calculation and Rounding Mismatches</strong></h4><p>Financial systems taught me very quickly that:<br><strong>“Almost correct” is still incorrect.<br></strong>Enterprise financial applications often need to process combinations of:</p><ul><li>Multiple currencies</li><li>Exchange rates</li><li>Taxes</li><li>Fees</li><li>Instalment amounts</li><li>Credit and debit values</li><li>Decimal precision</li><li>Settlement calculations</li></ul><p>A calculation could appear perfectly correct to a user while still producing a different value somewhere else in the system.</p><p>For example, problems can occur when:</p><ul><li>One component rounds a value to two decimal places while another keeps four</li><li>Exchange-rate calculations follow slightly different formulas</li><li>Backend calculations differ from values displayed by the frontend</li><li>Credit or debit amounts change slightly during conversion</li><li>Reports calculate totals differently from transaction-processing services</li></ul><p>Imagine this simplified scenario:</p><p><strong>Displayed amount:</strong> $1,250.48<br><strong>Backend calculated amount:</strong> $1,250.4756<br><strong>Settlement amount:</strong> $1,250.47</p><p>The difference looks tiny.</p><p>But when the same behaviour happens across thousands of transactions, it can contribute to:</p><ul><li>Reconciliation discrepancies</li><li>Accounting mismatches</li><li>Customer enquiries</li><li>Reporting inconsistencies</li><li>Audit concerns</li></ul><p><strong>What this taught me</strong></p><p>When testing financial applications, I learned to go beyond confirming that a calculation <em>looks correct</em>.</p><p>Now I think about:</p><ul><li>What decimal precision is required?</li><li>Where does rounding occur?</li><li>Does the frontend use the same calculation rules as the backend?</li><li>What happens with extremely small or large values?</li><li>What happens when currencies use different decimal conventions?</li><li>Do reports, APIs, databases and user interfaces all agree?</li></ul><p><strong>QA takeaway</strong></p><p>When testing financial calculations, validate:</p><ul><li>Decimal precision</li><li>Rounding rules</li><li>Currency conversion logic</li><li>Tax and fee calculations</li><li>Boundary values</li><li>Large and small transaction amounts</li><li>Multi-currency scenarios</li><li>Frontend versus backend calculations</li><li>Database values</li><li>Reporting and settlement values</li></ul><p>This experience reinforced one principle that has stayed with me:<br><strong>In financial software, accuracy is part of functionality.</strong></p><h4><strong>2. Role-Based Access and Authorisation Defects</strong></h4><p>Another type of production issue that can be deceptively serious involves <strong>user permissions</strong>.<br>Imagine a system containing several user roles:</p><p><strong>Administrator → Manager → Operations User → Read-Only User</strong></p><p>Each role should have different capabilities.<br>After a change or deployment, however, a user with limited permissions may unexpectedly gain access to functionality that should be restricted.</p><p>The application itself may still work perfectly.<br>Pages load.<br>Buttons work.<br>APIs respond.<br>Nothing crashes.<br>Yet from a security and business perspective, the system is behaving incorrectly.</p><p><strong>Why these bugs are dangerous</strong></p><p>A role-based access defect can potentially allow someone to:</p><ul><li>View sensitive information</li><li>Modify records</li><li>Approve transactions</li><li>Access administrative screens</li><li>Perform actions outside their responsibility</li></ul><p>It taught me an important distinction:<br><strong>Authentication answers:<br></strong><em>Who are you?<br></em><strong>Authorisation answers:<br></strong><em>What are you allowed to do?</em></p><p>Testing login functionality alone is therefore not enough.</p><p><strong>QA takeaway</strong></p><p>Regression testing should include a clear <strong>role-permission matrix covering access levels</strong> such as:</p><p><strong>Administrator:</strong> Full access <br><strong>Manager:</strong> Operational + approval access <br><strong>Standard User:</strong> Limited operational access <br><strong>Read-Only User:</strong> View only</p><p>Testing should also verify permissions beyond the UI level.</p><p>Hiding a button does <strong>not</strong> necessarily mean the underlying API is protected.</p><p>Whenever possible, I check permissions across:</p><p><strong>UI → API → Backend → Database/business rules</strong></p><p><strong>Question for testers</strong></p><p>If a restricted button disappears from the screen, would you consider the authorisation test complete?<br><em>I wouldn’t.</em></p><h4><strong>3. An API That Worked in QA but Failed in Production</strong></h4><p>This is one of those classic situations that almost every experienced tester eventually encounters.</p><p>Everything works in the QA environment.<br>Integration testing passes.<br>Regression testing passes.<br>The API responds correctly.<br>Then the application reaches production and something fails.</p><p>Why?</p><p>The code may be identical, but the <strong>environment is not</strong>.<br>Differences can exist in:</p><ul><li>Environment variables</li><li>API endpoints</li><li>Authentication credentials</li><li>Certificates</li><li>Network rules</li><li>Firewall configuration</li><li>Database configuration</li><li>Timeout values</li><li>Third-party integrations</li><li>Feature flags</li><li>Infrastructure settings</li></ul><p>That experience changed the way I think about the phrase:<br><strong>“It works in QA.”</strong></p><p>Passing QA is important, but it does not automatically prove that every production dependency is ready.</p><p><strong>QA takeaway</strong></p><p>Before a release, I now think beyond application functionality and ask:</p><ul><li>Are production configuration values validated?</li><li>Are endpoints correct?</li><li>Are certificates valid?</li><li>Are required services reachable?</li><li>Are secrets and credentials configured?</li><li>Are timeout and retry settings appropriate?</li><li>Are dependent services available?</li><li>Are feature flags configured correctly?</li><li>Is there monitoring if something fails?</li></ul><p>A strong release process requires collaboration between:</p><p><strong>QA + Development + DevOps + Infrastructure + Business</strong></p><p>It is also why production-like test environments are extremely valuable.</p><p>The closer a testing environment resembles production, the fewer surprises we are likely to discover after release.</p><h4><strong>4. A Mobile UI That Broke Only on Certain Devices</strong></h4><p>Mobile testing taught me another important lesson:<br><strong><em>A screen that looks perfect on one device proves very little about how it looks elsewhere.</em></strong></p><p>A mobile interface may work correctly on most devices while breaking under one combination of:</p><ul><li>Android version</li><li>Screen size</li><li>Resolution</li><li>Aspect ratio</li><li>Orientation</li><li>Device manufacturer</li><li>Font size</li><li>Operating-system behaviour</li></ul><p>The application may technically function, yet the experience can become unusable.<br>Buttons may overlap.<br>Text may disappear.<br>Fields may move outside the visible area.<br>Important controls may become impossible to select.</p><p><strong>How I approached this in practice</strong></p><p>In my previous testing experience, I used device-farm platforms such as Sauce Labs to test mobile applications across a broader range of devices and operating systems.</p><p>Instead of relying only on the physical devices available to the QA team, a device farm allowed me to validate the same application across different:</p><ul><li>Android versions</li><li>Device models</li><li>Screen sizes and resolutions</li><li>OS configurations</li><li>Browser and mobile environments</li></ul><p>It was particularly useful for reproducing issues that appeared only on specific devices.</p><p>For example, if a layout defect was reported on a particular Android version or screen size, I could use the device farm to reproduce the same environment, investigate the behaviour, validate the fix, and then run additional compatibility checks across other devices before release.</p><p>It helped make device compatibility testing more systematic and significantly reduced the risk of assuming that a feature was working correctly simply because it passed on one or two test devices.</p><p><strong>Why this matters</strong></p><p>Unlike many desktop environments, mobile applications run across a wide variety of hardware and operating system combinations.<br>That means mobile testing requires more than confirming whether the functionality works.</p><p>We also need to verify whether users can <strong>use the application correctly across the devices they depend on.</strong></p><p><strong>QA takeaway</strong></p><p>My mobile-testing checklist now considers:</p><ul><li>Different screen sizes</li><li>Multiple OS versions</li><li>Portrait and landscape orientation</li><li>Small and large displays</li><li>Different resolutions</li><li>Font scaling</li><li>Keyboard behaviour</li><li>Scrolling and navigation</li><li>Network interruptions</li><li>Background and foreground transitions</li><li>Real devices as well as emulators</li><li>Cloud-based device farms such as Sauce Labs</li></ul><p>When testing time is limited, I prefer a <strong>risk-based device coverage strategy</strong> rather than testing every possible combination.</p><p>I prioritise devices using factors such as:</p><p><strong>User base + OS usage + device popularity + business criticality + technical risk</strong></p><p>This experience taught me that device farms are not simply a convenience for mobile testers — they are an effective way to increase compatibility coverage, reproduce device-specific defects, and improve release confidence.</p><h4><strong>5. Data Synchronisation Delays Between Systems</strong></h4><p>Modern enterprise applications rarely operate alone.<br>A transaction may travel through several systems:</p><p><strong>Application → API → Backend Service → Database → Integration Service → External System</strong></p><p>Now imagine that one stage is delayed. <br>S<em>ystem A shows that an operation has completed.<br>System B still displays the old information.</em><br>A few minutes later, everything synchronises.<br>Technically, nothing may have been permanently lost.<br>But to the user, the system appears inconsistent.<br>And inconsistent systems quickly reduce confidence.</p><p><strong>Why synchronisation bugs are difficult</strong></p><p>These problems often involve:</p><ul><li>Asynchronous processing</li><li>Background jobs</li><li>Message queues</li><li>Scheduled tasks</li><li>APIs</li><li>Caching</li><li>Database replication</li><li>Third-party systems</li></ul><p>They may also be timing-dependent, making them difficult to reproduce consistently.</p><p><strong>QA takeaway</strong></p><p>Testing distributed systems requires more than testing individual components.<br>We also need to validate the <strong>complete business journey</strong>.</p><p>For example:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ml9IX2y--zK6yeO9NvVWCQ.png" /></figure><p>Important questions include:</p><ul><li>How long should synchronisation take?</li><li>What happens if one service is unavailable?</li><li>Is retry logic implemented?</li><li>Can duplicate records be created?</li><li>Does the user receive a meaningful status?</li><li>Is eventual consistency acceptable?</li><li>Can failed transactions be identified?</li><li>Are failures logged and monitored?</li></ul><p>This experience strengthened my appreciation for<em> </em><strong><em>end-to-end integration testing, observability and production monitoring</em></strong><em>.</em></p><h4><strong>What These Production Bugs Changed About My Approach to QA</strong></h4><p>Earlier in my career, I naturally thought about testing mainly as a sequence:<br><strong>Requirement → Test Case → Execution → Defect → Retest<br></strong>Real production experience changed that thinking.</p><p>Today, when I look at a feature, I don’t only ask:<br><strong>“Does it work?”</strong></p><p>I also ask:</p><ul><li>What could fail?</li><li>Who could be affected?</li><li>What is the business impact?</li><li>What happens when systems interact?</li><li>What happens with unexpected data?</li><li>What changes between environments?</li><li>What security, permission or compatibility risks exist?</li><li>How would we detect the problem if it reached production?</li></ul><p>My testing mindset has therefore evolved into something broader:</p><p><strong>Requirement → Business Risk → User Behaviour → Integration → Data → Security → Environment → Edge Cases → Monitoring → Production Impact</strong></p><p>For me, this is where traditional software testing becomes <strong>quality engineering</strong>.</p><h4><strong>The Value of Experience in Modern QA</strong></h4><p>One of the most important lessons from my experience is that QA should never be treated as the final checkpoint before release.</p><p>The real value of QA starts much earlier.</p><p>A strong QA professional can help teams clarify requirements, identify business and technical risks, design meaningful test coverage, validate integrations and data flows, challenge assumptions, and improve release confidence before problems reach customers.</p><p>Finding a defect during testing is valuable.<br>Finding a critical defect before a customer experiences it is even more valuable.<br>But the highest level of quality thinking is often:</p><p><strong><em>Understanding the risk early enough that the defect never needs to exist in the first place.</em></strong></p><p>Software systems have changed significantly during my career. Applications increasingly depend on APIs, cloud services, mobile devices, distributed systems, third-party integrations, databases, automation and security controls.<br>Yet the lessons behind these five production defects remain highly relevant.</p><p>A small calculation discrepancy can become a financial reconciliation issue.<br>A permission defect can become a security risk.<br>A configuration difference can cause a production API failure.<br>A device-specific UI issue can affect an entire group of users.<br>A delayed integration can make two functioning systems display conflicting information.</p><p>These are different defects, but they share one important lesson:</p><p><strong><em>Quality problems rarely respect system boundaries.</em></strong></p><p>That is why experienced QA professionals need to understand more than whether a test case passes or fails. We need to understand how the <strong>product, users, data, integrations, environments and business processes connect</strong>.</p><p>Experience has taught me to think in terms of:<br><strong>Risk. Impact. Data. Integration. User behaviour. Production reality.<br></strong>It has also taught me that quality is a shared responsibility.<br>The strongest outcomes occur when <strong>QA works closely with developers, business analysts, product teams, DevOps and business stakeholders</strong> and asks difficult questions early.</p><p>As I continue building the next stage of my QA and Test Engineering career in Australia, this is the mindset I carry forward.</p><p>Tools will change.<br>Automation frameworks will evolve.<br>AI will continue reshaping software development and testing.<br>Systems will become more connected and complex.<br>But organisations will still need people who can look at a system and ask:</p><p><strong>“Where is the real risk?”</strong></p><p>They will still need QA professionals who understand the difference between something that technically works and something that is genuinely <strong>reliable, secure, accurate and ready for customers</strong>.</p><p>For me, that is the real value of experience in QA.<br>It is not simply knowing more testing techniques.</p><p>It is developing better judgement about <strong>where quality can fail, what the impact could be, and where testing can create the greatest business value.</strong></p><h4><strong>One Final Question</strong></h4><p>If you work in QA, software development, product or engineering:</p><p><strong>What production defect changed the way you think about software quality?<br></strong>Was it related to data, security, APIs, performance, mobile compatibility, configuration, integration or something completely unexpected?</p><p>I’d be interested to hear not only what the defect was, but also <strong>what it changed about the way you approach quality today.</strong></p><p>Because some of the most valuable lessons in software engineering come from the problems we never want to see twice.</p><p><strong><em>“A production bug may begin as a defect.<br>Handled well, it becomes experience.<br>Applied properly, that experience becomes better engineering.”</em></strong></p><p><em>Thanks for reading my first article on Medium. I’ll be sharing more practical insights from software testing, Quality Assurance, API testing, mobile testing, automation, quality engineering and real-world software delivery.</em></p><p><strong>If these topics are part of your world too, I’d be glad to have you follow along.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5b0832a2594c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>