<?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 Xyzzklk on Medium]]></title>
        <description><![CDATA[Stories by Xyzzklk on Medium]]></description>
        <link>https://medium.com/@xyzzklk?source=rss-42770bc6d4d7------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*2zt_ii1X3C-QPB9g2jE37A.jpeg</url>
            <title>Stories by Xyzzklk on Medium</title>
            <link>https://medium.com/@xyzzklk?source=rss-42770bc6d4d7------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 12:03:55 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@xyzzklk/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[Destructively Bypassing Huorong Internet Security 5.0 to Create Persistence]]></title>
            <link>https://medium.com/@xyzzklk/destructively-bypassing-huorong-internet-security-5-0-to-create-persistence-987e0012ea08?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/987e0012ea08</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[endpoint-security]]></category>
            <category><![CDATA[penetration-testing]]></category>
            <category><![CDATA[reverse-engineering]]></category>
            <category><![CDATA[malware-analysis]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Sat, 21 Nov 2026 20:31:01 GMT</pubDate>
            <atom:updated>2026-11-21T20:31:01.433Z</atom:updated>
            <content:encoded><![CDATA[<h4>A Defence Evasion Technique via Hijacking Windows System Binaries by Abusing a Privilege Modification Blind Spot in Huorong’s Defence Rules</h4><blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/126999464">originally documented in 22 Sep 2022</a>, when I was in 10th grade, and now republished on Medium. So, if you realise that the whole article is about the legacy Huorong Internet Security 5.0 version (Individual Edition) which official support has long ended, please don’t mind.</blockquote><blockquote>The vulnerability described in this article is rather simple, rough and not that kind of “god-level” one and has specific limitations. This evasion technique was tested in both Windows 7 Ultimate and 10 Pro environments, against Huorong and additionally, the latest version of Qihoo 360 at time of writing.</blockquote><blockquote>I had <a href="https://bbs.huorong.cn/forum.php?mod=viewthread&amp;tid=110725&amp;extra=page=1&amp;mobile=no">reported this flaw to Huorong officially and it was then confirmed and patched in version 5.0.73.12</a>. Viewing the thread outside of China might result in being blocked by their sensitive WAF, though.</blockquote><blockquote>⚠️<strong> Disclaimer</strong></blockquote><blockquote>This article is published strictly for educational and defensive research purposes. Understanding how persistence mechanisms and detection blind spots operate helps defenders, blue teams and security developers build stronger monitoring and protection rules. All testing was conducted in an isolated lab environment. Readers are expected to use this information responsibly and only in authorised testing environments.</blockquote><h3>The Context</h3><p>Consider the following scenario—a rogue attacker wants to deploy some malware, either some ransomware or prankware that locks or destroys the victim’s machine, and being persistent across reboots so the malware never comes off unless the victim pays, or just stays there forever (e.g. for revenge).</p><p>In this case, since the goal is to lock the whole OS, there is no worry about being stealthy.</p><p>To make a malicious executable always start upon reboot, the attacker must utilise some persistence mechanism.</p><p>Common Windows persistence mechanisms that immediately come up in our heads are all well-defended by the Host Intrusion Prevention Systems (HIPS) from different vendors:</p><ul><li>Creating a System or User-level Service</li><li>Registry</li><li>Scheduled Tasks</li></ul><p>Directly exploiting any one of these will immediately trigger a huge alert and probably result in the initiating binary being quarantined immediately.</p><p>So, here’s a cleverer approach.</p><h3>Still From Registry, but Without Value Modification</h3><p>An attacker could swap a system executable, for example, explorer.exe, that always starts on boot or login with their own binary so that persistence is achieved. Why swap? Won’t this cause the legitimate explorer.exe to not be able to start? It will, but since the entire goal is to create panic, making the whole screen black (true explorer not starting anymore) and losing all control widgets like the taskbar only achieves better threatening effect on an average user.</p><p>explorer.exe itself also starts from registry, but the binary swapping approach is way “safer” than pointing the registry record that holds the location of explorer.exe to a different binary, as any registry move or even starting the reg.exe command-line binary could trigger a serious alert, for most HIPS.</p><p>To understand what I mean by “pointing the registry record of explorer.exe to a different location”, we need to look at how explorer.exe itself is started from registry. By default, when a user logs on, explorer.exe is started through a process like this:</p><ol><li>winlogon.exe authenticates user credentials.</li><li>It queries the registry key HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon for the Userinit value, which is C:\WINDOWS\system32\userinit.exe, by default. It calls Windows API CreateProcessAsUser to launch userinit.exe under the logon user’s name.</li><li>userinit.exe reads the Shell value from the same HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon key, which is explorer.exe by default.</li><li>userinit.exe spawns explorer.exe and immediately terminates itself, so explorer.exe looks orphaned without a parent process.</li></ol><p>So, it simply means that directly editing something like Shell is too risky.</p><p>But will swapping explorer.exe be unnoticed?</p><p><strong>That was where the flaw was.</strong></p><h3>Detailed Exploit</h3><p>We all know that to prevent system files from being arbitrarily modified, Windows sets their owner to NT SERVICE\TrustedInstaller by default and only gives Administrators the Read and Read &amp; execute NTFS permissions. So a process running with Administrator privileges cannot rename, move or replace system files.</p><p>Let us look at one system file, C:\Windows\System32\cmd.exe as an example.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/544/1*km-t_YT6dvbDhIv4goIrIg.png" /><figcaption>NT SERVICE\TrustedInstaller permissions (machine translated from ZH-CN Windows UI)</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/544/1*C2MS9sRZndrfel7fS6tdew.png" /><figcaption>Administrators permissions (machine translated from ZH-CN Windows UI)</figcaption></figure><p>To bypass this and swap system files, an attacker can simply change permissions the built-in Administrators group has to give Administrators Full Control over the target file, and this only requires changing the owner to Administrators beforehand.</p><p>The only requirement is to do all this as an Administrator. An attacker can simply ship a malicious executable that requires UAC to run. Even though this technically creates a small barrier, from my own observation and experience, most average humans just blindly allow it, from a psychological or social engineering view. Plus, some legitimate app installers require leveraged UAC to run as well, further weakening the phsychological effect of that warning. Even worse, on Windows 7 machines, many users are using the built-in default Administrator account, which prompts nothing when running an executable that requires Administrators privileges.</p><p>Unfortunately, Huorong had zero defence rules for preventing ownership taking and arbitrary permission modification of system files, <strong>even with its optional </strong>System Directory<strong> file protection rule enabled.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6VCSlKJrFHDl9uPPcSQ92g.png" /><figcaption>Huorong System Hardening Rules UI (machine translated from ZH-CN)</figcaption></figure><blockquote>In fact, the same applies to Qihoo 360, which only prompts a yellow low-level warning with no immediate action taken automatically, for the latest version at time of experimenting.</blockquote><p>So an automated exploit targeting explorer.exe written in Batch Script could be like:</p><pre>@echo off<br>takeown /A /F C:\Windows\explorer.exe<br>icacls C:\Windows\explorer.exe /grant Administrators:F<br>echo Privilege stealing complete! Ha, ha, ha!<br>pause &gt;nul<br>exit</pre><p>And now we can do whatever we want. For example, replacing it with something evil.</p><pre>REM Screen goes black!<br>taskkill /f /im explorer.exe<br><br>rename C:\Windows\explorer.exe C:\Windows\explorerOriginal.exe<br>move PATH_TO_MALICIOUS_BIN\evil.exe C:\Windows\explorer.exe<br><br>REM Optional: start the impostor now<br>start explorer.exe</pre><p>Even if we don’t run the new, evil C:\Windows\explorer.exe now, it will be automatically triggered on the next reboot.</p><p>The most interesting thing is that an attacker could technically leave a secret recovery script that, when triggered, automatically fixes everything to the original state by swapping the evil impostor version of explorer.exe back to the original one, say explorerOriginal.exe in this example.</p><pre>taskkill /f /im explorer.exe<br>rename C:\Windows\explorer.exe C:\Windows\explorerEvil.exe<br>rename C:\Windows\explorerOriginal.exe C:\Windows\explorer.exe<br><br>REM Resets to default privileges, so inherits from C:\Windows again<br>icacls C:\Windows\explorer.exe /reset<br>icacls C:\Windows\explorer.exe /setowner &quot;NT SERVICE\TrustedInstaller&quot;<br><br>start explorer.exe</pre><p>This is obviously a blind spot in Huorong’s rules.</p><h3>The Fix</h3><h4>Solution 1: Add a Custom Defence Rule</h4><p>If you’re using a version before the official fix rolled out, use the following custom defence rule to alert when permission changing commands are run from command line, as supported by Huorong for flexible tuning.</p><pre>{<br>    &quot;ver&quot;:&quot;5.0&quot;,<br>    &quot;tag&quot;:&quot;hipsuser&quot;,<br>    &quot;data&quot;:[<br>        {<br>            &quot;id&quot;:1,<br>            &quot;power&quot;:1,<br>            &quot;name&quot;:&quot;Changing File Permissions from Command Line&quot;,<br>            &quot;procname&quot;:&quot;*&quot;,<br>            &quot;treatment&quot;:1,<br>            &quot;policies&quot;:[<br>                {<br>                    &quot;montype&quot;:0,<br>                    &quot;action_type&quot;:16,<br>                    &quot;res_path&quot;:&quot;C:\\Windows\\System32\\takeown.exe&quot;<br>                },<br>                {<br>                    &quot;montype&quot;:0,<br>                    &quot;action_type&quot;:16,<br>                    &quot;res_path&quot;:&quot;C:\\Windows\\system32\\icacls.exe&quot;<br>                },<br>                {<br>                    &quot;montype&quot;:0,<br>                    &quot;action_type&quot;:16,<br>                    &quot;res_path&quot;:&quot;C:\\Windows\\SysWOW64\\icacls.exe&quot;<br>                },<br>                {<br>                    &quot;montype&quot;:0,<br>                    &quot;action_type&quot;:16,<br>                    &quot;res_path&quot;:&quot;C:\\Windows\\SysWOW64\\takeown.exe&quot;<br>                }<br>            ]<br>        }<br>    ]<br>}</pre><h4>Solution 2: Updating to the Latest Version</h4><p>This flaw was officially assigned a Bug ID 38826 and patched in version 5.0.73.12.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EgvUJaxDe9wJanAoZ_2FFg.jpeg" /><figcaption>Official thread (machine translated from ZH-CN)</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=987e0012ea08" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[FreePBX 17: SIP Trunk Going Offline Without Configuration Change? Here’s Why]]></title>
            <link>https://medium.com/@xyzzklk/freepbx-17-sip-trunk-going-offline-without-configuration-change-heres-why-ce836f30262b?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/ce836f30262b</guid>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[voip-telephone-system]]></category>
            <category><![CDATA[voip]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Fri, 20 Nov 2026 04:31:04 GMT</pubDate>
            <atom:updated>2026-11-20T04:31:04.558Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/148404706">originally documented in 2025</a> and now republished on Medium. If you find some details being legacy or not matching the latest version of FreePBX, please don’t mind.</blockquote><h3>Problem Background</h3><p>Sometimes, we want to establish SIP trunking between two FreePBX servers so that an extension on one can call an extension on the other. For example, two sites of an organisation want to make calls between them, and each site has a server.</p><h3>Symptoms of the Problem</h3><p>Sometimes, because of a network disruption, a SIP trunk may disconnect.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/721/0*2mUbfNJoI56rkfIY.png" /></figure><p>However, the dumb thing is that Asterisk (the core engine used by FreePBX underneath) cannot seem to auto-reconnect by itself.</p><p>Because the settings haven’t been changed, we know it is not a configuration problem.</p><h3>Reasoning</h3><p>Usually, Asterisk <em>does</em> try to auto-reconnect. But if the attempt keeps failing, for example, when a server is powered on too long after its remote peer has, it might eventually give up. In fact, we can see some logs like this:</p><pre>[2025-06-03 17:54:16] WARNING[1688]: res_pjsip_outbound_registration.c:1249 schedule_retry: No response received from &#39;sip:192.168.51.3:5060&#39; on registration attempt to &#39;sip:To_Site_B@192.168.51.3:5060&#39;, retrying in &#39;60&#39;<br>[2025-06-03 17:54:20] WARNING[1688]: res_pjsip_registrar.c:1239 find_registrar_aor: AOR &#39;&#39; not found for endpoint &#39;To_Site_B&#39; (192.168.51.3:5060)</pre><p>So this is not a big problem as long as we manage to let them reregister to their peers.</p><p>Note that before attempting anything, make sure they are still reachable at network level as PJSIP requires connectivity to the remote peer’s IP address and port, obviously.</p><h3>The Fix</h3><p>Let us try reloading FreePBX on each server.</p><pre>fwconsole reload</pre><p>No effect.</p><p>It could be just restarting FreePBX itself but not many of the underlying services at Asterisk level, so let’s try:</p><pre>fwconsole restart</pre><p>to restart both Asterisk and FreePBX. This forces the SIP trunk to reregister.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/684/0*kaXhyFlbVLHFxufd.png" /></figure><p>After waiting for a long time, we see that they’ve finished restarting. After waiting for another while, we see that the trunk comes back online.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/228/0*cAR--mHT-ynOzxsU.png" /></figure><p>Problem solved.</p><h3>Insights</h3><p>It is likely to observe two FreePBX peers unable to register the trunk if they cannot connect for a while after booting.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ce836f30262b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When the Very First Attempt on FreePBX 17 Fails: Ajax Request Broken or Aborted for an Unexpected…]]></title>
            <link>https://medium.com/@xyzzklk/when-the-very-first-attempt-on-freepbx-17-fails-ajax-request-broken-or-aborted-for-an-unexpected-50d66261254b?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/50d66261254b</guid>
            <category><![CDATA[asterisk]]></category>
            <category><![CDATA[freepbx]]></category>
            <category><![CDATA[voip]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Wed, 18 Nov 2026 04:54:43 GMT</pubDate>
            <atom:updated>2026-11-18T12:20:40.450Z</atom:updated>
            <content:encoded><![CDATA[<h3>When the Very First Attempt on FreePBX 17 Fails: Ajax Request Broken or Aborted for an Unexpected Reason</h3><blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/148403798">originally documented in 2025</a> and now republished on Medium. If you find it mentioning some details that no longer match the latest versions of FreePBX or that FreePBX has already fixed the issue described in this article, please don’t mind.</blockquote><h3>Background of the Problem</h3><p>After installing the operating system of a fresh FreePBX 17 server, we will be required to perform an initial configuration after logging in to its web interface. This includes setting up the admin account and password, enabling the responsive firewall and activating any licenses etc.</p><h3>The Problem</h3><p>During the initial configuration steps, we will be asked about which networks and IP addresses to trust. For example, whether to trust the local LAN the server is in (in my case, it is 192.168.51.0/24) and the current IP address I am accessing FreePBX from.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/999/0*fa8OejVxpun8XYci.png" /></figure><p>Whether to trust the entire local LAN depends on the situation, but for the current IP, we almost always choose to trust it because otherwise, you will not be able to continue with the rest of the setup.</p><p>So, we choose Yes.</p><p>However, after waiting for a while, the webpage just stayed still, and finally prompted a nasty error message.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*C6LkwDpvvAkTReGu.png" /><figcaption>Ajax request broken or aborted for an unexpected reason. Please check console logs for more details.<br>The network you are currently using (192.168.51.0/24) to manage this server isn’t marked as Trusted.<br>If this is a known secure network, you should add it to the Trusted zone<br>Please ensure that you are not inadvertently allowing unauthorized hosts access to your machine. You should only select ‘Yes’ if you are sure the network (above) is not accessible by any unknown third parties.</figcaption></figure><p>After that, no matter what on the webpage we click, the browser will simply prompt “The page is not responding”.</p><h3>Analysis</h3><p>If the remote port is still reachable but not listened, the browser’s error message should be something like “Connection Refused”, or that the remote port should return some data our browser cannot parse.</p><p>But reality is that our browser request kept “spinning” for a long time and never heard a response. This strongly suggests some firewall blocking (“filtered”). In the worst case, FreePBX still blocked us despite that we chose to whitelist our IP. Let’s verify this.</p><h3>Solving the Problem</h3><p>Let’s explicitly try to add our IP to the firewall’s trust list and see if the behaviour changes.</p><pre>fwconsole firewall trust 192.168.51.21/32</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*6U5F3VzYtEvtwQno.png" /></figure><p>Refresh the web console. We see ourselves accessible again.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Xdowgi2kJeXLU3Ph.png" /></figure><p>Problem solved.</p><h3>Conclusion</h3><p>FreePBX should have trusted our IP and allowed us to access, yet it did the complete opposite, even though we clicked Yes.</p><p>Sangoma’s products, such as FreePBX, are not stable. In many cases, it is not the admin but the software’s own fault. So, next time when you see another error, blame yourself less. It could be the thing’s own problem.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=50d66261254b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Solving Hyper-V Debian Guest VM Boot Error: The singed image’s hash is not allowed]]></title>
            <link>https://medium.com/@xyzzklk/solving-hyper-v-debian-guest-vm-boot-error-the-singed-images-hash-is-not-allowed-2e21012be1c3?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/2e21012be1c3</guid>
            <category><![CDATA[hyper-v]]></category>
            <category><![CDATA[microsoft]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[virtualization]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Wed, 11 Nov 2026 04:31:01 GMT</pubDate>
            <atom:updated>2026-11-18T12:19:58.505Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/147747338">originally documented in 2025</a> and now republished on Medium.</blockquote><h3>The Problem</h3><p>Sometimes, we want to run Linux virtual machines in Hyper-V, like Debian. We want to use a Debian ISO image to install a fresh operating system.</p><p>However, after we have created a Generation 2 VM in Hyper-V and assigned the ISO image to have the highest boot priority, booting the VM results in the following error.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/684/0*ePFdeQCnRcqpTdA2.png" /><figcaption>Microsoft Hyper-V UEFI<br>Virtual Machine Boot Summary<br>1. SCSI DVD (0,1) The signed image’s hash is not allowed (DB)</figcaption></figure><h3>Reasoning</h3><p>This is because Gen 2 VMs in Hyper-V use secure boot by default. Secure boot only allows the VM to boot from a certain “whitelist” of ISO images to make sure it does not boot from a dangerous one, and Hyper-V checks if your image is in the list by calculating its hash.</p><p>Unlike other hypervisors and firmware, this list in Hyper-V is particularly short.</p><h3>The Solution</h3><p>Disable secure boot, and power on normally.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/674/0*1Q-Pt9eHLIxbiqm1.png" /><figcaption>Disabling secure boot in Hyper-V settings</figcaption></figure><p>Now the VM boots without errors.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2e21012be1c3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Installing Kali Linux in GNS3 for Low-Level Network Attack Simulation—The Reliable Setup Guide]]></title>
            <link>https://medium.com/@xyzzklk/installing-kali-linux-in-gns3-for-low-level-network-attack-simulation-the-reliable-setup-guide-bef23a8d86db?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/bef23a8d86db</guid>
            <category><![CDATA[cisco]]></category>
            <category><![CDATA[kali-linux]]></category>
            <category><![CDATA[cisco-packet-tracer]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Tue, 10 Nov 2026 05:22:58 GMT</pubDate>
            <atom:updated>2026-11-18T12:19:10.541Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/147477647">originally documented in 2025</a> and now republished on Medium. If you find it mentioning some details that no longer match the latest versions of GNS3 or if that GNS3 has already fixed some of the issues described in this article, please don’t mind.</blockquote><h3>GNS3’s Problematic Default Setup</h3><p>It is often required to install Kali Linux in GNS3, an emulator software for learning computer networking without real devices, to simulate low-level network attacks against networking infrastructure, such as Cisco routers and switches.</p><p>In this case, we will need to add a real Kali Linux virtual machine in our current GNS3 project to simulate an attacker’s device.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/178/0*w1LSxuzyfzme1q6K.png" /></figure><p>GNS3 does provide such capability. The default, official way is to add a QEMU (uses KVM, so it fits GNS3 VM or Server) Kali Linux VM that is <em>live booted</em> from a CD image. This does, technically, provide great simplicity by bypassing the complex OS installation process every time, but it creates a big issue—there are too many limitations under live boot, especially with installing certain tools.</p><ul><li>apt produces weird behaviours and could not properly install packages under live boot.</li><li>This directly affects our ability to use offensive tools that require separate installation, like <a href="https://github.com/tomac/yersinia">Yersinia</a>, making attack simulation impossible.</li></ul><p>Also, the current latest version of GNS3 (2.2.54) still only supports Kali Linux 2021.1, which causes weird errors when forcefully using with the current latest Kali CD image 2025.1a.</p><blockquote>Again, this article was not written at the same time it was republished here, so it references the latest versions when it was written.</blockquote><p>The workaround is to create a machine that boots from <em>hard disk</em> with the OS properly installed. GNS3 is able to store the contents of the hard disk file to create a “template” of a certain device, so we don’t need to install a brand-new OS every time we need a Kali machine.</p><h3>Preparing for Install</h3><p>First, make sure you have the latest version of GNS3, which can be downloaded from <a href="https://www.gns3.com/software/download">its official website</a>.</p><p>If you have an older version and want to upgrade it to the latest, you can reference <a href="https://medium.com/@xyzzklk/upgrading-gns3-from-client-to-vm-server-the-complete-walkthrough-9646a1c7c80e">this article</a>.</p><p>You may use a GNS3 VM with nested virtualisation enabled or GNS3 Server along with your GNS3 client. In my case, I’m using Hyper-V to run a GNS3 VM. If you ever come across issues with Hyper-V checkpoints, you can reference <a href="https://medium.com/@xyzzklk/the-nightmare-with-hyper-v-checkpoints-unable-to-edit-virtual-hard-disk-c3985733c20c">this article</a>.</p><h3>Installing The Kali Template in GNS3</h3><h4>Installing a New Template</h4><p>Start the GNS3 client. I’ll need to make sure that my VM starts with the client as well. The client needs Administrator privileges to automatically turn on a VM in Hyper-V (related details can be found <a href="https://medium.com/@xyzzklk/upgrading-gns3-from-client-to-vm-server-the-complete-walkthrough-9646a1c7c80e">here</a>).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UfLsTu25yLi4k7qF.png" /></figure><p>Now we choose to add a template.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*vIFWbaVpNWuMAFTf.png" /></figure><p>We choose to Install an appliance from the GNS3 server (recommended). This simply means to use a pre-defined device template provided by GNS3 officially, which information is stored on the VM and comes with the VM’s initial installation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QHuA2IKxt21e0sJ7.png" /><figcaption>New template</figcaption></figure><p>We choose a Kali machine that runs as a VM (QEMU), not as a Docker container.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FCU0DRGHS0MPe8Cu.png" /></figure><p>Then we install this template to our GNS3 VM. Our client machine is Windows 11, which doesn’t really provide any option for running KVM, so the VM is our only choice.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*fyqSAppqEC6uIWPB.png" /></figure><p>GNS3 confirms that it will use the AMD64 engine of QEMU to run this VM. We agree.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ZFoKdrl5GDAVrv8R.png" /></figure><p>We notice that by default, support only goes up to 2021.1. Let’s create a new template version just to prevent confusion.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jA9K7UDvYhc871W1.png" /></figure><p>In our case, we name it 2025.1a. You can name it according to the version of the Kali image you use.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/782/0*Rn6Bo_rvRzivK02E.png" /></figure><p>Even though we won’t be using live boot, GNS3 still forces us to specify a file name for the CD image that might be used by this template, if there is any. We will follow the official naming convention like its pre-installed other Kali versions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/416/0*YEFePXO-lpeAgYdV.png" /></figure><p>We do the same for the hard disk file.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/442/0*xzbJ100IklnL28G5.png" /></figure><p>Now, the new version will be shown at the bottom of the window.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*08viOj927s1Xm8Xf.png" /></figure><p>For official templates, checksums for image and disk files are enabled to prevent a user from accidentally uploading a wrong image file that doesn’t match the device model they intend to add. For our custom version, we can ignore this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*JA_MYDCIpwtPwo0C.png" /></figure><p>GNS3 indicates that we still need to upload image files to create a machine of such template.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*J_Obe9vgueyKlKTr.png" /></figure><p>So, we go to <a href="https://old.kali.org/kali-images/">Kali’s official website</a> to download files for 2025.1a.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*X8FKfp9NaZsSP_2S.png" /></figure><p>Find the AMD64 CD image and the pre-installed VM disk and download them. We won’t really be using the CD here as we’re booting directly from the virtual disk but just using it as a place filler and backup.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*L6HetegdvjOBhwVR.png" /></figure><p>After the download has finished, extract the 7-zip file that contains the VMWare virtual disk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/0*GYwOXy5unKHH85mU.png" /></figure><p>Then upload the files to GNS3. Repeat the same steps for both files. It might take a long time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1df4aqv9CTs0HP7K.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/472/0*JwCyLxxZZa7YIteK.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7a_28oSPuM57Fvco.png" /></figure><p>Now GNS3 shows that all required files are present. We can go to the next step and install it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ozHw3i-CS6iBxgg3.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/726/0*3yWGfKN39jQbH0_g.png" /></figure><p>The template creation wizard is about to finish. Even though it prompts that the default username and password are root and toor for Kali, the actual defaults for the latest version are kali and kali.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*moJEgGsR3GHP2Hsr.png" /></figure><p>Click Finish to finish the wizard. We see that our new device has been successfully added.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/964/0*zTEzw-rNmR8HCAwO.png" /></figure><p>It also shows up in the device list.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/526/0*zTsNcCem_9lXiAWU.png" /></figure><h4>Configuring the Kali Template in GNS3</h4><p>Yet we still cannot create an instance from this machine template and start it at this stage because GNS3 will try to start from the CD image file using live boot by default, which defeats the purpose of using a hard disk. We need to change its configuration to inform it to always start from hard disk.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/691/0*tiLg__sjFyLvyPUK.png" /></figure><p>We change Boot Priority to HDD to let GNS3 know to always start an instance of this device from hard disk. We also change memory RAM to 2048 MB to make sure there are enough resources to make Linux properly run. 1GB is enough for older Kali versions, but not the latest.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nq1gu8oM8WYiL5U9.png" /></figure><h4>Test Run</h4><p>We’ve finally done all the necessary configurations. Drag a Kali device into our working area and power the machine on!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1_OsZQ___F1vUmrB.png" /></figure><p>Click Terminal. We will be redirected automatically to a graphical VNC connection to this VM.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iHr3AEQoCNq8_qst.png" /></figure><p>All done!</p><h3>Managing Kali Linux in GNS3</h3><h4>Connecting Kali to The Internet</h4><p>But this is not the end. If you try to start an attack simulation by installing tools like Yersinia right now, you will notice that Internet is not available, which is something necessary to download packages from Kali sources.</p><p>Thankfully, unlike Cisco Packet Tracer, we have the ability to connect GNS3 devices to the real internet through the GNS3 VM (or GNS3 Server). This is achieved by connecting devices to a special device called “Cloud”, which is also the symbol we use in network topologies to represent the internet or a wide area network.</p><p>There are 2 types of “Clouds” in GNS3. The normal “Cloud” and “NAT Cloud”. The only difference is that the latter uses NAT (Network Address Translation), which makes your emulated devices internet accessible but not able to accept traffic proactively initiated from other devices in the same LAN your GNS3 VM is in because of PAT (Port Address Translation).</p><p>However, there are some other things to watch out:</p><ul><li>On my PC, for some unknown reason, there is always only one “Cloud” usable, and if one works, the other always doesn’t, with some weird error.</li><li>Both “Clouds” also act as DHCP servers, so a device connected to it <em>can</em> use DHCP to get its IP configuration automatically.</li><li>But there are exceptions to the previous one. On my device, sometimes “NAT Cloud” hands out DHCP leases timely, other times it just never responds, and I will need to manually configure IP addresses for devices connected to it. Just remember that in that case, the network the “NAT Cloud” is in is 192.168.122.0/24, and the “NAT Cloud” is 192.168.122.1.</li></ul><p>So, let’s try connecting Kali to one of the clouds to gain internet access, starting with connecting Kali to the normal “Cloud”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*vFTwKCLQbAMWf8Ue.png" /></figure><p>In this trial, Kali got a lease automatically via DHCP. Normal “Cloud” doesn’t perform NAT, so Kali got this IP address through DHCP from the Hyper-V NAT switch my GNS3 VM is in, hence the 172.17-31.X.X/20 IP address.</p><p>When using “NAT Cloud”, if the cloud doesn’t offer any leases, configure a static IP and gateway info on Kali to gain proper internet access, which can be done in the following interface by changing Method.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*s_uKs8cNO7TLQ3-G.png" /></figure><p>You can also leverage a Cisco router. With this way, you can add more end devices to internet connection through a single access point. If you enable DHCP on the router, you get extra bonus of not needing to configure static IP info on every end device because the router will be handing out leases automatically.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/552/0*sw4KWA6JDscuCcCj.png" /></figure><p>The only thing to watch out is that free Cisco images for education and experiment purposes cap total network throughput at 1Mbps, so expect slow internet access for all LAN devices.</p><p>To make sure the router itself uses the normal “Cloud” to access internet, perform the following configurations. In my case, it automatically gets an IP address from Hyper-V’s NAT switch.</p><pre>R1#config terminal<br>Enter configuration commands, one per line.  End with CNTL/Z.<br>R1(config)#interface g0/1<br>R1(config-if)#ip address dhcp<br>R1(config-if)#no shutdown<br>R1(config-if)#<br>*Apr 25 02:05:15.166: %LINK-3-UPDOWN: Interface GigabitEthernet0/1, changed state to up<br>*Apr 25 02:05:16.166: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/1, changed state to up<br>R1(config-if)#<br>*Apr 25 02:05:23.981: %DHCP-6-ADDRESS_ASSIGN: Interface GigabitEthernet0/1 assigned DHCP address 172.20.23.39, mask 255.255.240.0, hostname R1<br>R1(config-if)#do show ip interface brief<br>Interface                  IP-Address      OK? Method Status                Protocol<br>GigabitEthernet0/0         unassigned      YES unset  administratively down down<br>GigabitEthernet0/1         172.20.23.39    YES DHCP   up                    up  <br>GigabitEthernet0/2         unassigned      YES unset  administratively down down<br>GigabitEthernet0/3         unassigned      YES unset  administratively down down<br>R1(config-if)#do ping 8.8.8.8<br>Type escape sequence to abort.<br>Sending 5, 100-byte ICMP Echos to 8.8.8.8, timeout is 2 seconds:<br>!!!!!<br>Success rate is 100 percent (5/5), round-trip min/avg/max = 16/17/19 ms<br>R1(config-if)#</pre><p>Now we have to configure NAT on the router itself. Otherwise, the Hyper-V NAT switch doesn’t know any routes to networks on the other side of the router, resulting in traffic behind the router only being able to send out but any responses not being able to travel back using the same route. In our case, the Kali LAN behind the router is 192.168.1.0/24, and the router’s interface in that LAN is 192.168.1.1.</p><pre>R1#config terminal<br>Enter configuration commands, one per line.  End with CNTL/Z.<br>R1(config)#interface g0/0<br>R1(config-if)#ip address 192.168.1.1 255.255.255.0<br>R1(config-if)#no shutdown<br>R1(config-if)#<br>*Apr 25 02:10:46.350: %LINK-3-UPDOWN: Interface GigabitEthernet0/0, changed state to up<br>*Apr 25 02:10:47.350: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/0, changed state to up<br>R1(config-if)#ip nat inside<br>R1(config-if)#<br>*Apr 25 02:11:43.017: %LINEPROTO-5-UPDOWN: Line protocol on Interface NVI0, changed state to up<br>R1(config-if)#interface g0/1<br>R1(config-if)#ip nat outside<br>R1(config-if)#access-list 1 permit 192.168.1.0 0.0.0.255<br>R1(config)#ip nat inside source list 1 interface g0/1 overload<br>R1(config)#</pre><p>Finally, we configure the router to be a DHCP server itself, so we wouldn’t need to configure IP addresses manually on every end device connected to it, which gives us convenience.</p><pre>R1(config)#ip dhcp pool MYPOOL<br>R1(dhcp-config)#network 192.168.1.0 /24<br>R1(dhcp-config)#dns-server 8.8.8.8<br>R1(dhcp-config)#default-router 192.168.1.1<br>R1(dhcp-config)#end<br>R1#<br>*Apr 25 02:18:18.654: %SYS-5-CONFIG_I: Configured from console by console<br>R1#show ip dhcp binding<br>Bindings from all pools not associated with VRF:<br>IP address          Client-ID/              Lease expiration        Type<br>                    Hardware address/<br>                    User name<br>192.168.1.2         010c.9a59.0e00.00       Apr 26 2025 02:17 AM    Automatic<br>R1#</pre><p>At this point, Kali is able to access internet.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/967/0*xs-ykt10Bl7UuqXr.png" /></figure><p>Sometimes, Kali may not detect the change of a new DHCP server (the router) coming online after you finished the router configuration, and it might not know that it should reach out for any offers proactively. By default, Kali doesn’t have dhclient installed, so you cannot force a discovery like on Windows. The simplest way here is just to manually restart the Kali machine. It will perform DHCP discoveries on start-up, so it will detect and use your new router DHCP server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/920/0*vtolMpk-XwtG7-Ka.png" /><figcaption>Kali successfully getting a lease from our router DHCP server</figcaption></figure><p>If you ever decide to replace “Cloud” with “NAT Cloud” in the same topology we had above, the configuration is going to be a bit different.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/369/0*ciNR_bUr3cq7jrcn.png" /></figure><p>If the NAT cloud doesn’t respond to the router’s DHCP discoveries, you will need to configure a static IP on the router’s selected WAN interface.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/485/0*_ffkHAlpQlWmpX0i.png" /></figure><pre>R1#config terminal<br>Enter configuration commands, one per line.  End with CNTL/Z.<br>R1(config)#interface g0/1<br>R1(config-if)#ip address 192.168.122.100 255.255.255.0<br>R1(config-if)#no shutdown<br>R1(config-if)#do show ip route<br>Codes: L - local, C - connected, S - static, R - RIP, M - mobile, B - BGP<br>       D - EIGRP, EX - EIGRP external, O - OSPF, IA - OSPF inter area<br>       N1 - OSPF NSSA external type 1, N2 - OSPF NSSA external type 2<br>       E1 - OSPF external type 1, E2 - OSPF external type 2<br>       i - IS-IS, su - IS-IS summary, L1 - IS-IS level-1, L2 - IS-IS level-2<br>       ia - IS-IS inter area, * - candidate default, U - per-user static route<br>       o - ODR, P - periodic downloaded static route, H - NHRP, l - LISP<br>       a - application route<br>       + - replicated route, % - next hop override, p - overrides from PfR<br><br>Gateway of last resort is not set<br><br>      192.168.1.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.1.0/24 is directly connected, GigabitEthernet0/0<br>L        192.168.1.1/32 is directly connected, GigabitEthernet0/0<br>      192.168.122.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.122.0/24 is directly connected, GigabitEthernet0/1<br>L        192.168.122.100/32 is directly connected, GigabitEthernet0/1<br>R1(config-if)#ip route 0.0.0.0 0.0.0.0 g0/1 192.168.122.1<br>R1(config)#do show ip route<br>Codes: L - local, C - connected, S - static, R - RIP, M - mobile, B - BGP<br>       D - EIGRP, EX - EIGRP external, O - OSPF, IA - OSPF inter area<br>       N1 - OSPF NSSA external type 1, N2 - OSPF NSSA external type 2<br>       E1 - OSPF external type 1, E2 - OSPF external type 2<br>       i - IS-IS, su - IS-IS summary, L1 - IS-IS level-1, L2 - IS-IS level-2<br>       ia - IS-IS inter area, * - candidate default, U - per-user static route<br>       o - ODR, P - periodic downloaded static route, H - NHRP, l - LISP<br>       a - application route<br>       + - replicated route, % - next hop override, p - overrides from PfR<br><br>Gateway of last resort is 192.168.122.1 to network 0.0.0.0<br><br>S*    0.0.0.0/0 [1/0] via 192.168.122.1, GigabitEthernet0/1<br>      192.168.1.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.1.0/24 is directly connected, GigabitEthernet0/0<br>L        192.168.1.1/32 is directly connected, GigabitEthernet0/0<br>      192.168.122.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.122.0/24 is directly connected, GigabitEthernet0/1<br>L        192.168.122.100/32 is directly connected, GigabitEthernet0/1<br>R1(config)#do ping 8.8.8.8<br>Type escape sequence to abort.<br>Sending 5, 100-byte ICMP Echos to 8.8.8.8, timeout is 2 seconds:<br>!!!!!<br>Success rate is 100 percent (5/5), round-trip min/avg/max = 16/17/18 ms<br>R1(config)#</pre><p>Other than that, no configuration changes need to be made for other services we configured on the router before (DHCP, NAT).</p><h4>Properly Managing apt and Downloading Layer2/3 Attack Frameworks</h4><p>Kali is able to access internet, so we can do this easily, e.g. installing Yersinia, finally.</p><pre>sudo apt update &amp;&amp; sudo apt install yersinia</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/0*D1qChaa3Zyd-GPyw.png" /></figure><h4>Saving the “Running-Config” of Kali</h4><p>You probably already know that, with Cisco devices, to permanently save the running config, you need to issue special commands, otherwise all config changes will be lost on the next reboot.</p><pre>R1#write<br>Building configuration...<br>[OK]<br>R1#<br>*Apr 25 02:53:38.315: %GRUB-5-CONFIG_WRITING: GRUB configuration is being updated on disk. Please wait...<br>*Apr 25 02:53:38.976: %GRUB-5-CONFIG_WRITTEN: GRUB configuration was written to disk successfully.<br>R1#</pre><pre>SW1&gt;enable<br>SW1#copy running-config startup-config<br>Destination filename [startup-config]?<br>Building configuration...<br>Compressed configuration from 3203 bytes to 1711 bytes[OK]<br>SW1#<br>*Apr 25 02:57:09.164: %GRUB-5-CONFIG_WRITING: GRUB configuration is being updated on disk. Please wait...<br>*Apr 25 02:57:09.817: %GRUB-5-CONFIG_WRITTEN: GRUB configuration was written to disk successfully.<br>SW1#</pre><p>With a Kali machine, if you are using live boot, then all changes will be lost on the next reboot. But thankfully, we are booting from a virtual hard disk, so all changes will be permanently saved automatically.</p><p>Let’s do a test. Create a text file on Kali’s desktop.</p><pre>cd ~/Desktop<br>printf &quot;Some text&quot; | tee test.txt</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OaSfqbF1YUw_03U3.png" /></figure><p>Shut down the entire GNS3 client directly (GNS3 VM will be stopped as well) and re-open the same project. We see our text file got preserved, proving that changes to the hard disk are non-volatile.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/349/0*LAQqukLztkhGhoXL.png" /></figure><p>Finally, you can start doing whatever you want now. Enjoy!</p><h3>One Note on Yersinia</h3><p>It seems that the GUI is broken on Kali Linux 2025.1 if you were to launch it with the -G option. <a href="https://github.com/tomac/yersinia/issues/86">This is a known compatability issue on newer versions of Kali</a> and still haven’t been fixed at the time of this writing.</p><p>There are some rough workarounds like:</p><pre>git clone https://github.com/tomac/yersinia /opt/yersinia<br>apt install autoconf libgtk-3-dev libnet1-dev libgtk2.0-dev<br>cd /opt/yersinia<br>./autogen.sh<br>./configure<br>make<br>make install</pre><p>But I just ended up using the TUI, which was enough for me to simulate DHCP starvation attacks against DHCP servers.</p><p>Enjoy!</p><h3>Summary</h3><p>The friction of installing Kali Linux in GNS3 mainly comes from outdated version support, but as we have seen, there are solid workarounds for it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bef23a8d86db" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Upgrading GNS3—From Client to VM (Server)—The Complete Walkthrough]]></title>
            <link>https://medium.com/@xyzzklk/upgrading-gns3-from-client-to-vm-server-the-complete-walkthrough-9646a1c7c80e?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/9646a1c7c80e</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[gns3]]></category>
            <category><![CDATA[cisco]]></category>
            <category><![CDATA[servers]]></category>
            <category><![CDATA[networking]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 05:18:49 GMT</pubDate>
            <atom:updated>2026-11-18T12:18:09.247Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was originally documented in 2025 (<a href="https://blog.csdn.net/xyzzklk/article/details/147447551">here</a>, and <a href="https://blog.csdn.net/xyzzklk/article/details/147447075">here</a> for the troubleshooting part) and now republished on Medium. If you find it mentioning some details that no longer match the latest versions of GNS3, please don’t mind.</blockquote><p>GNS3 is a network emulator for learning computer networking without using real devices (routers, switches), similar to Cisco Packet Tracer. Personally, I prefer it over Packet Tracer because it runs real device images, providing behaviour closer to real networking devices, unlike the fully simulated Packet Tracer, in which devices can produce many behaviour misalignments compared to real Cisco hardware.</p><p>The trade-off is that GNS3 requires a Linux based environment to run real device OSes with KVM, which means that if your computer doesn’t run Linux, you’ll need to run a separate VM (basically a GNS3 Server) with nested virtualisation to run real device images in it. In my case, I’m running GNS3 client on my Windows 11 PC, with a Hyper-V GNS3 VM on the same machine with nested virtualisation enabled serving as the “GNS3 VM” running router and switch images.</p><p>We often need to upgrade GNS3. This means we’ll need to do it for both the client and the VM every time, which can make things super complicated, and this deserves a thorough guide.</p><p>Let’s start from the client.</p><blockquote><strong>Note on GNS3 VM with Hyper-V</strong></blockquote><blockquote>Sometimes, after running for a while, all network communications are stalled among virtual devices within the VM, even with the correct configuration.</blockquote><blockquote><a href="https://github.com/GNS3/gns3-gui/issues/3640">This is a known problem</a> and the maintainers have yet to come up with a proper explanation or solution. Some suspect this is due to a problem with Hyper-V itself and suggest switching to another hypervisor like VMWare.</blockquote><h3>Upgrading the GNS3 Client</h3><h4>Downloading the Latest Installer</h4><p>This part is relatively easy. Simply <a href="https://www.gns3.com/software/download">download the new installer</a> and install, and it’ll automatically overwrite the old version. I’ll choose the version matching my OS, Windows.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GJ4I_yNSw3ZaJFjp.png" /></figure><p>Annoyingly, GNS3 forces us to log in or sign up. I already have an account, so I log in.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*u044e3J4VYyt6Sgq.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_VpP-jH4VLL3cB5K.png" /></figure><h4>Install</h4><p>Double-click to run the downloaded installer .exe.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/0*Mpsbv-ea07hfq8LS.png" /></figure><p>Accept the agreement.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/745/0*ARYl4RH8kQa2zXS8.png" /></figure><p>I’ll use the default start menu folder.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/738/0*NkmtBNTPcQJHpLZh.png" /></figure><p>We will choose to install default components selected. We’ll update GNS3 VM separately later, so we don’t tick it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/739/0*sngJeV9uxGCxec_l.png" /></figure><p>We specify the path of the current GNS3 installation to let the new installer properly overwrite.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/741/0*5FW1nX2J6eFz-ujG.png" /></figure><p>Then we continue and start the installation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/744/0*_EkcMPWVLaEVb_qo.png" /></figure><p>After a century, it finally finished. We see this pop-up of SolarWinds advertising their product. I just don’t want to bother with it and chose No.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/744/0*-nNSJvAnuOSZqaz7.png" /></figure><p>Now, we choose <em>not to</em> start GNS3. This is because we finished updating the client but not the Server/VM yet. Starting it now will produce errors because of the version mismatch.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/742/0*TMO2LN4IN2tlX0WT.png" /></figure><h3>Updating the GNS3 VM (Hyper-V)</h3><h4>Preparing the VM</h4><p>In my case, since I’m using Hyper-V, I’ll start the VM alone without starting the client.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_ElYPHNm10L11Xaa.png" /></figure><h4>Connecting to the VM</h4><p>Open the VM in Hyper-V. You’ll see a TUI similar to this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kwDWSYzxAn0tL3kN.png" /></figure><p>Press Enter to go back to the main menu.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5fOxcx6CyclN3YA4.png" /></figure><h4>Selecting a Release Channel (Optional)</h4><p>By default, GNS3 VM only upgrades to the next stable version. You can choose to upgrade to an unstable or development version if you want.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PYSX28__16su5EvO.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/787/0*ci17sjmBmWOr3G0z.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/943/0*0qB_EtQhEGm2xaz5.png" /><figcaption>Select the GNS3 release channel and next release to upgrade to</figcaption></figure><p>For me, I’ll stick with the current stable release, so I keep 2.2 highlighted and select OK.</p><h4>Upgrading the VM</h4><p>Go back to the main menu. Select Upgrade.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*lUw4eKuH7kv9VX9-.png" /></figure><p>GNS3 VM warns it’s best to create a checkpoint before upgrading in case something fails.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*ZWZn_VCCyLoty4TS.png" /></figure><p>This is optional. If you do want to make sure things don’t go wrong, create one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*27opXzseNwJdksg5.png" /></figure><p>But for me, I don’t really have anything important saved in the VM, so I’ll just ignore it.</p><p>Choose Yes, and now we see it upgrading.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*mwWuNbqHoCcdIUKJ.png" /></figure><h4>Completing the Upgrade</h4><p>After long waiting, we see the server rebooted with a menu displaying a higher version.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0p1Z-7UlSjeQy7Ig.png" /></figure><p>So, this part is all done.</p><h3>Testing the Upgraded Version</h3><p>The moment of truth has arrived. Before launching the client (VM is still running now), make sure you always launch the executable with Administrator privileges. Otherwise, it may fail to call Hyper-V to start the VM automatically if later launching the client directly from a cold start (VM off).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SulWVI7_pyx8GxX9.png" /></figure><p>Start the client and wait for a while.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-UIKjdf3k2NDVWAI.png" /></figure><p>All good!</p><h3>Troubleshooting</h3><p>This might not be applicable to you, but I encountered one weird error when upgrading GNS3 VM.</p><p>In an initial attempt, the VM stalled at this display right after I selected Yes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7XDpjXwaLjLXfsYy.png" /></figure><p>This is <em>unhelpful </em>and doesn’t provide any reason.</p><p>Go back to the main menu, and test network connection, just to see if anything’s wrong.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nK-I61x8xYYya1-y.png" /></figure><p>After about 5 seconds, GNS3 claims that “it was not able to connect to the Internet”.</p><p>Choose the Shell option to enter the Linux command line, and ping a random website to test Internet connectivity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*pJp8R6k44eyMB7vj.png" /></figure><p>But ping works, so no connectivity issues at all at routing level.</p><p>Later on, in my case, I found that I was accidentally connected to WARP (Cloudflare’s VPN). Bruh.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/517/0*bI3SlzOw7u1FrfxI.png" /></figure><p>Switched it off, and Internet connectivity test succeeded.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*aGoTZgmCfriyJ1ll.png" /></figure><p>This time, selecting Yes in the Upgrade page leads us to a page showing something like this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*RSNPQz0tmBZrXHNQ.png" /></figure><p>The VM finally makes it to the next step and shows a bunch of versions selectable and asks us which one we want to upgrade to.</p><p>That was kind of silly, but problem solved.</p><h3>Summary</h3><p>Upgrading GNS3 is a long and complicated process, but if steps are performed correctly in order, the process will be relatively smooth.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9646a1c7c80e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Nightmare with Hyper-V Checkpoints: Unable to Edit Virtual Hard Disk]]></title>
            <link>https://medium.com/@xyzzklk/the-nightmare-with-hyper-v-checkpoints-unable-to-edit-virtual-hard-disk-c3985733c20c?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/c3985733c20c</guid>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[windows]]></category>
            <category><![CDATA[virtualization]]></category>
            <category><![CDATA[hyper-v]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Mon, 02 Nov 2026 08:31:01 GMT</pubDate>
            <atom:updated>2026-11-18T12:17:23.422Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/147439231">originally documented in 2025</a> and now republished on Medium. If you find it mentioning some details that no longer match the current interface of Hyper-V, please don’t mind.</blockquote><h3>The Situation</h3><p>Hyper-V checkpoints have the reputation of being unstable, and today we encountered one of these problems ourselves.</p><p>In my environment, we see this virtual machine and want to edit its virtual hard disk file.</p><figure><img alt="Edit is not available because checkpoints exist for this virtual machine." src="https://cdn-images-1.medium.com/max/1024/0*MXLEWqEAQZ7duAyX.png" /><figcaption>Warning: Edit is not available because checkpoints exist for this virtual machine.</figcaption></figure><p>Yet we can see the Edit button greyed out, with a warning message “Edit is not available because checkpoints exist for this virtual machine.”</p><p>This is, well, by itself, not a problem. Logically, Hyper-V checkpoints almost share the same mechanism of how a parent and child disk work together in a <em>differencing disk</em> setting. The child disk (or the “checkpoint file” in the case of a checkpoint, which you can almost think of as just a regular child disk file with a special checkpoint suffix) stores <em>only changes made</em> relative to the parent disk. If the child disk is in-use, the parent disk must remain unmodified; otherwise, it will corrupt the relationship, and that’s why we can’t edit the “disk”, which is actually the parent disk when there are checkpoints exist.</p><p>The solution is… well, <em>should</em> be simple. Either delete or merge the checkpoint, and the lock is removed. In our case, we don’t really have any changes before the checkpoint to keep, so merging would be the best solution.</p><p>Yet Hyper-V Manager simply shows no checkpoints existing…</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*R0JpK29in_2VG9aD.png" /><figcaption>The selected virtual machine has no checkpoints.</figcaption></figure><p>…while checkpoint files being present in the same directory where the parent disk lives. Weird, huh?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/943/0*-aGo24_N2oFz0LFK.png" /></figure><blockquote>Before performing any alteration, make sure you back up all the disk files. Some users have reported the VM being entirely unrecoverable after performing steps outlined in the rest of this article. This could be due to</blockquote><blockquote>1. Logical errors in this article that the author is unaware of, or</blockquote><blockquote>2. The steps in this article only suiting a narrow scenario: checkpoint metadata mismatches with actual checkpoint files, <strong>while the file exists but metadata are missing.</strong> If your scenario is different from this, be cautious with any suggestion given in this article.</blockquote><blockquote>If you can provide more insight or find the logical flaw that trapped the author, feel free to leave it in the comments.</blockquote><h3>Weird Behaviours</h3><p>As we discussed earlier, let’s merge our checkpoint to our parent disk. Since the GUI doesn’t work, let’s try PowerShell. Note that even though this command says Remove, the actual job it performs is merging checkpoints.</p><pre>Get-VMCheckpoint -VMName &quot;YourVMName&quot; | Remove-VMCheckpoint</pre><p>But this command throws an error saying that “no checkpoints exist for this VM”.</p><p>Even if we try an alternative command, it errors out with some other long, nasty exception.</p><pre>Merge-VHD Path = &quot;ChildDiskPath&quot; DestinationPath = &quot;ParentDiskPath&quot;</pre><h3>Final Solution</h3><p>Lucky enough, Hyper-V manager still provides another way to do it in the GUI. If you go to Hyper-V Manager and click Edit Disk:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*DIT-bKCGefTKexJ-.png" /></figure><p>And select the checkpoint file, not the parent disk:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YdGt0a02Gu0H79Jj.png" /></figure><p>Then select Merge:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bi36--srk1-F2BiP.png" /></figure><p>And merge “To the parent virtual hard disk”:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FCJqc8Z3nFtzvbMg.png" /></figure><p>Then wait for the process to complete:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*I_3Nh9SF0hX-Ed1q.png" /></figure><p>Disk will be editable now, and the checkpoint metadata that Hyper-V manager knows matches the actual files.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9mrFZ6utWYMFXRH5.png" /></figure><blockquote>If Virtual hard disk: under Media still shows the path of the checkpoint merged that no longer exists, try deleting then adding the parent disk file back under your VM’s IDE (or SCSI) controller to remove all old clutter.</blockquote><blockquote>Furthermore, the key point is to match the actual disk files with checkpoint metadata. If for some reason it becomes the opposite (e.g. metadata says that a checkpoint exists but there is no corresponding file) or you messed up in some way, then in the worst case, the safest way is to re-create a brand new VM on the same virtual hard disk files.</blockquote><h3>Summary and Root Cause Analysis</h3><p>The reason for this trouble is unknown, but most likely due to accidental mismatch between Hyper-V’s checkpoint metadata and actual disk files.</p><p>This could happen if a VM is stored on a portable hard drive and has been moved between different machines. A checkpoint was created on one machine with its metadata saved in its local Hyper-V Manager, while the other machine only sees a checkpoint file that suddenly popped up with no matching local metadata when the drive is plugged into it next time.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c3985733c20c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Pitfall That Tripped Many Over When Connecting a Cisco Switch to the Internet or Another…]]></title>
            <link>https://medium.com/@xyzzklk/the-pitfall-that-tripped-many-over-when-connecting-a-cisco-switch-to-the-internet-or-another-ff66d62c9517?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/ff66d62c9517</guid>
            <category><![CDATA[ccna]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[computer-networking]]></category>
            <category><![CDATA[switch]]></category>
            <category><![CDATA[cisco]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Sun, 04 Oct 2026 19:31:01 GMT</pubDate>
            <atom:updated>2026-11-18T12:16:33.854Z</atom:updated>
            <content:encoded><![CDATA[<h3>The Pitfall That Tripped Many Over When Connecting a Cisco Switch to the Internet or Another Network</h3><blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/146429735">originally documented in 2025</a> and now republished on Medium.</blockquote><h3>Background of Problem</h3><p>It is often needed that we manage a Cisco switch remotely, like to SSH into it. However, by default, switches operate at the data link layer. This means it must have a network layer address to accept inbound connections based on the internet protocol (IP). To achieve this, we create an SVI (Switch Virtual Interface) — a virtual interface with an IP address to give it layer-3 access.</p><p>In our example, a switch is in the network 192.168.1.0/24, and it has the IP address 192.168.1.253. The default gateway of this local area network is 192.168.1.254.</p><pre>Switch&gt;enable<br>Switch#configure terminal<br>Enter configuration commands, one per line.  End with CNTL/Z.<br>Switch(config)#interface Vlan 1<br>*Mar 23 02:46:56.686: %LINEPROTO-5-UPDOWN: Line protocol on Interface Vlan1, changed state to down<br>Switch(config-if)#ip addres 192.168.1.253 255.255.255.0<br>Switch(config-if)#no shutdown<br>Switch(config-if)#<br>*Mar 23 02:47:16.454: %LINK-3-UPDOWN: Interface Vlan1, changed state to up<br>*Mar 23 02:47:17.454: %LINEPROTO-5-UPDOWN: Line protocol on Interface Vlan1, changed state to up</pre><p>But this setup is not enough. To make it talk to another machine on the internet or any other machine outside this LAN, it needs to know the default gateway to forward traffic to, so we set 192.168.1.254 as its default gateway.</p><pre>Switch(config)#ip default-gateway 192.168.1.254<br>Switch(config)#do show running-config | include default-gateway<br>ip default-gateway 192.168.1.254<br>Switch(config)#</pre><h3>Symptoms</h3><p>However, when another machine in a different LAN (10.0.0.2/30) pings our switch, it never responds.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*N670KO2ocxiqn1Zg.png" /></figure><h3>Troubleshooting</h3><p>We rechecked the existing configuration on our switch, and didn’t find any obvious errors. However, there is one thing worth noting that is unique in our scenario — the switch is a layer-3 switch.</p><p>If routing has been enabled on a layer-3 switch, it will ignore default gateway settings and rely on its routing table. But even then, it is often said that routing is disabled on a layer-3 switch by default. To make it safe, it is worth checking the configuration.</p><pre>Switch&gt;enable<br>Switch#show running-config | section ip<br>ip cef<br>no ipv6 cef<br> ip address 192.168.1.253 255.255.255.0<br>ip default-gateway 192.168.1.254<br>ip forward-protocol nd<br>ip http server<br>ip http secure-server<br>ip ssh server algorithm encryption aes128-ctr aes192-ctr aes256-ctr<br>ip ssh client algorithm encryption aes128-ctr aes192-ctr aes256-ctr<br>Switch#</pre><p>Well, we didn’t see a line like ip routing. But is it really safe? Let’s check if we can access the routing table.</p><pre>Switch#show ip route<br>Codes: L - local, C - connected, S - static, R - RIP, M - mobile, B - BGP<br>       D - EIGRP, EX - EIGRP external, O - OSPF, IA - OSPF inter area<br>       N1 - OSPF NSSA external type 1, N2 - OSPF NSSA external type 2<br>       E1 - OSPF external type 1, E2 - OSPF external type 2<br>       i - IS-IS, su - IS-IS summary, L1 - IS-IS level-1, L2 - IS-IS level-2<br>       ia - IS-IS inter area, * - candidate default, U - per-user static route<br>       o - ODR, P - periodic downloaded static route, H - NHRP, l - LISP<br>       a - application route<br>       + - replicated route, % - next hop override, p - overrides from PfR<br><br>Gateway of last resort is not set<br><br>      192.168.1.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.1.0/24 is directly connected, Vlan1<br>L        192.168.1.253/32 is directly connected, Vlan1<br>Switch#</pre><p>Surprisingly, <em>the routing table is accessible! </em>Which means routing is <em>on</em>.</p><p>Even weirder, the running-config only reflects the disabled state when we explicitly issue no ip routing.</p><pre>Switch(config)#no ip routing<br>Switch(config)#do show running-config | section ip<br>no ip routing<br>no ip cef<br>no ipv6 cef<br> ip address 192.168.1.253 255.255.255.0<br> no ip route-cache<br>ip default-gateway 192.168.1.254<br>ip forward-protocol nd<br>ip http server<br>ip http secure-server<br>ip ssh server algorithm encryption aes128-ctr aes192-ctr aes256-ctr<br>ip ssh client algorithm encryption aes128-ctr aes192-ctr aes256-ctr<br>Switch(config)#</pre><p>This could be a specific behaviour of the switch model Cisco IOSvL2 15.2. Most people expect the running-config to show nothing about ip routing by default (disabled), and only show this line when it’s enabled. The root cause of it not replying to 10.0.0.2 is because it is relying on its routing table for packet forwarding, and the routing table contains no potential routes to 10.0.0.2, so the switch can do nothing but silently drop it.</p><h3>Solving The Problem</h3><h4>Solution 1:</h4><p>After issuing no ip route, the switch will act according to the default gateway information and disable the routing table.</p><pre>Switch(config)#no ip routing<br>Switch(config)#do show ip route<br>Default gateway is 192.168.1.254<br><br>Host               Gateway           Last Use    Total Uses  Interface<br>ICMP redirect cache is empty<br>Switch(config)#</pre><p>And now 10.0.0.2, another Cisco device, is able to ping this switch.</p><pre>Type escape sequence to abort.<br>Sending 5, 100-byte ICMP Echos to 192.168.1.253, timeout is 2 seconds:<br>.!!!!<br>Success rate is 80 percent (4/5), round-trip min/avg/max = 2/2/3 ms</pre><h4>Solution 2:</h4><p>Of course we can choose to keep routing on, depending on your needs. In this case, we will need to add a default route, which serves the same purpose as a default gateway configuration, or any other route to make 10.0.0.2 reachable.</p><pre>Switch(config)#ip route 0.0.0.0 0.0.0.0 192.168.1.254<br>Switch(config)#do show ip route<br>Codes: L - local, C - connected, S - static, R - RIP, M - mobile, B - BGP<br>       D - EIGRP, EX - EIGRP external, O - OSPF, IA - OSPF inter area<br>       N1 - OSPF NSSA external type 1, N2 - OSPF NSSA external type 2<br>       E1 - OSPF external type 1, E2 - OSPF external type 2<br>       i - IS-IS, su - IS-IS summary, L1 - IS-IS level-1, L2 - IS-IS level-2<br>       ia - IS-IS inter area, * - candidate default, U - per-user static route<br>       o - ODR, P - periodic downloaded static route, H - NHRP, l - LISP<br>       a - application route<br>       + - replicated route, % - next hop override, p - overrides from PfR<br><br>Gateway of last resort is 192.168.1.254 to network 0.0.0.0<br><br>S*    0.0.0.0/0 [1/0] via 192.168.1.254<br>      192.168.1.0/24 is variably subnetted, 2 subnets, 2 masks<br>C        192.168.1.0/24 is directly connected, Vlan1<br>L        192.168.1.253/32 is directly connected, Vlan1<br>Switch(config)#</pre><h3>Conclusion</h3><p>Behaviour differences in switch models cause issues.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ff66d62c9517" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Completely Uninstalling WSUS]]></title>
            <link>https://medium.com/@xyzzklk/completely-uninstalling-wsus-c8e4d9d478d3?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/c8e4d9d478d3</guid>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[servers]]></category>
            <category><![CDATA[system-administration]]></category>
            <category><![CDATA[microsoft]]></category>
            <category><![CDATA[windows-server]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Wed, 30 Sep 2026 11:31:01 GMT</pubDate>
            <atom:updated>2026-11-18T12:13:37.768Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/146493575">originally documented in 2025</a> and now republished on Medium. If you find it mentioning some legacy technology like the Windows Server Update Servies and Windows Internal Database, please don’t mind.</blockquote><h3>Cookie Crumbs of WSUS</h3><p>Sometimes we want to perform a clean removal of WSUS, no matter for troubleshooting or just a regular uninstall. However, it is common to find a lot of its components still left around even after a standard uninstall, making tasks that require a clean removal, like troubleshooting, harder. This often results in unnecessary consumption of storage space or a re-install with all previous data preserved. The latter makes troubleshooting requiring cache clearing painful.</p><h3>Performing a Clean Removal</h3><h4>A Standard Removal</h4><p>Many might think that removal is to simply untick WSUS in Server Manager, which is indeed how a standard removal looks like.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Xa9Ts7TAHhrM9iNS.png" /></figure><p>But this will leave many related components, features and cache behind. As a result, we must explicitly choose to remove them as well.</p><h4>Dealing with Dependent Components</h4><p>To start with, IIS is something that WSUS heavily relies on. In my scenario, nothing else is dependent on IIS, so we can safely remove it entirely. Make sure to click Remove Features when Windows Server asks to remove dependent features of IIS with their management tools. If you accidentally missed that pop-up window, you could close Remove Roles and Features Wizard entirely and repeat the process again, so the window will pop up the next time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ChymZGsNVSEKFJrT.png" /></figure><p>Also deselect management tools of WSUS to remove them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/759/0*PTDdB_mg7ubHBFx_.png" /></figure><p>And most importantly, don’t forget WID, which stands for “Windows Internal Database”. It is the default database WSUS uses if Microsoft SQL Server is not selected.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/713/0*enix0bnNHUuQwML6.png" /></figure><p>Finally, those are the most important related components of WSUS, and we can click Remove.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/785/0*0_lMzixuJlSEMGmU.png" /></figure><p>Reboot your server after the process finishes.</p><h4>Removing Remaining Cache</h4><p>This doesn’t mark the end of journey of our removal process, as there are still many stale files hanging around.</p><p>Our “troublesome” WID still leaves an entire folder of cache not being automatically deleted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/671/0*tf7FLwJcLshH9RXp.png" /></figure><p>Delete the entire C:\Windows\WID folder and its contents.</p><p>There are still cache in WSUS’s default installation directory, C:\Program Files\Update Services. Delete this folder as well.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/712/0*zAiuOEQt--fjKXB6.png" /></figure><p>And the one last thing — WSUS content, which consists of two subfolders. WsusContent is where WSUS stores update binary files like Windows Updates, drivers and applications. UpdateServicesPackages is where third-party updates (e.g. custom applications, custom drivers etc) you want to publish are stored.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/628/0*jVskcScuy8zv9VoX.png" /></figure><p>You can delete the whole WSUS Content folder.</p><p>Up to this point, you have completed a clean removal of WSUS.</p><h3>Conclusion</h3><p>Performing a clean removal of WSUS is tricky, as the default uninstall path leaves many components and cache behind. One could argue that this is potentially for safety reasons, but it causes severe frustration if one is not familiar with the locations of cache stored or what kinds of cache are left behind.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c8e4d9d478d3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Solving VMWare Error “Error while opening the virtual machine: Internal error.”]]></title>
            <link>https://medium.com/@xyzzklk/solving-vmware-error-error-while-opening-the-virtual-machine-internal-error-c7c84256415c?source=rss-42770bc6d4d7------2</link>
            <guid isPermaLink="false">https://medium.com/p/c7c84256415c</guid>
            <category><![CDATA[servers]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[virtualization]]></category>
            <category><![CDATA[vmware]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <dc:creator><![CDATA[Xyzzklk]]></dc:creator>
            <pubDate>Sun, 16 Aug 2026 00:01:02 GMT</pubDate>
            <atom:updated>2026-11-18T12:12:22.459Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>This article was <a href="https://blog.csdn.net/xyzzklk/article/details/146103053">originally documented in 2025</a> and now republished on Medium.</blockquote><h3>The Problem</h3><p>Sometimes you might try to open a VMWare virtual machine and see this frustrating error.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/453/0*uYAnLoWlqHW93tIb.png" /><figcaption>Error while opening the virtual machine: Internal error.</figcaption></figure><h3>Solving the Problem</h3><p>First, navigate to the directory where the VM is at. Then, you’ll notice a folder named [VM&#39;s name].vmx.lck.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*y-OsRdg8JV7UNPtN.png" /></figure><p>Delete this folder.</p><p>Problem solved.</p><h3>Reasoning</h3><p>Why would this happen?</p><p>Honestly, I am not sure, but here is a possible explanation.</p><p>The [VM&#39;s name].vmx.lck folder holds “lock files”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7HgvwATEPqmGiNXV.png" /><figcaption>Lock files end with the “.lck” suffix.</figcaption></figure><p>Their purpose is to ensure that the VM files are not modified by other processes when the VM is running, which could cause the entire VM to fail. However, sometimes, due to reasons like the VM not being shut down properly, VMWare may fail to delete the lock files even when the VM is shut down, causing its files to remain in a “protected state”, resulting in even VMWare itself not being able to access it properly.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c7c84256415c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>