<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/">
    <channel>
        <title>CertAutoPilot Blog</title>
        <link>https://www.certautopilot.com/docs/blog</link>
        <description>CertAutoPilot Docs Blog</description>
        <lastBuildDate>Mon, 09 Nov 2026 00:00:00 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <language>en</language>
        <copyright>Copyright © 2026 CloudNativeWorks.</copyright>
        <item>
            <title><![CDATA[Which of your TLS servers use post-quantum key exchange?]]></title>
            <link>https://www.certautopilot.com/docs/blog/post-quantum-key-exchange-inventory</link>
            <guid>https://www.certautopilot.com/docs/blog/post-quantum-key-exchange-inventory</guid>
            <pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[Hybrid ML-KEM key exchange (X25519MLKEM768) protects today's TLS traffic from harvest-now-decrypt-later attacks. How to check a server by hand with OpenSSL, what the results mean, and how to measure it across a whole estate.]]></description>
            <content:encoded><![CDATA[<p>Post-quantum migration has two very different halves. Replacing certificate signatures (RSA, ECDSA) with post-quantum ones is still years away for public TLS. <strong>Key exchange is already here</strong>: browsers and current TLS libraries negotiate the hybrid group <strong>X25519MLKEM768</strong> by default, and every connection that uses it is protected against traffic recorded today and decrypted by a future quantum computer.</p>
<p>The question for an operator is simple: which of my servers already do this, and which do not?</p>
<!-- -->
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="checking-one-server">Checking one server<a href="https://www.certautopilot.com/docs/blog/post-quantum-key-exchange-inventory#checking-one-server" class="hash-link" aria-label="Direct link to Checking one server" title="Direct link to Checking one server" translate="no">​</a></h2>
<p>OpenSSL 3.5 and later name the negotiated group. Against a server that supports hybrid key exchange:</p>
<div class="language-bash codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-bash codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token builtin class-name" style="color:rgb(189, 147, 249)">echo</span><span class="token plain"> </span><span class="token operator">|</span><span class="token plain"> openssl s_client </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-connect</span><span class="token plain"> www.example.com:443 </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-servername</span><span class="token plain"> www.example.com </span><span class="token operator file-descriptor important">2</span><span class="token operator">&gt;</span><span class="token file-descriptor important">&amp;1</span><span class="token plain"> </span><span class="token operator">|</span><span class="token plain"> </span><span class="token function" style="color:rgb(80, 250, 123)">grep</span><span class="token plain"> </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-i</span><span class="token plain"> group</span><br></div><div class="token-line" style="color:#F8F8F2"><span class="token plain"></span><span class="token comment" style="color:rgb(98, 114, 164)"># Negotiated TLS1.3 group: X25519MLKEM768</span><br></div></code></pre></div></div>
<p>When the server only offers classical groups, there is no such line; OpenSSL prints the classical key instead:</p>
<div class="language-text codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-text codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token plain">Peer Temp Key: X25519, 253 bits</span><br></div></code></pre></div></div>
<p>Two things to know when reading the result:</p>
<ul>
<li class=""><strong>Hybrid key exchange needs TLS 1.3.</strong> A server that negotiates TLS 1.2 is classical, whatever its library supports.</li>
<li class=""><strong>The server's preference wins.</strong> A server can support the hybrid group and still choose X25519 when the client offers both. Supporting a group and using it are different facts.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="what-usually-decides-the-answer">What usually decides the answer<a href="https://www.certautopilot.com/docs/blog/post-quantum-key-exchange-inventory#what-usually-decides-the-answer" class="hash-link" aria-label="Direct link to What usually decides the answer" title="Direct link to What usually decides the answer" translate="no">​</a></h2>
<p>In our measurements, a server built on a current TLS library negotiates the hybrid group out of the box: Node.js 24 with OpenSSL 3.5, and an <code>openssl s_server</code> from OpenSSL 3.5 or later, both chose X25519MLKEM768 with no configuration. Servers come out classical when they are pinned to an explicit group list that predates ML-KEM, when they are limited to TLS 1.2, or when they run an older library. Appliances and load balancers follow their own firmware schedule.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="measuring-a-whole-estate">Measuring a whole estate<a href="https://www.certautopilot.com/docs/blog/post-quantum-key-exchange-inventory#measuring-a-whole-estate" class="hash-link" aria-label="Direct link to Measuring a whole estate" title="Direct link to Measuring a whole estate" translate="no">​</a></h2>
<p>Checking servers one by one does not scale, and the answer changes as libraries are upgraded and configurations drift. CertAutoPilot's discovery records the key exchange of every TLS server it scans and shows it on a <a class="" href="https://www.certautopilot.com/docs/discovery/pqc-readiness">quantum readiness</a> page:</p>
<ul>
<li class="">how many servers use hybrid or pure ML-KEM key exchange, how many are classical, and how many are still on TLS 1.2;</li>
<li class="">optionally, which groups each server <em>supports</em> beyond the one it chose, to find servers that could be hybrid with a configuration change;</li>
<li class="">a daily trend, and a notification when a server that used to negotiate a post-quantum group falls back to a classical one.</li>
</ul>
<p>It also classifies certificate keys, so the second half of the migration, signatures, is visible in the same place when the time comes.</p>]]></content:encoded>
            <category>post-quantum</category>
            <category>pqc</category>
            <category>tls</category>
            <category>discovery</category>
        </item>
        <item>
            <title><![CDATA[47-day TLS certificates: what changes and how to prepare]]></title>
            <link>https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates</link>
            <guid>https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates</guid>
            <pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[The CA/Browser Forum's SC-081 ballot cuts public TLS certificate lifetimes from 398 days to 47 by 2029. What the timeline means in renewals per year, and a practical checklist to get ready.]]></description>
            <content:encoded><![CDATA[<p>Public TLS certificates are getting much shorter lives. The CA/Browser Forum ballot <strong>SC-081</strong> lowers the maximum validity of a publicly trusted TLS certificate in four steps, ending at <strong>47 days in March 2029</strong>. A process that renews a certificate by hand once a year has to do it roughly eight times a year by then, for every certificate.</p>
<!-- -->
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="the-timeline">The timeline<a href="https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates#the-timeline" class="hash-link" aria-label="Direct link to The timeline" title="Direct link to The timeline" translate="no">​</a></h2>






























<table><thead><tr><th>Certificates issued</th><th>Maximum validity</th><th>Renewals a year, per certificate</th></tr></thead><tbody><tr><td>until 15 March 2026</td><td>398 days</td><td>about 1</td></tr><tr><td>from 15 March 2026</td><td>200 days</td><td>about 2</td></tr><tr><td>from 15 March 2027</td><td>100 days</td><td>about 4</td></tr><tr><td>from 15 March 2029</td><td>47 days</td><td>about 8</td></tr></tbody></table>
<p>The same ballot also shortens how long a CA may reuse a domain validation it already performed, so the proof that you control a domain has to be repeated more often too. In practice this means DNS or HTTP validation has to be automated along with the renewal itself.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="why-it-is-harder-than-it-looks">Why it is harder than it looks<a href="https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates#why-it-is-harder-than-it-looks" class="hash-link" aria-label="Direct link to Why it is harder than it looks" title="Direct link to Why it is harder than it looks" translate="no">​</a></h2>
<p>Renewing is rarely the hard part. The work around it is:</p>
<ul>
<li class=""><strong>Finding every certificate.</strong> Certificates that nobody tracks are the ones that expire in production. Load balancers, Windows servers, appliances and Kubernetes clusters each keep their own copy.</li>
<li class=""><strong>Deploying the new certificate everywhere it is used.</strong> One certificate can sit on an F5 virtual server, three IIS sites and a Kubernetes Secret. Renewing it at the CA fixes none of them.</li>
<li class=""><strong>Proving it worked.</strong> A renewal that reached the CA but not the server still expires on schedule.</li>
<li class=""><strong>Keeping humans out of the loop without losing control.</strong> Approvals, audit trails and rollback have to keep working when nobody clicks "renew" any more.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="a-checklist">A checklist<a href="https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates#a-checklist" class="hash-link" aria-label="Direct link to A checklist" title="Direct link to A checklist" translate="no">​</a></h2>
<ol>
<li class=""><strong>Inventory first.</strong> Scan your networks, DNS zones and Certificate Transparency logs and list every certificate, where it is served and who owns it.</li>
<li class=""><strong>Measure your exposure.</strong> Count certificates that are renewed manually, validated with manual DNS records, or deployed by hand. Those are the ones that will break.</li>
<li class=""><strong>Automate validation.</strong> Use DNS-01 with an API-driven DNS provider, or HTTP-01 where a route to the challenge path exists. Manual TXT records do not scale to eight renewals a year.</li>
<li class=""><strong>Automate deployment.</strong> Every place a certificate is installed needs a deployment step that runs after each renewal, without someone logging in to the device.</li>
<li class=""><strong>Verify after deploying.</strong> Check the fingerprint the endpoint actually serves, not the one you uploaded.</li>
<li class=""><strong>Keep a way back.</strong> Being able to roll a target back to the previous certificate version turns a bad deployment into a minor event.</li>
<li class=""><strong>Watch the calendar.</strong> Renewal load concentrates on certain days; know your busiest day before the CA does.</li>
</ol>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="how-certautopilot-helps">How CertAutoPilot helps<a href="https://www.certautopilot.com/docs/blog/preparing-for-47-day-certificates#how-certautopilot-helps" class="hash-link" aria-label="Direct link to How CertAutoPilot helps" title="Direct link to How CertAutoPilot helps" translate="no">​</a></h2>
<p>CertAutoPilot's <a class="" href="https://www.certautopilot.com/docs/web-ui/readiness-47-day">47-Day Readiness dashboard</a> scores a project on exactly these points: how much is automated, how much of it is deployed automatically, and how much still relies on manual DNS. It projects renewals per year at 47 days and shows the busiest renewal day ahead, so you can see the gap before the 2027 and 2029 deadlines rather than after.</p>
<p>Issuance and renewal run through ACME (Let's Encrypt, Google Trust Services, Sectigo and private ACME CAs) or Microsoft AD CS, and each renewal is deployed to its targets automatically, then checked against what the endpoint serves.</p>]]></content:encoded>
            <category>47-day</category>
            <category>renewal</category>
            <category>automation</category>
        </item>
        <item>
            <title><![CDATA[Windows Server 2016 rejects AES-encrypted PFX files]]></title>
            <link>https://www.certautopilot.com/docs/blog/windows-server-2016-aes-pfx</link>
            <guid>https://www.certautopilot.com/docs/blog/windows-server-2016-aes-pfx</guid>
            <pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[A PFX exported with modern AES-256 encryption fails to import on Windows Server 2016 with 'The specified network password is not correct', even with the right password. Why, and how to export a PFX that older Windows accepts.]]></description>
            <content:encoded><![CDATA[<p>A PKCS#12 (<code>.pfx</code>) file exported with today's defaults can fail to import on <strong>Windows Server 2016</strong> with <em>"The specified network password is not correct."</em>, although the password is right. Newer Windows accepts the same file. The difference is the encryption inside the file, not the password.</p>
<!-- -->
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="what-we-measured">What we measured<a href="https://www.certautopilot.com/docs/blog/windows-server-2016-aes-pfx#what-we-measured" class="hash-link" aria-label="Direct link to What we measured" title="Direct link to What we measured" translate="no">​</a></h2>
<p>On Windows Server 2016 (build 14393, PowerShell 5.1.14393) we imported the same certificate and key packaged two ways:</p>




















<table><thead><tr><th>PFX encryption</th><th>Windows Server 2016</th><th>Windows Server 2022</th></tr></thead><tbody><tr><td>AES-256-CBC with PBKDF2 / SHA-256 MAC (OpenSSL 3 default)</td><td><strong>fails</strong></td><td>imports</td></tr><tr><td>TripleDES with SHA-1 (<code>pbeWithSHA1And3-KeyTripleDES-CBC</code>)</td><td>imports</td><td>imports</td></tr></tbody></table>
<p>Both ways of importing fail, each with a password error: loading the file with <code>X509Certificate2</code> says <em>"The specified network password is not correct."</em>, and <code>Import-PfxCertificate</code> says the file <em>requires either a different password</em> or a different protection. Both messages are misleading: Windows Server 2016 cannot read the AES-based protection and reports it as a password problem.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="how-to-export-a-pfx-that-older-windows-can-read">How to export a PFX that older Windows can read<a href="https://www.certautopilot.com/docs/blog/windows-server-2016-aes-pfx#how-to-export-a-pfx-that-older-windows-can-read" class="hash-link" aria-label="Direct link to How to export a PFX that older Windows can read" title="Direct link to How to export a PFX that older Windows can read" translate="no">​</a></h2>
<p>With OpenSSL 3 or later, ask for TripleDES and a SHA-1 MAC explicitly:</p>
<div class="language-bash codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-bash codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token plain">openssl pkcs12 </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-export</span><span class="token plain"> </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-in</span><span class="token plain"> cert.pem </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-inkey</span><span class="token plain"> key.pem </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-out</span><span class="token plain"> legacy.pfx </span><span class="token punctuation" style="color:rgb(248, 248, 242)">\</span><span class="token plain"></span><br></div><div class="token-line" style="color:#F8F8F2"><span class="token plain">  </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-certpbe</span><span class="token plain"> PBE-SHA1-3DES </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-keypbe</span><span class="token plain"> PBE-SHA1-3DES </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-macalg</span><span class="token plain"> sha1</span><br></div></code></pre></div></div>
<p>Check what you produced:</p>
<div class="language-bash codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-bash codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token plain">openssl pkcs12 </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-in</span><span class="token plain"> legacy.pfx </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-info</span><span class="token plain"> </span><span class="token parameter variable" style="color:rgb(189, 147, 249);font-style:italic">-noout</span><br></div></code></pre></div></div>
<p>The output should name <code>pbeWithSHA1And3-KeyTripleDES-CBC</code> for the key and certificate bags and <code>MAC: sha1</code>. A modern file shows <code>PBES2, PBKDF2, AES-256-CBC</code> instead.</p>
<p>TripleDES is the weaker protection, so use it only where an older Windows version needs to read the file, keep the PFX password strong, and delete the file once it is imported.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="in-certautopilot">In CertAutoPilot<a href="https://www.certautopilot.com/docs/blog/windows-server-2016-aes-pfx#in-certautopilot" class="hash-link" aria-label="Direct link to In CertAutoPilot" title="Direct link to In CertAutoPilot" translate="no">​</a></h2>
<p>When CertAutoPilot deploys to IIS or Exchange it builds the one-time PFX in the TripleDES form, so Windows Server 2016 hosts work without any setting. Where you take a PFX yourself (a download, a file written by a Linux or Windows target, or an e-mail attachment), you choose between <strong>Modern (AES-256)</strong>, the default, and <strong>Legacy (TripleDES)</strong> for older systems. See <a class="" href="https://www.certautopilot.com/docs/certificates/download">downloading certificates</a> and the <a class="" href="https://www.certautopilot.com/docs/distribution/iis">IIS target</a>.</p>]]></content:encoded>
            <category>windows</category>
            <category>pfx</category>
            <category>iis</category>
        </item>
        <item>
            <title><![CDATA[WinRM returns 401 for a local administrator: LocalAccountTokenFilterPolicy]]></title>
            <link>https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac</link>
            <guid>https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac</guid>
            <pubDate>Mon, 09 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[A local administrator account gets HTTP 401 from WinRM over NTLM although the password is right. The cause is UAC remote restrictions; how to confirm it in the Security log and fix it with LocalAccountTokenFilterPolicy.]]></description>
            <content:encoded><![CDATA[<p>You point a deployment tool at a Windows server over WinRM, use a <strong>local administrator</strong> account with <strong>NTLM</strong>, and get <strong>HTTP 401</strong>. The password is right; the account is in the local Administrators group. The reason is usually not the password at all but <strong>UAC remote restrictions</strong>.</p>
<!-- -->
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="what-happens">What happens<a href="https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac#what-happens" class="hash-link" aria-label="Direct link to What happens" title="Direct link to What happens" translate="no">​</a></h2>
<p>When UAC is on, Windows gives a <em>local</em> account that signs in over the network a filtered token with administrator rights stripped, unless it is the built-in <code>Administrator</code> account. WinRM requires administrator rights for a remote shell, so it refuses the request with 401 even though the sign-in itself succeeded. Domain accounts are not filtered this way, which is why the same setup often works with a domain account.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="how-to-confirm-it">How to confirm it<a href="https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac#how-to-confirm-it" class="hash-link" aria-label="Direct link to How to confirm it" title="Direct link to How to confirm it" translate="no">​</a></h2>
<p>Look at the <strong>Security</strong> log on the target server at the time of the attempt:</p>
<ul>
<li class=""><strong>Event 4624</strong> (successful logon, logon type 3) followed by the 401 means the password was accepted and the refusal is about rights. This is the UAC case.</li>
<li class=""><strong>Event 4625</strong> means the sign-in itself failed. The <em>Sub Status</em> code says why: <code>0xC000006A</code> wrong password, <code>0xC0000064</code> no such user (often a domain name filled in for a local account), <code>0xC0000072</code> disabled, <code>0xC0000234</code> locked out.</li>
</ul>
<p>Then check the setting:</p>
<div class="language-powershell codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-powershell codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token function" style="color:rgb(80, 250, 123)">Get-ItemProperty</span><span class="token plain"> HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System </span><span class="token punctuation" style="color:rgb(248, 248, 242)">|</span><span class="token plain"></span><br></div><div class="token-line" style="color:#F8F8F2"><span class="token plain">  </span><span class="token function" style="color:rgb(80, 250, 123)">Select-Object</span><span class="token plain"> LocalAccountTokenFilterPolicy</span><span class="token punctuation" style="color:rgb(248, 248, 242)">,</span><span class="token plain"> EnableLUA</span><br></div></code></pre></div></div>
<p><code>EnableLUA</code> = 1 means UAC is on. If <code>LocalAccountTokenFilterPolicy</code> is missing or 0, local administrators are filtered on remote sign-in.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="the-fix">The fix<a href="https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac#the-fix" class="hash-link" aria-label="Direct link to The fix" title="Direct link to The fix" translate="no">​</a></h2>
<div class="language-powershell codeBlockContainer_Ckt0 theme-code-block" style="--prism-color:#F8F8F2;--prism-background-color:#282A36"><div class="codeBlockContent_QJqH"><pre tabindex="0" class="prism-code language-powershell codeBlock_bY9V thin-scrollbar" style="color:#F8F8F2;background-color:#282A36"><code class="codeBlockLines_e6Vv"><div class="token-line" style="color:#F8F8F2"><span class="token function" style="color:rgb(80, 250, 123)">New-ItemProperty</span><span class="token plain"> HKLM:\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System `</span><br></div><div class="token-line" style="color:#F8F8F2"><span class="token plain">  </span><span class="token operator">-</span><span class="token plain">Name LocalAccountTokenFilterPolicy </span><span class="token operator">-</span><span class="token plain">Value 1 </span><span class="token operator">-</span><span class="token plain">PropertyType DWord </span><span class="token operator">-</span><span class="token plain">Force</span><br></div></code></pre></div></div>
<p>No restart is needed. Two notes before you apply it everywhere:</p>
<ul>
<li class="">It applies to <strong>every</strong> local administrator account on the server, not only the one your tool uses. Where you can, prefer a domain account (and Kerberos), which needs no such exception.</li>
<li class="">Set it through Group Policy or your server image rather than by hand, so it survives a rebuild.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="other-401-causes-worth-ruling-out">Other 401 causes worth ruling out<a href="https://www.certautopilot.com/docs/blog/winrm-401-local-administrator-uac#other-401-causes-worth-ruling-out" class="hash-link" aria-label="Direct link to Other 401 causes worth ruling out" title="Direct link to Other 401 causes worth ruling out" translate="no">​</a></h2>
<p>We reproduced the common ones on a lab server; the authentication schemes the server offers in its 401 response narrow it down:</p>

























<table><thead><tr><th>Symptom</th><th>Cause</th></tr></thead><tbody><tr><td>401, server offers <code>Negotiate</code> only, on port 5985</td><td>Unencrypted HTTP is not allowed (<code>AllowUnencrypted=false</code>); use HTTPS on 5986, or allow it</td></tr><tr><td>401, server offers <code>Basic</code> only</td><td>Negotiate (NTLM) authentication is turned off on the WinRM service</td></tr><tr><td>401 with event 4624 and no 4625</td><td>Authenticated but not authorized: not an administrator, or the UAC filtering above</td></tr><tr><td>HTTP 500 "Access is denied"</td><td>The account is only in <em>Remote Management Users</em>; a remote shell needs Administrators</td></tr></tbody></table>
<p>CertAutoPilot's IIS, Windows (WinRM) and Exchange targets report which of these it saw in the error message itself. The full catalogue is in the <a class="" href="https://www.certautopilot.com/docs/distribution/winrm-troubleshooting">WinRM troubleshooting guide</a>.</p>]]></content:encoded>
            <category>windows</category>
            <category>winrm</category>
            <category>iis</category>
            <category>troubleshooting</category>
        </item>
    </channel>
</rss>