<?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 Prem Sumetpong on Medium]]></title>
        <description><![CDATA[Stories by Prem Sumetpong on Medium]]></description>
        <link>https://medium.com/@sumetpong?source=rss-f36f9ed49c60------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dmbNkD5D-u45r44go_cf0g.png</url>
            <title>Stories by Prem Sumetpong on Medium</title>
            <link>https://medium.com/@sumetpong?source=rss-f36f9ed49c60------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 11:47:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sumetpong/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[Patch eMudhra, Then Inventory Your Trust Stores While You’re There]]></title>
            <link>https://medium.com/@sumetpong/patch-emudhra-then-inventory-your-trust-stores-while-youre-there-8ef7ed414076?source=rss-f36f9ed49c60------2</link>
            <guid isPermaLink="false">https://medium.com/p/8ef7ed414076</guid>
            <category><![CDATA[certificate-authority]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[cryptographic-inventory]]></category>
            <category><![CDATA[public-key-cryptography]]></category>
            <category><![CDATA[cyber-security-awareness]]></category>
            <dc:creator><![CDATA[Prem Sumetpong]]></dc:creator>
            <pubDate>Wed, 28 Oct 2026 02:50:34 GMT</pubDate>
            <atom:updated>2026-10-28T03:29:57.756Z</atom:updated>
            <content:encoded><![CDATA[<p>When eMudhra’s keys broke, the question was simple: which of our clients trust this root? Is your organization able to answer it? A second question is harder. For the third-party systems you connect to, if someone forged one of their certificates, would you know?</p><p>For most backend clients, the answer is no. A script error in eMudhra’s April 2025 key ceremony produced eight intermediate CA private keys small enough to fit in one byte. One of them was the number 4. Anyone can recover them from the public certificates and sign a certificate for any domain, including a partner’s. Nobody stole these keys. They were born broken.</p><p>Browsers were covered within days, Chrome within hours, and no attack has been reported. But the keys stay public, so a forgery can arrive at any time, and a backend client whose trust store was never fixed will still accept it.</p><p>Patching is only the first step. While you patch, inventory where each client’s trust lives, then standardize it.</p><h3>Where trust lives outside the browser</h3><p>A browser is one TLS client. Every other TLS connection your organization makes happens outside it: a Java service calling a payment API, a CI runner pulling packages, a Python job posting to a webhook, a monitoring agent reporting home. Each one gets its trust from somewhere, and most stacks touch several of these places:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S81VOotXVJrESie7eanyxw.png" /></figure><h3>One key, two chains</h3><p>Each of the eight weak keys had two certificates. eMudhra issued one under its own emSign ECC Root CA — G3. Sectigo issued the other, a cross-signature from its AAA Certificate Services root (listed as “Comodo AAA Services root” in some stores). Eight keys, two issuers each: 16 certificates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qeVaNCtuu0OVfdIF_8n9Lg.png" /></figure><p>A client checking a forged certificate does not care which chain it finds. It builds whichever path ends at a root it trusts. Distrusting emSign closes the left path. The right one stays open for any client that still trusts AAA Certificate Services, and an unpatched JDK’s cacerts still does.</p><p>Sectigo revoked its cross-signed copies on October 19. That helps only clients that check revocation, and most clients outside the browser do not. For them the exposure has no end date of its own. It lasts until someone fixes the trust store, or until the certificates expire.</p><p>An inventory that stores trust as a list of root names can answer “is emSign in our stores?” and still miss this.</p><h3>Three lists, not one</h3><p>We have an inventory of certificates issued by our trusted CAs. That answers one part of the eMudhra question. Google and AWS set replacement deadlines of October 27 and 28. A team that knew which of its certificates came from eMudhra, and already had a second CA, could reissue before those dates. A team that did not was searching against a deadline.</p><p>Then come the questions that inventory cannot answer. Do we know who the TLS clients in our ecosystem are? Which CAs are in their app trust bundle? Where does the trust bundle that each app loads actually live? That is the second list: what we trust. It is spread across the OS store, cacerts, certifi, the roots built into Node.js and a Cisco appliance&#39;s root bundle, and it needs an inventory of its own, down to the roots and cross-signs each one holds.</p><p>And the last question: would we know if one of those TLS clients connected to a site whose certificate was forged? Answering it starts with a third list: the external services our applications call, and the CA behind each one. When a CA is distrusted, a partner’s certificate can stop validating in our clients as soon as our trust stores update. AWS’s advisory names two: identity providers behind a load balancer, and origins behind a CDN.</p><p>Usually we would not know. The forgery chains to a root the client trusts, so the handshake succeeds and nothing logs an error. Browsers would catch it through CT. Backend clients do not check CT, and an attacker aiming at them has no reason to log the certificate, so the partner’s own CT monitoring sees nothing either.</p><p>The defenses that would catch it have to be chosen in advance. The strongest is giving that connection a trust store that holds only the CA the partner actually uses, or a private CA you both agree on. Pin only where you control both ends. A pin breaks the moment the partner changes CA, which is exactly what happens in an incident like this one. Revocation helps only if the client checks it from cached revocation lists. Live checks either fail open or cause outages when the responder is down. The narrowed trust store is only possible if the third list exists.</p><p>Planning to switch CAs quickly means preparing all three before the next incident:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3D8ZehG_cxeH8oi27Aoaag.png" /></figure><p>Inventorying is costly, and it may still not be complete. A new base image, a vendor update or a pinned package adds a trust store nobody listed. Inventory alone is the weaker recommendation. Pair it with standardizing: point applications at a small set of approved trust stores, and use the inventory to find drift from that standard. The inventory then does not need to be perfect. It needs to catch the clients that left the standard.</p><p><a href="https://unmitigatedrisk.com/2026/10/the-webpkis-unmanaged-software-supply-chain/">Ryan Hurst</a> goes further and recommends narrowing each application’s trust to the CAs it actually needs. Standardize first, then narrow. Narrowing takes more effort, because you first have to analyze what each application actually uses.</p><h3>One incident, a stack of advisories</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*U8V5MfvP1vCgRmVQKiUnTQ.png" /></figure><p>This table is a snapshot from October 27. For the latest status across vendors, and test hosts to check your own clients, see the <a href="https://emudhra.we-have-a-problem.com/">disclosure site</a>. Each vendor’s own advisory remains the authority for that vendor.</p><p>Spring Boot published nothing, and it did not need to: a Spring Boot service trusts whatever JVM it runs on. That is the gap in one example. A team looking for a Spring Boot advisory finds none, while its exposure sits one layer down, in the JDK’s cacerts.</p><h3>So what does it take to be protected?</h3><p>To be definitively protected, every one of these has to be true at once:</p><ol><li>The vendor has shipped a fix. An announcement that a fix is coming is not a fix: Apple’s command-line fix will arrive “in a future software update”, and Cisco dates some fixes to December 8. Oracle’s fix covers only Java versions under Premier or Extended Support.</li><li>Someone has tested the patch or instructions and pushed the change to production.</li><li>They did it in every location: every host, every base image, every rebuilt binary, every appliance.</li></ol><p>I have experienced all three: waiting for the vendor first, then for a team to incorporate the fix and test it, and then for a change to push it to production.</p><p>If any one of them is false, that client is still exposed. For a router, printer or appliance on frozen firmware, the first condition may never be true, and it will trust these certificates until they expire in 2039, as Scott Helme points out.</p><p>Each condition sits with someone else: the vendor’s release date, the team that patches, the team that owns the image. That is software supply chain risk. As <a href="https://unmitigatedrisk.com/2026/10/the-webpkis-unmanaged-software-supply-chain/">Ryan Hurst</a> puts it, a root store is “a set of security dependencies, and we should manage it like one.” Your protection depends on other people’s schedules, and you cannot see those dependencies without knowing where your trust lives.</p><p>There is also no standard for where a CA bundle lives or how it gets updated, so each row in the advisory table has its own location and its own update path. That is why changing a CA, or responding to a compromised one, costs time and resources.</p><p>The same incident also arrived as at least four CVEs, scored anywhere from medium (6.8 for Red Hat’s system store) to critical (9.1 for Oracle Java), depending on the component. A scanner reports each one separately, and the components with no CVE of their own, like a Spring Boot service or an appliance, do not show up at all.</p><p>Then test, because a version number is a claim. The <a href="https://emudhra.we-have-a-problem.com/">disclosure site</a> runs test hosts that serve chains through the compromised intermediates. Point each client at one. A certificate error means it rejects the chain. A successful connection means it still trusts a key whose value is 4.</p><h3>The next one</h3><p>Chrome’s removal of the emSign ECC Root CA — G3 from its root store was scheduled for October 27. Your users will not notice.</p><p>The same inventory pays off again when the CA in trouble is yours, though not in the same way for every CA.</p><p>A compromised private root is a valid scenario in the threat model. If it happened, every copy of that root would have to come out: cacerts imports, custom bundles in container images, mutual TLS trust stores, appliances. No browser vendor will push a blocklist for you, and no distribution will ship an advisory. The inventory is the only map of where that root lives, and it lets you add a replacement root to every bundle before you switch. The standard keeps that map short enough to stay current.</p><p>If a public CA you buy certificates from were distrusted, the first job is reissuing your certificates from a second CA, which is the first list. The trust-store inventory helps wherever you narrowed trust to that CA, through pins or per-partner bundles, because each of those has to change too.</p><p>eMudhra’s own customers are living this now. The company announced on October 22 that it is leaving the global TLS business, and roughly 30,800 certificates have to move to another CA.</p><p>Follow up with the disclosure site and check with your vendors. As each fix is applied, take the opportunity to inventory: record where that app’s CA bundle lives, so the different flavors of CA bundle locations are known. Then follow up with standardization changes.</p><p>If one of your CAs were distrusted or compromised tomorrow, which list would you open first?</p><p>If it were a public CA, I would open the certificate inventory first and reissue from the second CA, then the trust stores where we narrowed trust to it. If it were our private CA, I would open the trust-store inventory first, because the new root has to be in every bundle before a single certificate can move.</p><h3>Sources</h3><ul><li><a href="https://scotthelme.co.uk/eight-bits-of-trust-emudhras-ca-keys-cracked-in-milliseconds/">Scott Helme: Eight bits of trust, eMudhra’s CA keys cracked in milliseconds</a></li><li><a href="https://unmitigatedrisk.com/2026/10/the-webpkis-unmanaged-software-supply-chain/">Ryan Hurst: The WebPKI’s unmanaged software supply chain</a></li><li><a href="https://bleepingcomputer.com/news/security/mozilla-google-apple-block-compromised-emudhra-ca-certificates">BleepingComputer: Mozilla, Google, Apple block compromised eMudhra CA certificates</a></li><li><a href="https://theprint.in/tech/whats-the-emudhra-security-scare-that-has-rattled-the-internet-all-you-need-to-know/3077744/">ThePrint: distrust vs. private key compromise</a></li><li><a href="https://inc42.com/buzz/emudhras-security-key-flaw-what-went-wrong-and-why-google-mozilla-stepped-in/">Inc42: Google’s October 27 replacement deadline</a></li><li><a href="https://cycode.com/blog/emudhra-intermediate-ca-compromise/">Cycode: why patching the OS is not enough</a></li><li><a href="https://bell-sw.com/blog/distrusting-compromised-emudhra-intermediates-in-java/">BellSoft: distrusting the intermediates in Java</a></li><li><a href="https://glaforge.dev/posts/2026/10/21/the-emudhra-ca-compromise-why-your-jvm-is-vulnerable-and-how-to-fix-it/">Guillaume Laforge: the cross-sign trap in cacerts</a></li><li><a href="https://emudhra.we-have-a-problem.com/">Disclosure site and test hosts</a></li><li>Vendor advisories: <a href="https://support.apple.com/en-us/149555">Apple</a>, <a href="https://www.microsoft.com/en-us/msrc/blog/2026/10/microsoft-response-to-compromised-emudhra-certificate-authority-keys">Microsoft (Windows)</a>, <a href="https://azure.github.io/AppService/2026/10/22/emudhra-certificate-issue-app-service-windows.html">Microsoft (Azure App Service)</a>, <a href="https://access.redhat.com/security/vulnerabilities/RHSB-2026-012">Red Hat</a>, <a href="https://ubuntu.com/security/notices/USN-8944-1">Ubuntu</a>, <a href="https://lists.debian.org/debian-security-announce/2026/msg00487.html">Debian</a>, <a href="https://aws.amazon.com/jp/security/security-bulletins/2026-150-aws/">AWS</a>, <a href="https://openjdk.org/emudhra">OpenJDK</a>, <a href="https://www.oracle.com/security-alerts/alert-cve-2026-91403.html">Oracle</a>, <a href="https://adoptium.net/news/2026/10/temurin-october-2026-emudhra">Eclipse Temurin</a>, <a href="https://www.ibm.com/support/pages/node/7293525">IBM</a>, <a href="https://go-review.googlesource.com/c/go/+/851047">Go</a>, <a href="https://discuss.python.org/t/information-for-python-users-regarding-emudhra-emsign-ca-certificate-security-incident/109576">Python</a>, <a href="https://nodejs.org/en/blog/vulnerability/october-2026-security-releases">Node.js</a>, <a href="https://blog.cloudflare.com/emudhra-certificate-compromise">Cloudflare</a>, <a href="https://www.veeam.com/kb4952">Veeam</a>, <a href="https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-emudhra-info-ScS22A1r">Cisco</a><br><a href="https://thehackernews.com/2026/10/emudhra-to-exit-global-ssltls-business.html">The Hacker News: eMudhra to exit global SSL/TLS business</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8ef7ed414076" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Writing “TLS 1.2+” in Your Security Baselines]]></title>
            <link>https://medium.com/@sumetpong/moving-beyond-tls-1-2-time-for-a-tls-1-3-first-mindset-b78d0d50e03d?source=rss-f36f9ed49c60------2</link>
            <guid isPermaLink="false">https://medium.com/p/b78d0d50e03d</guid>
            <category><![CDATA[cryptography]]></category>
            <category><![CDATA[cyber-security-awareness]]></category>
            <category><![CDATA[tls]]></category>
            <category><![CDATA[post-quantum-cryptography]]></category>
            <dc:creator><![CDATA[Prem Sumetpong]]></dc:creator>
            <pubDate>Sun, 06 Sep 2026 20:14:44 GMT</pubDate>
            <atom:updated>2026-09-09T06:22:54.199Z</atom:updated>
            <content:encoded><![CDATA[<h3>RFC 9851 froze TLS 1.2. The phrase in your baseline is a ceiling now, not a floor.</h3><p>I ran into “TLS 1.2+” three times over the last two months. Once in a vendor’s answer to a questionnaire on data in transit protection. Once as the TLS setting on a load balancer. And once where anyone can go and read it, in <a href="https://docs.cloud.google.com/appengine/docs/standard/secure-minimum-tls">Google App Engine’s own documentation</a>, where the console option is labelled “TLS 1.2+ (Modern ciphers) (Default)”.</p><p>The load balancer offered TLS_RSA_WITH_AES_128_GCM_SHA256. The cipher is AES-128-GCM, i.e. AEAD, modern, nothing a reviewer would flag. The key exchange is static RSA, so any session that lands on that suite has no forward secrecy, and can be read later by whoever holds the private key.</p><p>I did not check whether a client ever picked it, and most current clients would not offer it. That is a different question from why it was on the list at all.</p><p>Nobody did anything wrong. The vendor answered the question honestly. The documentation is accurate. The endpoint met the baseline it was given. “TLS 1.2+” was satisfied at every step, because “TLS 1.2+” does not ask about the key exchange.</p><p>That is why the phrase survives, i.e. nothing breaks and everything passes. And it now points at a protocol the IETF has stopped building on.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9rwbKTCNpbioGpfIM1yHtA.png" /></figure><h3>They are not peers</h3><p>When we tell developers to target “TLS 1.2+”, we treat TLS 1.2 and TLS 1.3 as peers.</p><p>They are not.</p><p>TLS 1.2 lets you configure your way to safety. Pick ECDHE, pick AEAD, refuse CBC, refuse static RSA, and you land somewhere defensible. Every one of those is a decision someone has to make correctly, in every template, on every endpoint, every time.</p><p>TLS 1.3 removed the decisions. Forward secrecy is mandatory. The handshake is encrypted. The suite list is five entries long and the bad options are not in it.</p><p>One protocol is safe when the operator is careful. The other is safe when the operator is busy. That is the difference the baseline should be encoding.</p><h3>What RFC 9851 closed</h3><p><a href="https://www.rfc-editor.org/info/rfc9851/">RFC 9851</a> is titled “TLS 1.2 is in Feature Freeze,” and it is blunter than the title suggests. No changes get approved for TLS 1.2 outside urgent security fixes, with two carve-outs, i.e. new Exporter Labels and new ALPN Protocol IDs. Everything else stops at the registry. New cipher suites, new extensions, new supported groups. The working group will keep fixing what breaks. It will not add anything.</p><p>That decides the next ten years in two places, and the first one is key exchange. Hybrid post-quantum key exchange (i.e. classical ECDHE mixed with ML-KEM) is a new supported group. It lands in TLS 1.3. It does not land in TLS 1.2, and not because nobody wrote the code. The registry is closed.</p><p>That sets a clock you do not control. Traffic captured today does not have to be readable today. Someone storing sessions and waiting needs one thing to be true, that the key exchange was classical.</p><p>This is where the forward secrecy conversation usually stops, and it stops one step early. PFS protects a session key against a future leak of the server’s private key. It does nothing against a machine that can solve the discrete log the session was built on. Hybrid PQC answers the second problem, and it answers it in one protocol version. If your endpoints accept TLS 1.2 because the negotiation allowed it, that exposure is already sitting on somebody else’s disk.</p><p>Certificates are the second place, and the one most baselines have not noticed yet. Post-quantum signatures are large enough to change what a certificate chain costs on the wire. <a href="https://datatracker.ietf.org/doc/draft-ietf-plants-merkle-tree-certs/">Merkle Tree Certificates</a> are the IETF PLANTS working group’s answer, and they are negotiated through Trust Anchor IDs, which is a new TLS extension. New extensions are exactly what RFC 9851 closed. The draft is at revision 05 and is not an RFC, so this is a direction rather than a shipping feature, but it is a direction TLS 1.2 has no registry path to follow.</p><p>The IETF is not the only one moving. NIST is revisiting SP 800-52 Rev 2, and its <a href="https://csrc.nist.gov/News/2026/tls-comment-on-sp-800-52-rev-2">call for comments</a> asks whether servers “should” continue to support TLS 1.2, or whether that can become “may”. Not whether the floor should rise. Whether TLS 1.2 belongs in the recommendation at all. The comment period closed in July, the <a href="https://csrc.nist.gov/projects/crypto-publication-review-project">responses are public</a>, and as of early September NIST has not announced a decision.</p><p>That is what turns “TLS 1.2+” into a ceiling. It names the highest thing your baseline will ever reach, and calls it a minimum.</p><h3>A tiered baseline, not a blanket ban</h3><p>Nobody gets to mandate TLS 1.3 everywhere on Monday. Industrial controllers, unpatchable B2B endpoints, and middleboxes older than the current org chart are all real. The answer is not to lower the baseline for everyone so those endpoints keep passing. It is to name them and fence them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LldFOpMvuq54_RlKrvtGBA.png" /></figure><p>The cipher row says ECDHE, not just AEAD, and that is the row I got wrong the first time I wrote this framework down. “AEAD only” is the usual shorthand for a hardened TLS 1.2 config, and TLS_RSA_WITH_AES_128_GCM_SHA256 satisfies it. AES-128-GCM is AEAD. The endpoint I opened this article with would have passed my own criteria. Constrain the key exchange or the criteria does nothing.</p><p><strong>Tier 1</strong> takes the downgrade path off the table, because there is nothing to downgrade to. It also removes most of what you audit: five cipher suites instead of a list you prune every year. Full handshakes finish in one round trip instead of two. (0-RTT exists for resumed sessions and carries a replay tradeoff, so do not count it as a free win.)</p><p><strong>Tier 2</strong> is where most enterprises actually live. It keeps older Android, older Windows Server, and embedded hardware reachable, and it buys you the logs to find out who is still negotiating down before you set a cutoff date. It also lets the edge deploy hybrid PQC natively without breaking the clients behind it.</p><p>A tier describes a connection, not a system, and that distinction is where CDNs and WAFs live. Either one terminates two TLS legs. The edge leg faces whatever clients you actually have, so its tier follows from that, usually Tier 2 for a public site and Tier 1 for an API or a mobile backend. The origin leg has exactly one client, the CDN or WAF itself, and you configure both ends of it. Nothing about a mixed client base applies there, which makes the origin leg the easiest Tier 1 in the estate.</p><p><strong>Tier 3</strong> is a decision, not a default. It needs a named owner, a review date, and an honest line item for what it costs to keep.</p><p>My own Tier 3 is not an industrial controller. It is two vendors, in two different use cases, both answering that TLS 1.3 is not yet supported. One gave a roadmap. The other gave no date.</p><p>TLS 1.3 was published in 2018, so “not yet” in 2026 is a position rather than a constraint. The two answers are not the same position. A dated roadmap is a vendor doing the work on their own schedule, and it hands you a date to hold them to. No date is a vendor who has not committed to doing it at all, and it hands you nothing. I do not know why the second answer came back that way, and I am not going to guess at a motive I cannot evidence.</p><p>It does not change what I do. Both become Tier 3 entries with a named owner and a review date. For the undated one, the review date is the contract renewal, because that is the only deadline that vendor will feel.</p><h3>What to change</h3><p><strong>Rewrite the words before you touch the configs.</strong> Ban “TLS 1.2+” from your baselines and templates. Replace it with “TLS 1.3 preferred, TLS 1.2 fallback restricted to ECDHE and AEAD.” The ECDHE is not decoration. Leave it out and TLS_RSA_WITH_AES_128_GCM_SHA256 walks straight through the new wording too. Engineers configure what the document says, which is why this is the cheapest change on the list and why everything else follows it.</p><p><strong>Stop inheriting your cloud provider’s widest compatibility window.</strong> Which default you get depends on how you create the resource, and the automated path is the older one.</p><p>AWS documents two defaults for the same Application Load Balancer. In the console you get ELBSecurityPolicy-TLS13-1-2-Res-PQ-2025-09, which AWS <a href="https://docs.aws.amazon.com/elasticloadbalancing/latest/application/describe-ssl-policies.html">recommends by name</a> for the post-quantum transition. From the CLI, CloudFormation or the CDK you get ELBSecurityPolicy-2016-08: TLS 1.0 through 1.2, no 1.3, and a cipher list that includes AES128-GCM-SHA256, the OpenSSL name for the suite at the top of this article. Azure arrives at the same place by a simpler route, defaulting new App Service apps to <a href="https://learn.microsoft.com/en-us/azure/app-service/overview-tls">a TLS 1.2 minimum</a> with TLS 1.3 “fully supported” once you switch it on.</p><p>None of this is a vendor failing. Defaults protect the widest client base, which is their job. It does mean the baseline you wrote is not the baseline you are running unless you pinned it, and pinned it where the resource is actually created:</p><p>nginx</p><pre># Tier 2: TLS 1.3 preferred, TLS 1.2 fallback restricted to PFS + AEAD<br>ssl_protocols TLSv1.3 TLSv1.2;</pre><pre># This list governs the TLS 1.2 fallback only.<br>ssl_ciphers &#39;ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-CHACHA20-POLY1305:ECDHE-RSA-CHACHA20-POLY1305&#39;;</pre><pre># TLS 1.3 suites are selected separately, not by ssl_ciphers.<br>ssl_conf_command Ciphersuites TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256;</pre><pre># ssl_prefer_server_ciphers is deliberately absent. It has no effect on<br># TLS 1.3, and the 1.2 list above already excludes everything we refuse.</pre><p><strong>Measure your handshake before PQC arrives, not after.</strong> ML-KEM public keys and PQC certificates run larger than classical ones, large enough that a hybrid ClientHello can cross an MTU boundary nobody has tested. The failure mode is not a clean error. It is a handshake that hangs on one network path and works everywhere else, which is the most expensive kind of ticket to close. Check MTU and fragmentation handling on the firewalls and load balancers now, while a failure is a lab result instead of an outage.</p><p>Tier 3 is the tier I argue about most. Mine came down to two vendors who have not shipped TLS 1.3 yet. If you run something where TLS 1.2 has to stay, I want to know what forced yours, i.e. the device, the vendor, or the contract. That is the part no standards body can fix for us.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b78d0d50e03d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>