TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Let’s Encrypt moving to new CA root

152 pointsby alblueover 4 years ago

10 comments

blitblitblitover 4 years ago
Assuming you aren&#x27;t able to update via a custom ROM on unsupported hardware, it should still be possible to import the new Root Cert on Android.<p>&gt; System-installed certificates can be managed on the Android device in the Settings -&gt; Security -&gt; Certificates -&gt; &#x27;System&#x27;-section, whereas the user trusted certificates are manged in the &#x27;User&#x27;-section there. When using user trusted certificates, Android will force the user of the Android device to implement additional safety measures: the use of a PIN-code, a pattern-lock or a password to unlock the device are mandatory when user-supplied certificates are used.<p>Source: <a href="http:&#x2F;&#x2F;wiki.cacert.org&#x2F;FAQ&#x2F;ImportRootCert#Android_Phones_.26_Tablets" rel="nofollow">http:&#x2F;&#x2F;wiki.cacert.org&#x2F;FAQ&#x2F;ImportRootCert#Android_Phones_.26...</a>
评论 #25329073 未加载
评论 #25351923 未加载
lxgrover 4 years ago
I wonder if this will end up pushing the burden of CA root management to app developers on Android for apps other than browsers too.<p>This really seems like something worth decoupling from Android system updates – possibly even more so than emoji [0]?<p>Realistically, though, it&#x27;s too late for Google to roll out something like this and we&#x27;ll have to live with static copies of Mozilla&#x27;s root CA set that will quickly go stale...<p>On the upside, it might convince apps that only talk to vendor-controlled backends (i.e. with a predictable set of CAs) to finally enable certificate pinning: If they need to think about their trusted root certs themselves, might as well only include the relevant ones.<p>[0] <a href="https:&#x2F;&#x2F;www.xda-developers.com&#x2F;google-prepares-decouple-new-emojis-android-system-updates&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.xda-developers.com&#x2F;google-prepares-decouple-new-...</a>
est31over 4 years ago
See also: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25008748" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=25008748</a>
toast0over 4 years ago
It would be nice if you could send two end-entity certs, with the same public key, and different intermediates, and the other side would pick which ever one it liked.<p>Otherwise, you have to figure out which one root CA all your clients like, because it&#x27;s probably hard to figure out which clients need which certs.
评论 #25330403 未加载
评论 #25330653 未加载
tialaramexover 4 years ago
This article has some errors which are unfortunate.<p>&gt; From the start, Let&#x27;s Encrypt used a top-level certificate (known as a CA Root) that was signed by ISRG Root X1 and cross-signed by IdenTrust<p>Their &quot;top-level&quot; root is in fact ISRG Root X1. This was <i>not</i> cross-signed by IdenTrust (that wouldn&#x27;t be impossible, several popular CAs took this approach, but ISRG, the people behind Let&#x27;s Encrypt, did not).<p>The CAs which were cross-signed were the Let&#x27;s Encrypt intermediates (sometimes called subCAs), so far these have been named Let&#x27;s Encrypt Authority X1, X2, X3 and X4 and now just R3 and R4 (the name has to be sent over the wire every time you connect anywhere, and just &quot;R3&quot; is fewer bytes than &quot;Let&#x27;s Encrypt Authority R3&quot;)<p>[ Why are there intermediates? Well, an issuing CA has to be connected, albeit not directly, to the public Internet. That&#x27;s obviously not the most secure situation conceivable. To protect the roots they can&#x27;t be kept online, they usually live in dedicated hardware modules used only for relatively rare ceremonies like creating a new intermediate. Then the intermediates are kept online. If anything goes wrong it&#x27;s relatively painless to distrust a mere intermediate, whereas distrusting a root causes massive disruption ]<p>Let&#x27;s Encrypt provide a handy diagram:<p><a href="https:&#x2F;&#x2F;letsencrypt.org&#x2F;certificates&#x2F;" rel="nofollow">https:&#x2F;&#x2F;letsencrypt.org&#x2F;certificates&#x2F;</a><p>Anyway, what changes in January is <i>not</i> anything about the certificates you actually get from Let&#x27;s Encrypt. Not one byte. What changes is essentially metadata, starting in January the Let&#x27;s Encrypt backend system will by default stop giving subscribers a chain of certificates that proves this can be traced back to Identrust&#x27;s DST Root CA X3, and instead give you a chain that traces back to ISRG Root X1.<p><i>Your</i> certificate is the same either way, for most people it will say it was issued by R3 (the shorter name I mentioned above, began being used at the start of December so many Let&#x27;s Encrypt users haven&#x27;t got a certificate from R3 yet but will do so automatically when next renewing). What changes is the accompanying certificate for R3 itself, there are two versions of that certificate.<p>If you need to service clients which don&#x27;t yet trust ISRG Root X1 you can (tell your software to) continue to request the other version of R3 that&#x27;s signed by Identrust&#x27;s &quot;DST Root CA X1&quot; until later in 2021 when that expires.<p>But there is <i>no change</i> to the actual end subscriber &quot;leaf&quot; certificate and that&#x27;s a mistake in the linked article.<p>I thought we might see a HN link for Let&#x27;s Encrypt new intermediate R3 going live in the middle of the week, but we didn&#x27;t. So I guess that means it went so well that nobody even noticed.
评论 #25329018 未加载
评论 #25328066 未加载
bawolffover 4 years ago
&gt; but the key problem is dealing with older operating systems that either don&#x27;t have the correct root certificate, or don&#x27;t have support for either the SHA-1 or ECDSA protocols that are used in the process of verifying a certificate.<p>Say what now? Pretty sure sha-1 support for certs is much older than android 7.
评论 #25327901 未加载
评论 #25327903 未加载
mproudover 4 years ago
Yeah, this is 1 month old: <a href="https:&#x2F;&#x2F;letsencrypt.org&#x2F;2020&#x2F;11&#x2F;06&#x2F;own-two-feet.html" rel="nofollow">https:&#x2F;&#x2F;letsencrypt.org&#x2F;2020&#x2F;11&#x2F;06&#x2F;own-two-feet.html</a>
评论 #25327779 未加载
评论 #25327592 未加载
olliejover 4 years ago
Does this mean we&#x27;re going to see a bunch of services fail due to poor certificate pinning implementations?
评论 #25328876 未加载
评论 #25328449 未加载
x87678rover 4 years ago
I wish there was a way to have multiple certificates (1st preference, 2nd for failover) so you can support two or more at at time - with feedback on how many using the failover. Its stressful to make a switch and then see what is broken.
评论 #25329804 未加载
one-word-answerover 4 years ago
Painful