<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Carddav on rHomelab</title><link>https://rhomelab.com/tags/carddav/</link><description>Recent content in Carddav on rHomelab</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 13 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://rhomelab.com/tags/carddav/index.xml" rel="self" type="application/rss+xml"/><item><title>Self-Hosted Calendar and Contacts: Radicale vs Baikal vs Nextcloud for CalDAV and CardDAV</title><link>https://rhomelab.com/self-hosted/self-hosted-calendar-contacts-caldav-carddav/</link><pubDate>Fri, 13 Nov 2026 00:00:00 +0000</pubDate><guid>https://rhomelab.com/self-hosted/self-hosted-calendar-contacts-caldav-carddav/</guid><description>Google Calendar and Google Contacts are two of the hardest Google services to actually leave. Here&amp;#39;s how Radicale, Baikal, and Nextcloud compare for self-hosting your own CalDAV and CardDAV server, and what the migration actually involves.</description><content:encoded><![CDATA[<p>Most self-hosting migrations start with something low-stakes, a media server, a password vault, a note-taking app. Calendar and contacts tend to get put off because the failure mode is scarier: miss an appointment, lose a phone number, and you feel it immediately. But the two protocols behind nearly every calendar and address book app, CalDAV and CardDAV, are open standards, not Google-proprietary, and self-hosting them is one of the more mature corners of the self-hosted world. The question isn&rsquo;t really &ldquo;can I do this,&rdquo; it&rsquo;s which server to run and whether your devices will actually sync against it without drama.</p>
<h2 id="caldav-and-carddav-are-protocols-not-apps">CalDAV and CardDAV are protocols, not apps</h2>
<p>This trips people up on the first search. CalDAV (calendars, events, tasks) and CardDAV (contacts) are both extensions of WebDAV, the same HTTP-based protocol family that&rsquo;s been around since the late 1990s. Any client that speaks CalDAV or CardDAV, your phone&rsquo;s native calendar app, Thunderbird, Apple&rsquo;s Contacts app, can talk to any server that implements the same protocol, the same way any email client can talk to any IMAP server. You&rsquo;re not locked into a matching client-server pair the way you are with, say, a proprietary sync API. This is why self-hosting this layer is viable at all: the client side is already solved, you only need to stand up the server.</p>
<p>The practical catch is the server implementation itself, not the protocol. CalDAV and CardDAV are large, fiddly specs with a long tail of edge cases (recurring event exceptions, timezone handling, free/busy queries), and servers vary in how completely and correctly they implement them. That&rsquo;s the real axis these three options differ on, not features in the marketing sense.</p>
<h2 id="radicale-the-one-that-gets-out-of-your-way">Radicale: the one that gets out of your way</h2>
<p>Radicale is a small Python CalDAV/CardDAV server that stores everything as plain <code>.ics</code> and <code>.vcf</code> files on disk, no database required by default, though it can use one. There&rsquo;s no web admin UI in the traditional sense, you manage users and rights through a config file and a simple auth backend (htpasswd-style files, PAM, or an external script), and calendars/address books get created automatically the first time a client connects and asks for one.</p>
<p>This makes Radicale the right choice when you want calendar and contact sync for yourself, or a small household, and nothing more. Resource footprint is tiny, it runs comfortably in an LXC with a fraction of a core and under 100MB of RAM, and because everything lands as plain text files, backup is just backing up a directory, and recovery from a bad state means reading an <code>.ics</code> file in a text editor to see what&rsquo;s actually in it. The tradeoff is exactly that minimalism: no built-in sharing UI between users, no calendar-invite workflow beyond what the spec itself provides, and if you want to see &ldquo;who owns what&rdquo; at a glance, you&rsquo;re looking at a filesystem tree, not a dashboard.</p>
<h2 id="baikal-a-lightweight-admin-layer-on-a-mature-engine">Baikal: a lightweight admin layer on a mature engine</h2>
<p>Baikal is a PHP application built on top of sabre/dav, the same underlying CalDAV/CardDAV library that powers a good chunk of the self-hosted calendar ecosystem, including pieces of Nextcloud&rsquo;s own implementation. What Baikal adds over using sabre/dav directly is a real web admin interface: create users, create calendars and address books, generate a user&rsquo;s CalDAV/CardDAV connection URLs, all through a browser instead of hand-editing config.</p>
<p>It stores data in SQLite by default (MySQL/MariaDB is supported for anything beyond a handful of users), and runs on any stack that can serve PHP, an LXC with nginx or Apache plus PHP-FPM is the common homelab pattern. This sits in a sensible middle ground: more approachable than Radicale if you want a UI for adding a second household member without editing text files, while still being a single-purpose tool with a much smaller footprint and attack surface than a full groupware suite. sabre/dav&rsquo;s spec coverage is generally regarded as some of the most complete and correct in the open-source CalDAV/CardDAV space, which matters more than it sounds like once you start testing against Apple&rsquo;s and Google&rsquo;s more pedantic client behavior.</p>
<h2 id="nextcloud-calendar-and-contacts-as-apps-not-the-point-of-the-install">Nextcloud: calendar and contacts as apps, not the point of the install</h2>
<p>If you&rsquo;re already running Nextcloud for file sync (see the earlier comparison against Syncthing on this site), its Calendar and Contacts apps give you CalDAV/CardDAV support for free, no separate service to stand up. For a lot of homelabbers this is the actual right answer, not because Nextcloud is the best calendar server technically, but because you already have the infrastructure, the backups, the user accounts, and the reverse proxy config running, and bolting on two more apps costs you nothing you don&rsquo;t already have.</p>
<p>The honest tradeoff: if calendar and contacts are the only thing you want, running the entire Nextcloud stack, PHP, a full database, the app ecosystem, background job cron, just to get CalDAV/CardDAV is a lot of extra moving parts and attack surface for a job Radicale does in a fraction of the resource footprint. Nextcloud&rsquo;s calendar app is also noticeably more feature-complete at the UI layer, resource scheduling, calendar sharing between users with permission levels, a real web calendar view, than either Radicale or Baikal offer out of the box, since neither ships a client-facing web UI at all, they&rsquo;re pure protocol servers you&rsquo;re expected to view through a separate app.</p>
<h2 id="client-support-the-part-that-actually-determines-your-experience">Client support: the part that actually determines your experience</h2>
<p>All three servers speak the same protocol, so client compatibility is less about the server and more about the client ecosystem:</p>
<ul>
<li><strong>iOS and macOS</strong> have native CalDAV/CardDAV support built into Settings, no app needed. Add an account, point it at your server&rsquo;s URL, done.</li>
<li><strong>Android has no native CalDAV/CardDAV support at all.</strong> This is the single biggest gotcha for anyone coming from Google&rsquo;s ecosystem. You need a sync adapter app, <strong>DAVx5</strong> is the de facto standard, open source, and actively maintained, that handles the sync and exposes the data to Android&rsquo;s native Calendar and Contacts apps.</li>
<li><strong>Thunderbird</strong> needs the Lightning calendar component (bundled by default in modern releases) and handles CardDAV contacts natively.</li>
<li><strong>Desktop Linux</strong> varies by distro and DE, GNOME&rsquo;s Online Accounts and KDE&rsquo;s Akonadi/KOrganizer both support CalDAV/CardDAV directly.</li>
</ul>
<p>Whichever server you pick, test the actual device mix you and anyone else syncing to it actually use before committing. A server that&rsquo;s spec-perfect but paired with a flaky client integration is a worse experience than a slightly less complete server with clients that behave well against it.</p>
<h2 id="migrating-off-google-calendar-and-google-contacts">Migrating off Google Calendar and Google Contacts</h2>
<p>Google will export both as standard files, Calendar as <code>.ics</code> (Settings &gt; Import &amp; Export), Contacts as <code>.vcf</code> or Google CSV (Contacts &gt; Export). The <code>.ics</code> export imports cleanly into any of the three servers above since it&rsquo;s the same format they natively store. The gotchas worth knowing before you start:</p>
<ul>
<li><strong>Recurring events with individual edited exceptions</strong> (a weekly meeting where one occurrence got moved) sometimes import with the exception detached or dropped, depending on the server&rsquo;s RRULE handling. Spot-check a few recurring series after import rather than assuming a clean migration.</li>
<li><strong>Google&rsquo;s CSV contact export is not the same as vCard</strong>, and loses some fields in translation. Export as vCard (<code>.vcf</code>) specifically if Google offers the choice, it&rsquo;s the format all three servers and all CardDAV clients expect natively.</li>
<li><strong>Shared/delegated calendars</strong> (someone else&rsquo;s calendar you have view access to) don&rsquo;t migrate as data at all, since they were never really &ldquo;yours,&rdquo; that access has to be re-established on the new system if it&rsquo;s needed.</li>
</ul>
<p>Keep the Google export files themselves even after a successful migration, they&rsquo;re your rollback path if something&rsquo;s subtly wrong with recurring events weeks later and you need to re-check what the original data actually said.</p>
<h2 id="backup-matters-more-here-than-it-sounds-like">Backup matters more here than it sounds like</h2>
<p>A lost photo collection is a loss. A calendar or contact list that silently goes stale, missed birthdays, a dentist appointment that got dropped during a bad sync, is a loss you might not notice until it costs you something. If you&rsquo;re running Radicale or Baikal, back up their data directory (or database, if you&rsquo;re using one) on the same schedule as anything else you&rsquo;d actually miss, the Restic/Borg pattern covered elsewhere on this site for Docker-based services applies directly here. If you&rsquo;re running this inside Nextcloud, it&rsquo;s already covered by whatever backup you have for that instance, one more reason it&rsquo;s a reasonable default for anyone already on that stack.</p>
<h2 id="picking-one">Picking one</h2>
<p><strong>Just you, or you and a partner, and you want the smallest possible footprint:</strong> Radicale. It does exactly one job, does it well, and gets out of the way.</p>
<p><strong>You want a real admin UI for managing multiple users without touching config files, but still don&rsquo;t want a full groupware suite:</strong> Baikal. This is the sweet spot for most small households and the option worth defaulting to if you&rsquo;re starting from scratch with no existing self-hosted infrastructure to build on.</p>
<p><strong>You&rsquo;re already running Nextcloud:</strong> just turn on Calendar and Contacts there. Don&rsquo;t stand up a second, separate CalDAV/CardDAV server next to infrastructure that already does the job.</p>
<p>Whichever you pick, the real win isn&rsquo;t the server software, it&rsquo;s that CalDAV and CardDAV being open standards means you&rsquo;re never locked into whichever one you choose today. Switching servers later is a data export and a changed URL in DAVx5 or your phone&rsquo;s account settings, not a migration project.</p>
]]></content:encoded></item></channel></rss>