Plakar v1.1.3 is here · Plakar | The Open Standard for Backup and Restore

Search...

Book a demo<br>Demo

Book a demo<br>Demo

Plakar v1.1.3 is here

Gilles Chehade

2179 words<br>June 16, 2026<br>11 min read

Edit this page on GitHub

You liked this article?

Share it

Join us

Help us

Gilles Chehade<br>CTO @ Plakar Korp, R&D software engineer

Back in January, we shipped v1.1.0-beta and called it &ldquo;the foundation for what&rsquo;s next&rdquo;.<br>A long beta, three release candidates, and several thousand more snapshots later, that foundation is now done.

plakar v1.1.3 is officially out.

You read that right: we jumped straight to v1.1.3.<br>We were so eager to get this one into your hands that we skipped right past v1.1.0, v1.1.1, and v1.1.2, three whole versions, in our hurry to ship.<br>(Alright, the honest version is that we cut a few quick follow-ups on the way out the door and v1.1.3 is the one you actually want. But &ldquo;too eager to release&rdquo; is a much better story, so let&rsquo;s go with that.)<br>More seriously: during this release we moved our integrations to a monorepo workflow, and we are still getting used to it, a few of those extra version bumps are simply us finding our footing with the new process.

It is, without contest, the largest release we have ever published.<br>That is not just a feeling, the numbers back it up.<br>Here is what went in across our core repositories since v1.0.6 (roughly six months of work, from late November to early June):

Repository<br>Commits<br>Pull requests

plakar<br>421<br>250

kloset<br>599<br>274

integrations<br>1,090<br>206

That is well over 2,000 commits and several hundred reviewed pull requests, spread across plakar, Kloset, the gRPC bridge, the SDK, and our 25 integrations.

If you want the full story behind the headline features, the v1.1.0-beta announcement still holds up and goes into much more detail than I will here, this post is the victory lap rather than the deep dive.<br>But since a lot of you discovered the v1.1.0 branch through that beta and then patiently helped us harden it, I want to recap what landed, what changed since the beta, and what comes next.

TL;DR

v1.1.3 is final, stable, and fully backward compatible. Your existing stores keep working as-is; creating a fresh store and syncing into it is optional and only needed to unlock the new-format optimizations (more on that below).

The big features from the beta all made it: new terminal UI , multi-directory backups (single source), rewritten FUSE mounting (plus HTTP mounts), a brand new package manager , and much simpler integration interfaces .

Reliability: the old agent is gone for good. A tiny background service called cached now handles shared cache and locking, while commands run directly in the CLI.

Performance: restores are roughly 95% faster in our Korpus tests, and most of the backup optimizations we held back during the beta have now landed too.

Memory and disk: peak RAM is down across the board, and the default cache footprint is much smaller .

What changed since beta: a lot of stabilization, a move to an integrations monorepo, mount refinements, prune -group-by, safer locking for repair, and better behaviour for cached on Windows.

Thank you to everyone who tested the beta and reported issues. This release is genuinely better because of you.

From beta to release

When we announced the beta, most of the heavy lifting was already behind us, the v1.1.0 branch had been baking through the second half of 2025, and we had already pushed tens of thousands of snapshots through it.<br>So the beta phase was never really about whether the code worked, it was about polishing the edges, catching the things only real users on real data find, and earning the confidence to call it stable.

That is exactly what happened.

We moved from beta to release candidate in spring, cut three RCs , and spent the time in between doing the unglamorous work that makes a release trustworthy:

Stabilization and testing. We pushed test coverage up significantly across both plakar and Kloset, reaching 100% on several packages and lifting the rest substantially. None of this shows up in a feature list, but it is exactly the kind of work that lets us ship the next release faster and with fewer regressions.

An integrations monorepo. We consolidated our many integration modules into a single monorepo, which dramatically simplifies how we cut releases and keep everything in lockstep. If you build from source, this should make your life easier too.

Mount refinements. plakar mount learned to handle direct access to filesystem paths without browsing from the root, and gained an -allow-others flag to pass fuse.AllowOther() at mount time, a frequent request for shared setups.

prune -group-by. Retention policies can now group snapshots when pruning, and the flag is properly documented. Kloset&rsquo;s locate also gained -group-by along with dataset and data-class filters.

Safer maintenance. repair now takes an exclusive lock, the builder no...