Changelog
Notable changes to Cloud Disk, newest first.
Beta
Cloud Disk is in beta and under active development, so behavior and defaults may change between releases.
2026-06-24
- Cloud Disk now caps its own memory use automatically, keeping the daemon's footprint predictable without manual tuning.
2026-06-22
- Kubernetes (CSI): non-root pods can now read and write Cloud Disk volumes.
- Kubernetes (CSI): volume provisioning is now strongly consistent — a newly created volume is immediately usable.
- Discarding/
TRIMnow reclaims space within partially written chunks, not just whole ones.
2026-06-19
- Write-back caching is now the default, so writes are acknowledged from the local cache and flushed to Tigris in the background — noticeably faster for write-heavy workloads. Opt into write-through if you need every write durable on Tigris before it's acknowledged. See Configuration & Tuning.
- Reads are now strongly consistent: data you just wrote is always reflected, even across mounts.
2026-06-17
- Crash-consistent write-back across hosts: after an unclean shutdown you can remount the disk — including on a different machine — and it recovers cleanly, without losing acknowledged writes.
- New
-presetoption to pick a performance profile when creating a disk.
2026-06-15
- New
cloud-disk resizecommand to grow a disk; the filesystem expands on the next mount.
2026-06-12
- Faster cold starts: the cache pre-warms and reads ahead automatically, so the disk is responsive sooner after mounting.
- CLI improvements: mount and unmount show progress, a
statuscommand reports a disk's state, config defaults are clearer, and aversioncommand was added.