wasmCloud

v2.10.3Orchestration & Management

Oct 2, 2026

A maintenance release with dependency updates and correctness fixes for workload and plugin state cleanup. It includes fixes for state that outlived unbind, failed-stop cleanup, plugin unbind handling, and storage recreation, with no announced operator configuration changes.

All 9 other recorded changesfixes 8 · value changes 1

fixes (8)

* fix(wash-runtime): release per-workload state that outlived unbind by @ricochet in https://github.com/wasmCloud/wasmCloud/pull/5634
* fix(runtime-operator): stop orphaning workloads around failed stops and departing hosts by @ricochet in https://github.com/wasmCloud/wasmCloud/pull/5632
- **postgres never released anything at unbind.** Unbind removed entries whose component id *started with* the workload id. Component ids are fresh UUIDv7s, so nothing ever matched. - The existing test used made-up ids like workload-a-component-0, which hid the bug; it now uses real UUIDs.
- **wasi:logging had no on_workload_unbind.**
- **A plugin failure during start could free the workload id too early.**
- **unbind_all_plugins stopped at the first HTTP unbind error.**
- **Egress state outlived workloads that serve no HTTP.** The HTTP handler heard about a stop only for components exporting wasi:http, and the service unbind never reached the egress transport. A messaging handler or service that *sent* HTTP kept its pooled connections, TLS session store and connection permits until idle expiry (60s). The handler is now notified once for every stopping workload.
- **In-memory keyvalue and blobstore recreated storage after unbind.**

value changes (1)

* build(deps): Bump the all-go group across 2 directories with 10 updates by @dependabot[bot] in https://github.com/wasmCloud/wasmCloud/pull/5635

Add wasmCloud to your stack

A weekly email arrives when a release needs action. Weeks with nothing to act on, like this one, send no email.

Add to stack →

Official release note ↗All wasmCloud releases →