0.56.1

What's Changed

• Fix host directory policy boundary checks by @fkorotkov-oai in #459. Thank you for the reposrt, @genforAI!

Full Changelog: 0.56.0...0.56.1

0.56.0

What's Changed

• [codex] Preserve worker last_seen on sparse updates by @fkorotkov-oai in #439
• Reuse exec SSH connections per VM by @fkorotkov-oai in #440
• Relicense under FSL-1.1-ALv2 by @fkorotkov-oai in #446
• Migrate CI from Cirrus CI to GitHub Actions by @fkorotkov-oai in #447

Full Changelog: 0.55.0...0.56.0

0.55.0

What's Changed

• Supporting reconnecting to /exec socket by @fkorotkov-oai in #434
• Add more options to /exec by @fkorotkov-oai in #435
• Retry VM exec SSH setup during concurrent bursts by @fkorotkov-oai in #437

New Contributors

• @fkorotkov-oai made their first contribution in #434

Full Changelog: 0.54.0...0.55.0

0.54.0

What's Changed

• Remove license tier validation by @fkorotkov in #428

Full Changelog: 0.53.0...0.54.0

0.53.0

Notable changes

Orchard now supports Linux workers powered by Vetu virtualization

To give it a try, run orchard worker run on a Linux machine and create a new VM with --runtime vetu set.

Make sure that Vetu is installed on that worker, and that the user under which orchard worker run is running has /dev/kvm access.

Vetu supports Tart VM image format and OCI registries, and if you're looking for an image to try out, check out Linux images published by Cirrus Labs.

VM architecture differentiation

Previously, Orchard assumed that all VMs and workers run on arm64. Now it's just the default.

With the introduction of Vetu, one might have an Orchard cluster with different architectures, e.g. arm64 and amd64.

Workers will report their architecture to the Controller automatically, but you need to explicitly specify the architecture when creating non-arm64 VMs, otherwise they won't be scheduled correctly.

Use --arch command-line argument to orchard create vm (or arch field when creating a VM using API) to do that.

VM's operating system is now explicit

Previously, Orchard assumed that a VM's operating system was always darwin (macOS). This worked for Linux VMs too, but not without pitfalls (#382). Now darwin is just a default.

You can now explicitly set a VM's operating system at creation time with --os command-line argument to orchard create vm (or using os field when creating via API).

This is useful to disable the Apple EULA 2 VMs limit check by Orchard, as this limit does not apply to Linux VMs on macOS host.

What's Changed

• Support Vetu virtualization on Linux in addition to Tart on macOS by @edigaryev in #419
• Bump the all-updates group across 1 directory with 13 updates by @dependabot[bot] in #421
• create vm: introduce --{os,arch,runtime} command-line arguments by @edigaryev in #422

Full Changelog: 0.52.0...0.53.0

0.52.0

What's Changed

• test(load): avoid unnecessary new Uint8Array(...) by @edigaryev in #414
• Replace Prometheus with OpenTelemetry by @edigaryev in #413
• Bump the all-updates group with 4 updates by @dependabot[bot] in #415
• Bump the all-updates group with 9 updates by @dependabot[bot] in #416
• README.md: document an alternative "Local Network" permission workaround by @edigaryev in #417
• Fix race in port-forward by @mpbb in #418

New Contributors

• @mpbb made their first contribution in #418

Full Changelog: 0.51.0...0.52.0

0.51.0

What's Changed

• controller(api): new "GET /vms/{name}/exec" WebSocket-based endpoint by @edigaryev in #408
• controller(listVMs): avoid copy of each element when filtering by @edigaryev in #401
• Worker: decouple updateWorker() and syncVMs() to use different triggers by @edigaryev in #403
• Enable Gin's context fallback by @edigaryev in #409
• Instantiate new client for each spawned worker by @edigaryev in #402
• Grafana k6 load script: handle exceptions and provide more checks by @edigaryev in #404

Full Changelog: 0.50.1...0.51.0

0.50.1

What's Changed

• controller(listVMs): reduce allocations by @edigaryev in #400

Full Changelog: 0.50.0...0.50.1

0.50.0

What's Changed

• Bump the all-updates group with 7 updates by @dependabot[bot] in #395
• Fix Grafana k6 script to not consume all system's CPU by @edigaryev in #396
• echoserver: do not treat client errors as fatal by @edigaryev in #398
• Worker: ignore HTTP 404 when updating VM by @edigaryev in #397
• Refactor listing VMs by @fkorotkov in #399

Full Changelog: 0.49.0...0.50.0

0.49.0

Notable changes

Server-side filtering of VMs

GET /vms endpoint now supports filter query parameter, which Orchard Workers uses to limit the VMs returned only to those assigned to it.

This greatly improves the performance for Orchard Controllers that have maintain lots of active VMs objects.

New compute:connect role

The new compute:connect role is a more fine-grained alternative to compute:write and allows to:

• initiate a port-forwarding session to a VM
• query VM's IP address
• SSH to a VM through the Orchard Controller's built-in SSH server (enabled manually via --listen-ssh)

What's Changed

• Support authentication and TLS in k6 load testing script by @edigaryev in #391
• Implement server-side filtering for VMs by worker by @edigaryev in #392
• Introduce "compute:connect" role by @edigaryev in #393

Full Changelog: 0.48.0...0.49.0