Release notes | Ballast
Skip to main content
Version: 1.0

Release notes

What changed in each release, newest first. This page covers the whole 1.0 line, so it stays here as new 1.0.x releases are added.

1.0.2, 1 October 2026​

Before you upgrade​

  • Update in this order: the centre, then the agent file, then the hosts. Upload the new centre and the new ballast-agent.exe in Ballast Manager (Updates), then use Settings, Agents, Update all outdated in the console. Updating Ballast walks through each step.
  • The console opens in a new layout. Every browser switches to it on its next load. The previous layout is still available as Classic under Settings, Profile, Appearance. If you run the console as an installed app, close its window and open it again after the centre is updated, or it keeps showing the build it last loaded.
  • Check for hosts on the development stub. An agent installed by hand without -hyperv powershell ran a stub that reported two fake adapters (NIC1 and NIC2) and no switches, storage or VMs, while the host showed online. If any host looks like that, update its agent from the console: that reinstalls it on the real backend.

Console​

  • Editorial is the default layout. Square panels one shade lighter than the page, across the top bar, object header, tabs, panels and tables. The inventory navigator is unchanged in either layout.
  • New virtual machine has the settings Edit VM already had: computer name, dynamic memory, nested virtualisation, TPM, console resolution, and an Integration services tab. The New virtual machine, New host and New cluster dialogs are now the same width as the console's other setup dialogs.
  • A host's Summary shows the same virtual machine table as its Virtual machines tab, with the filter and multi-select power actions.

Agent​

  • A shorter reconcile pass. The agent runs a pass's per-VM PowerShell work in one PowerShell process instead of starting one per VM, only writes a VM setting when it differs from what the host already has, and reads network adapters, VLANs, guest IP addresses and integration services once per host instead of once per VM. On a test host with 14 VMs a full pass went from about 86 seconds to about 61. Every VM is still checked on every pass.
  • No false alarms during a planned shutdown. VMs on a host that Ballast is shutting down or rebooting no longer report Degraded while it happens.
  • The agent manages the Hyper-V host by default. -hyperv powershell is now the default on Windows. The stub is only used when asked for with -hyperv stub, and says so in the agent log.

Fixes​

  • Adopting a VM no longer declares a network adapter Ballast cannot apply. An adapter not connected to a virtual switch is left undeclared, where it used to fail every pass with a raw PowerShell error. A connected adapter is matched to a distributed port for its switch and VLAN, reusing one that exists or creating one named for the VLAN.
  • Updating an agent from the console works on an agent installed by hand in a different folder. It used to fail with "service already exists" and leave the agent stopped.
  • The fabric summary's storage bar read 100% used for a standalone host's volumes, whatever their actual usage.
  • A host's Storage tab showed each drive-letter volume twice in the Volume column (D: with D:\ under it).
  • In a host's Summary, the virtual machine table's name column lost its row lines until a row was hovered.

1.0.1, 24 September 2026​

Before you upgrade​

  • Nothing changes until you choose it. Active Directory sign-in is off until you configure it, and your existing local accounts keep working exactly as they do now.
  • Update the centre first, then the agents. Upload the new centre in Ballast Manager (Updates), then use Settings, Agents, Update all outdated. The console marks any host still on the 1.0.0 agent as outdated until you do. The new volume actions below need the 1.0.1 agent.
  • Some Settings pages moved. If you go looking for one of these and it is not where you left it, see Settings below.

Active Directory sign-in​

Operators can sign in to the console with their Active Directory account, with your groups mapped to the Administrator, Operator and Read-only roles.

  • Sign in as user@domain or DOMAIN\user. Nested groups count, and a person in several mapped groups gets the most privileged role.
  • A person in none of the mapped groups cannot sign in, and is told why.
  • Local accounts are tried first and remain the break-glass login. A domain controller being down never locks anyone out, and directory sign-in cannot be turned on unless a local administrator exists.
  • A Test button checks the bind account and each mapped group, and can try a real sign-in, without saving anything.
  • Signed-in directory users are re-checked every 15 minutes. Removing someone from the mapped groups, or disabling or deleting their account, signs them out. If the directory cannot be reached, sessions already open are kept. A person refused at sign-in because they are no longer in any mapped group is signed out at once.
  • It is included in every tier, including the free Community tier. Agents are not affected: this is the console's sign-in only.

Start with Active Directory sign-in. If your domain controllers do not offer LDAPS yet, see LDAPS on a domain controller.

Settings​

  • Alarms is now its own page. The webhook and email settings used to sit under Authentication only because that was the first settings page built. Authentication now holds the inactivity timeout, Active Directory and the operator accounts, and each page saves only its own settings.
  • Ballast Manager has its own section, next to Backup, that says what Manager is for and links to it. Backup keeps a single line and the same link.
  • The agent service account provisioning (the script, the example command and the "already set up" tick) moved from Domain & DNS to Agents. The script and its download are unchanged. Domain & DNS is now just the DNS servers.
  • Fixed the breadcrumb on Settings pages. Every page other than Credentials, Authentication and ISO Library was headed "Settings > Credentials".

See the Settings reference.

Recovery runbooks​

  • A runbook can be deleted from the console: right-click it in the library list or the tree, or use its menu, then Delete runbook and type its name to confirm. Deleting one also clears its run history.
  • Right-click Recovery runbooks in the tree for New runbook, matching the button on the library page.
  • The Run history and Recovery runbooks tables no longer overflow their card when a run's message is long.

Standalone host disks and volumes​

  • A disk or volume with nothing on it (per Ballast's own VM inventory) can be emptied by right-click, on the disk row or its volume. Format volume reformats in place and keeps the drive letter. Delete volume wipes the disk back to raw and poolable. Both are refused on a cluster member and on the OS volume.
  • A data volume carved from spare space on the OS's own disk (D: or I: on hosts with that layout) can be formatted. The agent now reports which physical disk backs each volume and whether that disk also carries the boot or system partition, and Format checks the volume's own partition rather than the whole disk.

1.0.0, 20 September 2026​

Public launch. The centre, Ballast Manager and the Windows agent are on the Downloads page, and the agent's source is public on GitHub.