Hi Emran Moradian •
From what you’ve described, this looks less like a misconfiguration and more like a regression tied to the June 2026 cumulative updates on Windows Server 2025 build 26100. The fact that both guests fail identically, with vmicvss timing out at PID 0 despite healthy VMBus and VSS writers, strongly suggests the integration component itself is broken at the OS build level rather than anything you can fix inside the VM.
At the moment, Microsoft hasn’t published a formal hotfix or KB that directly addresses this vmicvss startup failure, but there are two practical workarounds: (1) roll back the June CUs to confirm the regression disappears, or (2) switch to standard checkpoints or crash‑consistent backups until the next servicing release. If you’re running production workloads, I’d also recommend opening a support case with Microsoft CSS so they can track this as a potential servicing bug that helps push it into the queue for an out‑of‑band fix. Rebuilding the guests won’t help since the issue is tied to the integration service binaries shipped in build 26100. For now, your mitigation of crash‑consistent backups is the safest path until a patch lands