Has anyone experienced problems with Azure Virtual Desktop session hosts after installing the July 2026 cumulative update? Since applying it over the weekend, several user sessions have become stuck, with orphaned processes and no apparent way to terminate the sessions. The affected session hosts can eventually become inaccessible, and removing the host from the host pool and registering it again has been the only reliable recovery method. The environment also uses a relatively recent version of FSLogix, although its event logs have not shown anything useful. I'm considering rolling back the update and would appreciate hearing whether others have seen similar behavior.
2 Answers
We’ve seen somewhat similar behavior on several RemoteApp session hosts this week. Affected VMs first show as unavailable and may report a domain trust problem. Soon afterward, the machine becomes unreachable over RDP, user sessions freeze, and a restart is needed to recover it. It has occurred randomly on three VMs, so the symptoms may be related, though the domain trust error is not present in every case.
A cautious approach would be to roll back the update on one session host and keep that host out of rotation while comparing it with the others. If the deadlocks stop, pause the July update elsewhere and collect the RDS and FSLogix logs from an affected machine before recycling it. That should give Microsoft something concrete to investigate.
That was my thought as well, although the main reason for posting was to find out whether anyone else had seen it. Things were stable over the last couple of days, so it may have been an intermittent issue. We’re continuing to monitor it.
That sounds broadly similar, but we never saw a domain trust error. Task Manager showed several orphaned processes without associated usernames, and the VM became unusable after logging out. Neither Azure controls nor PowerShell could shut it down; removing it from the host pool was the only way to clear the sessions, after which it had to be added back manually. This happened on two VMs, while the last couple of days have been stable. FSLogix logs didn’t reveal much.