We’re aware of issues with Windows OS updated published July 10th that have been causing issues in some Service Fabric clusters. The following symptoms have been noted:
- Application crashes due to failure to open a HTTP or TCP listener
- Processes being unkillable with error message “Access Denied”
- Nodes going down that do not come up automatically
- Upgrades getting stuck – both for application upgrades and for cluster upgrades
- “netsh http show urlacl” hangs on nodes
The Windows team has informed us that updated packages are now available for download via the regular release channels, including Windows Update.
For cluster running on Windows Server 2016 – the fix is a replacement to the July 10th version update, so installing the latest version should mitigate the problem. If your cluster is running the Patch Orchestration Application (POA), the POA may have updated your cluster. Please confirm that the OS is not running the impact KB (Knowledge Base ID) as shown below.
For clusters running on Windows Server 2012R2 – the fix is an additional upgrade on top of the version released in July. This means that if the nodes were not updated since before July 10th, you may need to apply the July 10th update first and then execute the Windows update again to resolve the problem.
Here are the corresponding impacted KBs for each OS in your cluster:
| Operating System | Impacted update |
| Windows Server 2016 | KB 4338814 |
| Windows Server 2012R2 | KB 4338824 |
| KB 4338815 |
You can check if your nodes have received these updates by opening Command Prompt (may require admin privileges) and running “systeminfo.exe | findstr KB4338814” (replace the KB value with the appropriate number if you are on Server 2012).
If your cluster has nodes that have received these updates, please update them with the latest bits that have been published. For those of you with very large clusters where upgrading each node is not an option and you require assistance, please reach out to Azure Support.