- Improved reliability of JEA endpoint creation.
- Improved reliability of blob service during PNU/FRU and CSV move.
- Improved diagnostics for failures in DSC resources.
- Improved error message generated by an unexpected failure in bare metal deployment script.
- Improved resiliency of engine code to ensure typos in machine name casing do not cause unexpected state in the ECE configuration when manual actions are used to add and remove nodes.
- Fixed an issue where some alerts were not paused during update.
- Added resiliency during physical node repair operations.
- Added resiliency to image creation code when the destination path is unexpectedly not present.
- Improved logic around bringing cluster nodes back online in rare cases where outside intervention puts them into an unexpected state.
- Added a health check to detect VM or physical node repair operations that were left in a partially completed state from previous support sessions.
- Added disk cleanup interface for ERCS VMs and ensured that it runs prior to attempting to install new content to those VMs.
- Improved quorum check for Service Fabric node repair in the auto-remediation path.
- Improved diagnostic logging around installation of content from NuGet packages during update orchestration.
- Improved reliability of storage upgrade by eliminating Windows Health Service WMI call timeout.
- Fixed the internal secret rotation failure for customers who use AAD as identity system, and block ERCS outbound internet connectivity.
- Improved HealthAgent update reliability.
- Added a round of health checks prior to the engine update, so that failed admin operations can be allowed to continue running with their original version of orchestration code.
- Fixed ACS backup failure when the ACSSettingsService backup finishes first.
- Improved MonitoringAgent update reliability.
Fixes rolled up from previous hotfix releases
- Improved extraction reliability by addressing idempotent/file locking issues.
- Host update reliability improvements by addressing the image locking issues.
- Improved orchestrator resiliency for the low memory situations on ERCS VMs.
- Fixed "out of memory exception" during internal secret rotation.
- Enhanced self-healing of update and repair operations when hard disk specialization fails unexpectedly.
- Corrected the mechanism used to detect expired or nearly expired external endpoint certificates to only verify the currently bound certificate, instead of all certificates in the certificate store.
- Addressed issues in the Network Controller host agent that can cause Software Defined Network (SDN) instability.
- Improved handling of Virtual Machine extensions and disk provisioning state.
- Addressed an issue that can lead to a memory leak on the Azure Consistent Storage VMs, affecting Object Based Storage performance and stability.
- Increased reliability for Compute Resource Provider updates.
- Improved monitoring and recovery of the RDAgent.
- Addressed additional cases where NonHAVMs may not start after an update.
- Improved reliability of VM disk resize operations.
- Addressed an issue where an incorrect data encryption key is used during key migration.
- Addressed an issue where an update is incorrectly reported as successful when it had actually failed.
- Improved reliability of the DiskRP update process.
- Updated host monitoring configurations to reduce overall system CPU load.
- Corrected an issue where tenant and admin portals fail to load because a resource provider is not responding.
- Improved the admin portal experience to show the download progress of Azure Stack Hub update packages.
- Added a bypass flag for the Privileged End Point command Set-bmccredential to address cases where the customer needs to set the BMC password independently from it being updated in the system configuration store.
- Corrected an issue where the compute resource provider causes elevated CPU load on systems with a large number of VMs deployed.
- Addressed an issue where setting the LicenseType of a VM to the value of None is not recognized as a valid LicenseType.
- Fixed an issue introduced in the 1910 release that impacts future servicing of Azure Stack Hub, which includes affecting the ability to take OEM updates or repair Azure Stack nodes.
- Fixed an issue in which filtering of Marketplace images in the Marketplace Management blade may be inconsistent.
- Fixed an issue in which the volume capacity in the Storage Capacity blade may show as not available or does not update.
- Fixed a performance issue in the Storage Resource Provider that can impact overall control plane actions.
- Fixed a performance issue in the Compute Resource Provider that can impact overall control plane actions.
- Fixed an issue in which a newly deployed VM might use the incorrect protocol in the Windows Azure Agent.
- Fixed an issue in which communications and data transfer to a resource behind an internal load balancer might be significantly impacted and subsequently time out.
- Fixed an issue in which during an OEM or full update, certain infrastructure VMs can't be started after the host is re-imaged.
- Increased reliability for Compute Resource Provider updates.
- Improved monitoring and recovery of the RDAgent.
- Addressed additional cases where NonHAVMs may not start after an update.
- Improved reliability of VM disk resize operations.
- Addressed an issue where an incorrect data encryption key is used during key migration.
- Addressed an issue where an update is incorrectly reported as successful when it had actually failed.
- Improved reliability of the DiskRP update process.