Azure Stack Hub hotfix 1.2102.30.148

Article
09/21/2022

Summary

Optimized reading of disk IOPS values to support VMs with a large number of data disks.
CRP will self-heal a VM with a SCSI disk that failed to attach instead of requiring operator removal of the disk from the VM.
Fixed an issue in VM power-off operation in which CRP service always ignored the value of the non-graceful VM shutdown parameter.
Fixed an issue that prevented health remediation of the Compute Host Agent service.

Fixes rolled up from previous hotfix releases

Fixed a bug in incremental disk snapshots in which a failed snapshot can block any future snapshot operation.
Fixed a mismatched cluster name issue in FRP.
Fixed an issue that could impact updating to 2108.
Resolved an issue in which the Create a Virtual Machine image dropdown displays image options that are not available or were not downloaded to the stamp.
Fixed an issue in which the PrivateWorkingSet queried value overflows if larger than 4 GB.
Removed some unnecessary detailed error information on the administrator portal.
Improvements to support tools.
Resolved an issue with some Key Vault applications being in an unhealthy state while updating from 2102 to later builds.
Add retry for get-volume requests in Test-AzureStack infra capacity check.
Shorten the "break-glass" tokens and make them human-readable.
Fixed a bug related to physical disk health when repairing a node with SED drives.
Improved Datapruner processing to minimize actor timeout alerts.
Enabled rotation of health agent SSL certificate as part of internal secret rotation.
Added graphs to Storage area that show volume performance.
Improved logic for incremental snapshot creation and deletion.
Improved resiliency in PEP startup script.
Fixed an issue in which System Center Operations Manager (SCOM) was unable to close operator portal alerts.
Improved auto-remediation workflow for memory utilization.
Improved incremental snapshot creation and deletion.
Addressed an issue in the Reset-CloudAdminPassword cmdlet.
Updated Network Controller to fix bugs in PA VIP allocation and IP-MAC leaks.
Improved performance when querying for usage data.
Improved resiliency of compute admin operations.
Removed overly verbose logging from Software Load Balancer VMs.
Fixed a memory leak issue and improved memory efficiency for monitoring services.
Fixed an issue in which the backup blade showed a "rainy" page when backup share was inaccessible.
Fixed an issue that prevented transcript collection when closing Privileged Endpoint (PEP) session.
Fixed a multi-tenancy website security issue.
Improved reliability of update from 2008 to 2102.
Fix for adding availability set to SQL VM in the SQL VM creation process.
Fix for setting storage size in SQL VM configuration to more than 1000 GB.
Fixed group-based authorization errors for users that require group expansion.
Newer API version support for Key Vault resource provider. The Key Vault control plane supports the 2019-09-01 API version, and the Key Vault data plane supports API version 7.1.
Added banner to warn users when a certificate will expire soon.
Fix to show accurate status of node in portal if a repair operation on it has failed.
ETW trace sessions are configured to use 16 MB of non-pagedpool memory. Trace sessions now report lost event count (if any) per session, during ETL file rotation in TraceCollector Eventlog. This releases critical non-paged pool memory for other services on hosts and VMs. Also fixed various bugs in Tracecollector.
Fixed bugs that increased memory pressure on infrastructure.
Fixed bugs in AD FS and Azure Stack Graph to improve deployment and upgrade reliability.
Fixed bugs in SRP and DiskRP in which performance counters were missing in the WAC client.
Fixed a bug in resource provider (for example, Event Hubs) deployment, update, or secret rotation. The operation previously failed with no apparent failure cause. The fix allows the operation to complete successfully.
Fixed a bug that incorrectly raised a PnPDevice.Attached alert for a set of devices (mouse, keyboard, etc.) that were safe.
Set memory alert to preview.
Addressed an issue with host agent monitors.
Fixed a trust relationship issue with container applications in FabricRing.
Improved RdAgent availability by removing empty RdAgent files.
Improved reliability of full update.
Extended the update readiness checks to cover more Service Fabric health and VM health checks; for example, memory usage and storage disk capacity checks.
Updated memory configuration for VM sizes Standard_NC16as_T4_v3 and Standard_NC64as_T4_v3.
Removed legacy SRP SQL instances and DB files to free up stamp resources.
Fixed a bug in which the cluster status can be stuck in "Configuring Storage" after adding a new node.
Fixed health package registration, removing duplicate artifact creation.
Fixed a bug that sometimes caused health blades in the operator portal to become unavailable.
Fixed a bug in which BCDR runner logs filled up MASLogs folders on physical hosts.
Added monitoring of WMIProvider health and remediation.
Added Network Controller IMOS size check to Test-AzureStack.
Fixed a bug in SSL certificate validation during internal secret rotation.
Fixed process watchdog memory leaks.
Updated Defender platform to version 4.18.2103.7.
Enabled the alert module for customers depending on Syslog for alerts. The services will continue to emit alerts to the Syslog pipeline.

Hotfix information

To apply this hotfix, you must have version 1.2102.28.82 or later.

Important

As outlined in the release notes for the 2102 update, make sure that you refer to the update activity checklist on running Test-AzureStack (with specified parameters), and resolve any operational issues that are found, including all warnings and failures. Also, review active alerts and resolve any that require action.