Azure Service Fabric 7.1 Release
Published Apr 20 2020 04:10 PM 11.1K Views
Bronze Contributor

Azure Service Fabric 7.1 is now available!

Due to the current COVID-19 crisis, and taking into consideration the challenges faced by our customers, we are making 7.1 available through the manual upgrade path, but clusters set to automatic upgrade will not automatically receive the upgrade at this time. Automatic upgrades will be paused until further notice to ensure that customers can apply upgrades when most appropriate for them, to avoid unexpected disruptions.

You will be able to update to 7.1 via through the Azure Portal or via an Azure Resource Manager deployment.

      Service Fabric clusters with automatic upgrades enabled will begin to receive the 7.1 update automatically once we resume the standard rollout procedure. We will provide another announcement before the standard rollout begins on the Service Fabric Tech Community Site.

      We also have published updates to end of support date for major releases starting from 6.5 till 7.1 here. 

 

What is new in Service Fabric 7.1?

We are excited to announce the next release of Service Fabric. This release is loaded with key features and improvements. Some of the key features are highlighted below:

Key Announcements

    • ClusterObserver (CO) 1.1 -runs on one node, captures cluster level health telemetry. In this release, ClusterObserver also monitors node status and emits telemetry when node is down/disabling/disabled for longer than user-specified time period.
    • FabricObserver (FO) 2.0 - runs on all nodes, generates health events, emits telemetry when user configured resource usage thresholds are reached. This release contains several enhancements across monitoring, data management, health event details,structured telemetry.

Improve Application life cycle experience

  • Preview: Request drain: During planned service maintenance, such as service upgrades or node deactivation, you would like to allow the services to gracefully drain connections. This feature adds an instance close delay duration in the service configuration. During planned operations, SF will remove the Service’s address from discovery and then wait this duration before shutting down the service.
  • Automatic Subcluster Detection and Balancing: Subclustering happens when services with different placement constraints have a common load metrics . If the load on the different sets of nodes differs significantly, the Service Fabric Cluster Resource Manager believes that the cluster is imbalanced, even when it has the best possible balance because of the placement constraints. As a result, it attempts to rebalance the cluster, potentially causing unnecessary service movements (since the “imbalance” cannot be substantially improved). Starting with this release, the Cluster Resource Manager will now attempt to automatically detect these sorts of configurations and understand when the imbalance can be fixed through movement, and when instead it should leave things alone since no substantial improvement can be made.  
  • Different Move cost for secondary replicas:  We have introduced new move cost value Very High that provides additional flexibility in some scenarios to define if a separate move cost should be used for secondary replicas.
  • Support for Health Probes for containerized services: Support for Liveness Probe mechanism for containerized applications. Liveness Probe help announce the liveness of the containerized application and when they do not respond in a timely fashion, it will result in a restart.
  • Run till completion/once for services

   Image Store improvements

  • Service Fabric 7.1 uses custom transport to secure file transfer between nodes by default. The dependency on SMB file share is removed from the version 7.1. The secured SMB file shares are still existing on nodes that contains Image Store Service replica for customer's choice to opt out from default and for upgrade and downgrade to old version.

Reliable Collections Improvements

  •  In memory only store support for stateful services using Reliable Collections
  • Preview: Service Fabric Backup Explorer: To ease management of Reliable Collections backups for Service Fabric Stateful applications, Service Fabric Backup Explorer enables users to
    • Audit and review the contents of the Reliable Collections,
    • Update current state to a consistent view
    • Create Backup of the current snapshot of the Reliable Collections
    • Fix data corruption

In addition, this release contains other new features, bug fixes, and supportability, reliability, and performance improvements. For the full list of changes, please refer to the release notes.

 

Breaking Changes

  • For customers using service fabric managed identities, please switch to new environment variables ‘IDENTITY_ENDPOINT’ and ‘IDENTITY_HEADER’. The prior environment variables'MSI_ENDPOINT' and 'MSI_SECRET' are now removed.
  • Service Fabric Managed Identity endpoint is now secure (HTTPS).
  • Currently Service Fabric ships the following nuget packages as a part of our ASP.Net Integration and support:
  • Microsoft.ServiceFabric. AspNetCore.Abstractions
  • Microsoft.ServiceFabric. AspNetCore.Configuration
  • Microsoft.ServiceFabric. AspNetCore.Kestrel
  • Microsoft.ServiceFabric. AspNetCore.HttpSys
  • Microsoft.ServiceFabric. AspNetCore.WebListener

These packages are built against AspNetCore 1.0.0 binaries which have gone out of support (https://dotnet.microsoft.com/platform/support/policy/dotnet-core). Starting in Service Fabric 8.0 we will start building Service Fabric AspNetCore integration against AspNetCore 2.1 and for netstandard 2.0. As a result, there will be the following changes:

  1. The following binaries and their nuget packages will be released for netstandard 2.0 only. These packages can be used in applications targeting .net framework <4.6.1 and .net core >=2.0
    1. Microsoft.ServiceFabric. AspNetCore.Abstractions
    2. Microsoft.ServiceFabric. AspNetCore.Configuration
    3. Microsoft.ServiceFabric. AspNetCore.Kestrel
    4. Microsoft.ServiceFabric. AspNetCore.HttpSys
  2. The following package will no longer be shipped:
    1. Microsoft.ServiceFabric. AspNetCore.WebListener:
      1.       Use Microsoft.ServiceFabric. AspNetCore.HttpSys instead.

Upcoming Features and Improvements

We want to let you know about the work that we are doing beyond the initial 7.1 release!  Work on these items is still ongoing. At this point we plan for these work items to GA in upcoming 7.1CUs, or as separate packages available in the upcoming months. The specific delivery dates will be communicated as they become closer to GA.

  • Cross Availability Zone Clusters – Support for using a single Cross-AZ Virtual Machine Scale Set. Today’s workloads require one scale set per AZ. 
  • Support for stateless node type and low priority/Azure Spot Virtual Machine Scale Sets in Service Fabric clusters
  • Support for “Multiple stateless service instances on the same node”  allows more than one instance of a stateless service to be placed on a single node.
  • Increased information about historical cluster and application events in Service Fabric Explorer
  • Provide customers with diagnostics and insights through Azure Portal
  • FabricHealer(FH)stateless Service Fabric service that will detect and then initiate repairs after inspecting unhealthy evaluations for well-known error codes emitted by FabricObserver (FO) instances running in the same cluster. 
  • One box experience for Service Fabric Managed Identities
  • Portal support for Ubuntu 1804

Improve Application life cycle experience

  • GA of VMSS Ephemeral OS disk support
  • GA of Request drain
  • Validation of downloaded image’s integrity to resolve corrupted image issues when hosting downloads images from Image Store to activate a service's replica.
  • Enable cleanup of local files when an application is no longer running on a node.
  • Support Readiness probe-based routing to replicas

Service Fabric Explorer

       Service Fabric Explorer Modernization with new observability features to improve cluster and application lifecycle management. New features include

  • Repair jobs: Displays a list of currently pending and completed repair tasks, along with a scrollable timeline. A “REPAIR JOBS” tab will appear for clusters that have the repairManager service enabled.  
  • Replicator status: Replicator status to provide more insight into the status of replicas that are in build for stateful partitions, tiles showing progress have been added. This data was previously much harder to find and could be hard to read, this view is intended to display relevant information for that part of the in-build process.  
  •  Improved Unhealthy Evaluations view: Unhealthy evaluations will provide a tree-based view in addition to the current list based one.

 

 

5 Comments
Version history
Last update:
‎Apr 22 2020 01:29 PM
Updated by: