Service Fabric

Do you have an idea, suggestion, or feedback based on your experience with Azure Service Fabric? We would love to hear from you.

Please take a few minutes to submit your feedback in one of the forums available on the right or vote up an idea submitted by another Azure Service Fabric customer. All the feedback shared in these forums are being monitored and reviewed by the Microsoft Azure Service Fabric team.

Please remember this forum is for Azure Service Fabric, and if you have feedback on other Microsoft products, services or platforms, please submit your feedback in the appropriate forum (found on the right hand side).

Thanks for your feedback – we look forward to improving your satisfaction with Azure Service Fabric

  1. Reliable pub-sub framework

    Pub-sub is a common pattern in cloud applications. It would be great to have a reliable implementation of it out of the box.

    1,067 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    under review  ·  17 comments  ·  Flag idea as inappropriate…  ·  Admin →
  2. Provide a possibility to shut down a SF cluster gracefullly

    Provide a possibility to shut down a SF cluster gracefully.

    Currently we only have the option to go to every VM and shut it down individually.

    It would be nice to have an option on the SF cluster itself to shut the whole cluster at once gracefully.
    All apps & data would be retained, but we were able to put a cluster into "cold storage" and awake it later with a single click instead of go to every VM and stop or restart them.

    84 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  3. Support Reactive Extension (RX) 3 API

    Support of RX 3 API
    https://github.com/Reactive-Extensions/Rx.NET
    can reduce service communication when subscribing to remote query

    Actor should be able to subscribe for changes made by other Actors
    (like the case with Orleans)
    Support Rx API will dramatically improve Complex Event Processing
    on multi Actors fabric.

    416 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    3 comments  ·  Flag idea as inappropriate…  ·  Admin →
  4. Simplify Configuration

    Currently to make a single configuration setting use by code that differs between environments I have to:
    A) Add the setting to a configuration section in the Settings.xml (mark as "MustOverride", but you still have to provide a value).

    B) Add a (exactly matching) setting to the configuration override section of the application manifest.

    C) Add a application parameter (with a application global name) to the application manifest. This needs a default value that probably matches the one you already provided in the previous step.

    D) Add the parameter to every single application parameters file that matches the parameter name.…

    140 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    under review  ·  1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  5. instead of having to use a vm to host service fabric services, have it as a service

    i don't want to have a vm anywhere in the mix, just have a service where we can publish our service fabric services to.

    thnx a lot

    52 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  6. Improve deployment performance

    Overall, this framework is amazing, and I love it. However, the one drawback is the turnaround time for slight tweaks. I'd love to have a debug-only hotswap which deployed much faster. Maybe something like: pause the cluster failover monitoring, halt all of the running services for an application, deploy code changes, start them again, and resume failover monitoring.

    122 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    2 comments  ·  Flag idea as inappropriate…  ·  Admin →
  7. SDK support for services implemented in c++.

    For high performant backend services (eg. quant + fintech), can the SDK support c++ services?

    28 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  8. Add process id/thread id in details of partition, replica view in explorer

    Add process id/thread id in details of partition, replica view in explorer so that it helps in debugging and also allow taking dump

    17 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  9. Publish A Single Service From Visual Studio

    A feature allowing you to publish a Single Service in a solution (providing the same replace, or upgrade preserve data) would be a huge time saver for development! Something similar to Debug > Start New Instance that will deploy the latest code to the specified cluster and attach the debugger (or not attach if just publishing). The rest of the services are likely already running and I want them to continue to run and test the new changes I just made in the service I made those changes in.

    F5 or Rt Click > Publish deploys all of the apps…

    74 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    6 comments  ·  Flag idea as inappropriate…  ·  Admin →
  10. Add chaos testing capabilities to the Service Fabric Explorer

    Having the possibility to start basic integrated chaos testing from within the Service Fabric Explorer without having to write any lines of code.

    Feasible ideas would be to shutdown random nodes to see how your system reacts to it.

    21 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  11. Unit test Actors

    I would like to be able to create unit tests for the Actors without spinning up a cluster. Something like:

    [Setup]
    public void Setup(){
    ActorTestProxy.Setup<IMyActor>(new MyActorImplementation);
    ActorTestProxy.Setup<IMyActorDependency>(new MyActorDependencyMock);
    }

    [Test]
    public async void ActorTest()[
    IMyActor actor = ActorTestProxy.Create<IMyActor>(0, "test");
    var result = await actor.Foo("hep");
    Assert.NotNull(result);
    }

    139 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    3 comments  ·  Flag idea as inappropriate…  ·  Admin →
  12. Capability to explore actor instances

    The Service Fabric Explorer shows you what services & actor types are deployed on the cluster but when developing Actors it is sometimes hard to determine if there are instances for a specific Actor or how many are currently running.

    In some scenarios it would be nice to explore the cluster and see all the instances that are present or even have a look at their state.
    This would allow me to see if I successfully instantiated an instance of an Actor instead of building functionality in my system for this and check with the ActorProxy.

    Or am I missing…

    56 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    3 comments  ·  Flag idea as inappropriate…  ·  Admin →
  13. Restart service/partition/replica action in explorer

    Similar to delete service action, add restart action to service or partition or replica as appropriate

    12 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  14. Be able to add node types after deployment

    When wanting to add new nodetypes to a cluster after deployment either updating the arm template or making the put calls to the provider with new node types the error

    {
    "error": {

    &quot;code&quot;: &quot;ManualScaleUpOrDownNotAllowed&quot;,
    
    &quot;message&quot;: &quot;VMInstance count cannot be modified.&quot;

    }
    }

    is returned. The count is not changed for any of the existing nodes,so the error is also a bit misleading.

    21 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  15. FabActUtil.exe needs to give failure messages

    By far the single most frustrating part of Service Fabric development is FabActUtil.exe returning -1 for every single solitary error at compile time. It gives no indication whatsoever of what went wrong.

    We have burned extensive hours (and even examined the EXE's code with Reflector) just to try to get the slightest hint of what has gone wrong.

    Sometimes it works locally but fails on the build server. Sometimes it works on one dev's machine and not another. Sometimes it fails on compile and then when immediately recompiling it works.

    Often, we figure out what is causing the failure and…

    15 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    3 comments  ·  Flag idea as inappropriate…  ·  Admin →
  16. Custom actor state serialization

    Like the custom serialization providers available in the Reliable Services, allow developers to control the serialization of actor state. For instance, encrypting the state while at rest or using alternate forms of serialization besides DataContract serializable.

    27 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  17. Actor model communications must implement reliable messaging

    At the moment when using ActorModel you may get duplicate messages sent from the client when client gets communication errors or server failure. That in fact means that in the every actor you have to have a logic that will need to understand if message was processed already ot not. This is a standard problem and must be handled by Fabric insfastructure by implementing "reliable messaging". So the framework should detect if it is a duplicated message that already been delivered & processed or not.

    388 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    under review  ·  2 comments  ·  Flag idea as inappropriate…  ·  Admin →
  18. Rebalance partitions for stateful services and actors when adding nodes

    Currently, you cannot change the partitioning of a stateful service/actor. I'm forced to anticipate future horizontal scale and create that many partitions, so that when nodes are added to the cluster, load is moved off of hot nodes to the new ones.

    NoSQL databases, such as MongoDB, rebalance the partitions when adding a node. If I"m doing a range partition and have 5 nodes, the range is sliced up into 5 equal(ish) shards. If I add node later, data is migrated until I have 6 equal(ish) shards.

    Having this functionality in Service Fabric for stateful services/actors would allow better horizontal…

    19 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  19. ServiceFabric OData

    Allow access to ServiceFabric metrics & configuration data via an odata API

    14 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  20. Provide tools for inspecting reliable collections

    If I'm going to move some of my core application state from an external state store to reliable collections, I would like to have equivalent tools for inspecting that state apart from the VS debug window.

    272 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Microsoft
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    planned  ·  12 comments  ·  Flag idea as inappropriate…  ·  Admin →
  • Don't see your idea?

Service Fabric

Categories

Feedback and Knowledge Base