When you search in the left column, it highlights matching terms.

If your services have long names, the highlighting won't be visible unless you know to expand the column (and even then the max-width might be too narrow).

Also, filtering reduces cognitive load and speeds up navigation as there is simply less content on the screen. If I'm trying to search through 6 applications, each with a dozen services, it can be hard for the brain to isolate the highlighted entries. This is especially important during emergency situations on a production cluster.

The "Connect to localhost" checkbox should support secure clusters locally as well, as SF will be transitioning to default secure clusters locally.

Two options as I see it (would love opinions on this :-))

1. Don't disable the text field after checking the box, and it's easy to add/remove the 's'
2. Probe https/http with short timeout to make the choice for the user

When opening up SF explorer. It seems like it's only doing a silent sign-on request and not handling if that fails. You have to close the browser to get another auth. prompt.

The summary is that when the partition is undergoing a reconfiguration, the replica role that is returned in our queries (and hence displayed in the SFX UI) is the target role of the replica when the reconfiguration is completed. However, it is important to understand that if the reconfiguration is still in progress, then the replica may not yet have completed the reconfiguration to its target role.

Therefore, when a reconfiguration is in progress, the Role string that we display should be something like "Reconfiguring. TargetRole=Primary" instead of simply "Primary". This would make it clear to the user that the replica role may still be in transition.

I believe there is a bug in SFX.
Repro:

1. Go to /Explorer/index.html#/nodes:
2. Go to a page that is not the first one. In this example, page 3, you can see a node is in Error state: https://imgur.com/a/wmLXg
3. Uncheck OK in Health State filter: https://imgur.com/a/NY6rH
   Expected: the list of nodes in Error state
   Actual: nothing

I think the bug is that the page somehow stays on page 3, which is empty when filtering down to Error state nodes.
If you do the same thing from page 1, it works as expected.

Enable users to easily identify the current version of SFX they are running. This we will need for bug reporting.

Have question buttons that show hints about what a concept / object means in Service Fabric world or points to online documentation where we can read more.

For example, if a page (eg. the cluster essentials page) talks about seed nodes - what are those?
Or unhealthy evaluations - link to the Service Fabric health model that explains them in depth.

cluster:19080/Explorer/index.html#/tab/manifest shows the cluster manifest in xml format. SFX should highlight the xml syntax for easier reading.
Similarly for application manifests.

To enable more space for the metric graphs, I suggest to stack the checkboxes above the metrics labels.

This is current view.

Many SF health reports contain links to http://aka.ms/sfhealth. These do not appear as links, which makes it hard for users to follow them.

An example:

Native image store (fabric:ImageStore) hosts the application package and provisioned application types in the cluster. We need a way to list what content is hosted in the image store. We can use Image store list REST API (https://docs.microsoft.com/en-us/rest/api/servicefabric/sfclient-api-getimagestorecontent) to get the contents and display it. UI can show the top root content, then it should allow users to drill down to look at sub-folders and files at each level.

Currently, the code and config cluster versions are shown on the Node details page. It would be good to have this info more readily available - it should be included in the cluster and the node essentials.

As electron just released the new version 1.8.2, SFX-Standalone should follow and apply the new version of electron.

Service Fabric is a complicated platform and there are tons of concepts to understand. To help users navigate these, I suggest we create a "Learn about {concept}", which can be attached to existing elements, where we find it appropriate. When clicking the "Learn about {concept}" icon, we can pop-out a small descriptive text with a link to documentation for further reading.

E.g.

What will be difficult is to find the right balance where to put these, to not have them being plastered all over the UI. A good rule of thumb would be to only allow this inside the highlighted Essentials pane and top tab navigation.

Why not a question mark? It's already used to indicate unknown or invalid state:

Why not "Learn More"? - https://www.nngroup.com/articles/learn-more-links/

Make service fabric explorer work also on macOS.

It appears that secondary builds using the "gulp Clean-Build", does not actually run build again. I have to manually delete the "gulp.js" output every time, to make it build the Sfx-Standalone project.

Node: v9.3.0
NPM: v5.5.1
TSC: v2.6.2

We need a way for service to fail over to a particular node which might be useful in development and investigating production issues.

I think it would be nice to be able to remember the URLs typed in the past, to avoid keeping them in a file or something. A combo-box (text input + dropdown) would be just fine. At minimum, the last one used would also be useful.

An alternate would be to be able to pass the URL as a parameter to the app, so that I could make shortcuts.

Thanks

Making SFX-Standalone modularized including:

• Make most of the features of SFX-Standalone be modules.
• Every modules can expose different components.
• Every components can be requested via Dependency Injection.
• Have Dependency Injection engine to drive the modularization model.
• Be prepared to support SFX-Standalone plugin feature.

For browser SFX, the browser search can be used to find info in the cluster manifest.
For desktop version of SFX, I didn't find a way to search for string in manifest.

The left hand side tree should have a different icon (or an extra indicator) for seed nodes, so it's easy to detect them at a glance.
Additionally, the search in left side tree can highlight the seed nodes when searching "seed".

I work in a 200+ person development team, with stateful and stateless services running on a service fabric cluster. In our cluster we have a few different node types and > 200 services. The service fabric explorer UX gives a summary for the cluster of the total number of services and replicas. It would be useful to us if we could drill in and have it also show:

1. The total number of stateless instances we have and on which node types.
2. How many stateful services we have on each node type.
3. The total number of stateful service partitions we have per node type.

Recently we wanted this information because we were evaluating service fabric resource governance (https://docs.microsoft.com/en-us/azure/service-fabric/service-fabric-resource-governance),
for services not in containers. We needed to determine if we have enough cores for each service to be allocated a core without running out cores. This is important because the service fabric documentation recommends against having services which are governed and non-governed on the same set of nodes. Additionally, you cannot allocate fewer than 1 core per service. Before we turned on service fabric governance for all services, (granting each process a core) we needed to ensure we had enough cores for all services. Otherwise, per the documentation, we run the risk of not having enough cores, which results in some services not being placed by service fabric.

Additionally, if we end up turning resource governance for all services we will need to monitor this data on an ongoing basis, to ensure that as more services are written, we continue to have enough cores in our cluster to support them. With a very large development team, and a microservices architecture, it is impossible to track each new service that is added - having a global dashboard would make this much easier.

This feature would also be valuable for other similar decisions in the future, as it is very difficult to make data driven decisions without being able to easily collect the correct data.

Thanks!
Alex

In the Nodes view (/Explorer/index.html#/nodes), it would be nice to be able to apply a column filter (checkbox) on Node Type, Upgrade Domain and Fault Domain columns like it is possible on Health State and Status.

Currently depend on an electron version with know vulnerability:

The electron dependency defined in package-lock.json has a known critical severity security vulnerability in version range >= 1.7.0,< 1.7.11 and should be updated.

This is a tracking item for addressing overall accessibility compliance for SFX. Currently there's work being done and testing progress, PR will be opened as they are ready to be integrated into master.

More information on accessibility is available here:

https://www.microsoft.com/en-us/accessibility/default.aspx

• Branch/PR are integrated with Travis CI and AppVeyor.
• Travis CI
  • Linux.
  • macOS.
• AppVeyor
  • Windows.

The reserved CPUCores metric load is using the "ClusterLoad" value, which is an integer and reports a rounded number. There is another value in the API called ""CurrentClusterLoad", which is a double value which reports the non rounded load.

The metrics view should use and report the non-rounded number.

Not sure if other views are effected by the same issue.

Found in 18.3.103 on Windows.

Basically pass the parameters to Service when starting a Chaos - how long this Chaos will run; how long is the interval between Chaos events, and etc.

It would be nice if SFX was sorting columns containing numbers alphanumerically.
For example, versions and node ID.

Semantically, it should be FE_1, FE_2, ..., FE_10, FE_11, not FE_1, FE_10, FE_11, ..., FE_2.

Hi Team,
Does SFx supports onpremise AD Authentication or windows integrated authentication ?

Thanks,
Naresh Khatri

See here: https://github.com/Azure/service-fabric-issues/issues/733

I installed the SFX with great hopes to be able to bundle the 3 fabric explorers I now continuously use in browser tabs.

Is it possible to add support for multiple clusters? With a list on the left, a bit like SQL Server Management? Or different tabs to choose from?

Basically we work with a local cluster, an on premisis cluster and a cloud cluster. So having the possibility to switch between them would be handy! I don't see a lot of reasons to choose the standalone over browsing now.

It would be great to have placement properties displayed in Service Fabric Explorer on Node view (currently this can be found in Cluster -> Manifest but this isn't right I think).

P.S. This was found on Azure Standalone Cluster for Windows Server version 6.1.467.9494

There should be some validation on the endpoint provided for the cluster. For example, if I use port 19000 instead of 19080 in the provided URL, SFX still comes up but just has "No items to display." or "Nothing to display".

• Create package manager module to manage SFX standaone extensions
• Add support of the public myget npm registry for SFX standalone extension packages.
• Add support of customized npm registry for SFX standalone extension packages.

Package manager UX will be done with another task.

When starting SFX without internet connectivity the following javascript error is thrown:

You get past the error and can run SFX.
SFX should show a user friendly error message.

Add ... next to the replicas with option to move the primary/secondary to another node. Useful for mitigation and for testing some scenarios.

We should add a connection timeout which will take the user back to the connection dialog, when the cluster does not respond. Find a good default value for now and enable overwrite later.

This will give a better experience when:

1. The user types a wring connection address or port (or don't specify port).
2. The user looses connection to the cluster.

When taking the user back to the connection dialog, we should persist the connection information which timedout.

Currently, the search has limited utility as it will only highlight what is visible. This means the user must manually expand every application type, then every application instance in order to search through all services.

For my team, that means 12 clicks before the search is useful.

It's very unusual to use double quote-marks in JavaScript/TypeScript source, what is the rational to go with double as oppose to single?

Keep forgetting this and get compile errors ;-)