We have this Pub/Sub system that you subscribe to via a callback mechanism in C# to recieve events from various things that happen within the database. This subscription has a callback signature attached to it that allows for the Pub / Sub system to callback any subscribers it has and notify them of the change on that Callback thread.
We are taking our windows application and migrating it into a web application. In doing so, I need a way to update this Web Application (The clients) with information from this Pub / Sub. I want to use SignalR, but not sure where to host it. I assume if I host it on the same Web Application as the Client, it won't be able to subscribe to the pubsub due to it not being able to do background threading.
Currently, I have it in a Console application hosting the SignalR server on a specific port. Obviously this is for testing and not ideal for a larger scale.
My question is.. is it really safe to be hosting SignalR outside of IIS? should I put this in a Windows Service? Web Service somehow? Can it go in a Web Application somehow?
Related
I have an architectural question with signalR
Let's say I have a web app (pure front JS) that use a web api and the web app query an API that do a long task and want to be notified when the task is finished.
So the web api create a fire and forget task, and we use SSE with signalR to notify the web app. It's working. Great, thanks to signalR.
But now, I want the long task to be run in another process, let's say with a msmq based system.
So, the web app query the API, the web api create a message, and the msmq service process the message asynchronously.
Can the msmq service hosted in another process (maybe another machine !) notify the web app that the task is finished ? It can be possible to put the connection id in the message, but the service can be able to send the notification ?
I would use a servicebus, you can then use this library to forward the message directly to the clients.
https://github.com/AndersMalmgren/SignalR.EventAggregatorProxy
We're building a real-time Web-based monitoring system for .NET applications (ASP.NET and Windows executable). Those applications can start a long-running operations and statistics are displayed in real-time on Web page.
For ASP.NET ones we found SignalR a perfect solution: Long running operation (even caused by simple WebForms form postback) periodically call JS client-side functions via SignalR RPC to update monitoring page. But we hit 2 caveats:
In ASP.NET we need to monitor several different apps located in several different virtual directories. How do we push data from those different apps onto a single HTML monitoring page?
Another app is a .NET Windows console executable that runs periodically on a schedule. How do we push its run-time statistics to the same monitoring HTML page? One thing comes to mind - have EXE store temporary statistics in a DB and have client pull same data from the DB, but we'd like to avoid polling. Another - periodically at a given intervals the EXE would call the WebApp, passing the data - and WebApp would pass it to client via the same SignalR call. But are there better ways?
One architecture that I've used is a small monitoring collection service, with embedded monitoring clients in every monitored application, Asp.net, Windows desktop app, console app, Windows service, or otherwise.
The collection service is always running. A webapp then connects directly to the service and requests the state of all monitored apps.
Monitored apps run some small embedded client that feeds back application-specific metrics to the monitoring service. The client can either provide data on events or timers, or the monitoring service an ask for it on a timer itself.
With this, we have a unified monitoring architecture - everything that runs just talks to the monitoring service to send updates, and the health viewer clients just ask the service for data using a unified protocol.
It's basically the Application Server pattern applied to monitoring, and takes a couple cues from the design of SNMP.
Very new to SignalR, didn't realize it has multiple clients for different platforms. We will go with SignalR .NET client for all the apps - they will all talk to main SignalR hub directly invoking server-side methods, which in turn update monitoring page.
I am having trouble wrapping my head around whether or not my scenario will require scale out. I have a process in a windows service that pushes messages to a hub hosted in a web app via the signalr .net client. These are user specific messages and are distributed using the Client(connectionid) approach. If this is deployed in a web farm scenario will I need to use a scale out approach? When a user joins I am storing that connection info in the database. I store the url of the webserver and connectionid so I can target that when I publish messages from the windows service.
I would use this if it is an option.
http://www.asp.net/signalr/overview/performance/scaleout-with-windows-azure-service-bus
Louis
I have the following scenario in my project :-
The client makes use of ASP.NET Web API to make HTTP service requests. The Web API sits on top of a couple of WCF services, which in-turn handle all the business logic. The client subscribes to a particular type of event with the Web API. Whenever the Web API receives notifications from the internal WCF services about the occurrence of the event, the Web API in-turn needs to notify (push events to) all the subscribed clients about the events along with their details.
I want to understand the different options which are available for
sending asynchronous callbacks from an ASP.NET Web API to the
clients.(Currently we are working on a prototype for which the
client is a C# Windows Forms application. Later we might opt for
ASP.NET MVC4 web application.).
I also want to know which option would be ideal to send asynchronous
notifications back to the client when the data that accompanies the notification is of large sizes. In our scenario, the notification data that is sent back from the service may be of large sizes (~ in the range of 5KB - 50 MB).
In our scenario which I described above, can SignalR be used for notifying the c# client from Web API, as and when the Web API receives the callback from the internal WCF services?
Note :- The Web API is currently hosted in a Windows Service and the client is a .NET Windows Forms application.
Any pointers to such code samples or directions on how this can be achieved would be extremely helpful.
Cheers
SignalR is a good fit for the scenario you're describing, so I'd suggest using it for the notifications (especially since you want to start with a WinForms application and later switch to browser clients - with SignalR, you'll be able to connect to the same server-side code).
However, I'd also suggest keeping the notication messages lightweight, so instead of sending the data to the client with them, I'd send a token the client can retrieve the data with from WebAPI (SignalR isn't really ideal for large file transfers).
I have 2 roles in my Azure cloud service application: a web role (signalr connections here) and a worker role.
The web role uses Azure service bus as its scaleout provider.
At certain points in time the worker role will emit certain events. I'd like to send this data directly to clients connected to a Hub.
My current implementation involves the worker role placing a message on a service bus queue which the web role subscribes to, the web role then forwards this message to clients via a HubContext call.
My question is: how can I send this message directly to connected clients from the worker role? So far I have considered 3 methods:
Configure signalr as in the web role so that they use the same servicebus topic. - this done not work as intended as worker role instances "steal messages" from topic subscriptions intended for the web role. This would seem to be the cleanest way of doing it but configuration is a problem.
Use the .Net client to send a hub message - this is not ideal as it places unnecessary load on the web role, as well as double the amount of service bus messages when compared to the above method.
Manually write a signalr compatible message to the topic - very hacky and succeptable to breaking changes.
I know that the team are currently rewriting scaleout for the next release but will this be possible at some point?
Edit:
I have noticed that this is supported in the RabbitMq implementation.
It seems an issue with my configuration was responsible for the first method not working.
However, it seems like that method is slower end to end (by about 150 ms) even with one less message in the loop.
I will wait and see if the scaleout work brings any improvements to this method before making any changes.