How to handle error in a block of TPL workflow

Question

I have created a pipeline of Transformblocks and finally ending with Actionblock. This pipeline will remain static and once created when the service starts and will be used to push data whenever there is api request from the client.

When one of the blocks internal function throws an unhandled error say Bblock in below code, the propagation completes and aggregated exception is thrown at the end. but the issue is I cannot send any more request in the blocks This is happening in UT. Will it work in real case scenario? I do not want any block in bad state as this can block the message pipeline endlessly which I can see in UT.

How to handle error in a block gracefully so next message is executed without an issue (don't want block in faulted state)?
I am sending cancellationtoken while sending message as this is will be new for every request. Will taskcancelledexception make the block in faulted state?

Code

public static ClassA
{
    
    private BufferBlock<Input> requestQueue;
    private ActionBlock<Input> CBlock;
    void init()
    {
        var options = new DataflowLinkOptions { PropagateCompletion = true };
    
        this.requestQueue.LinkTo(Ablock, options);
        Ablock.LinkTo(Bblock, options);
        Bblock.LinkTo(Cblock, options);
    }

    public async Task<bool> Execute(Input input, CancellationToken cancellationToken)
    {
        try
        {
            await this.requestQueue.SendAsync(input, cancellationToken);
            await this.CBlock.Completion;
        }
        catch (Exception ex)
        {
            if (ex is AggregateException)
            {}
        }
    }
}

Here in the above sample code. Bblock throws error.

If you don't want the block to become faulted then just add a try-catch around the code that may throw an exception, at the point you can decide what to do with the exception without faulting your entire pipeline. — JSteward
– JSteward, Commented Sep 24, 2020 at 22:03
Does this answer your question? Avoid shutting down an entire data flow network when one block is faulted — JSteward
– JSteward, Commented Sep 24, 2020 at 22:05

Harshikesh Kumar · Accepted Answer · 2020-10-09 13:28:35Z

0

Handling any exception using try catch is the only way to keep the pipeline alive and not go into faulted state.

answered Oct 9, 2020 at 13:28

Harshikesh Kumar

3752 silver badges9 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

How to handle error in a block of TPL workflow

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related