Using Task Parallel Library in API

Question

I have a method where I would like to run three queries in parallel. I am trying to use Task Parallel library to do so as shown:

        internal static async Task<RequestAppointmentDBMasterObject> GetDAMasterObject(string facility, string sites, string state)
        {
            
            RequestAppointmentDBMasterObject mo = new RequestAppointmentDBMasterObject();

            string querySites = query1;
            string queryPractitioners = query2;
            string queryExistingAppts = query3;

        using (IDbConnection db = new OdbcConnection(connectionString))
        {
            await Task.Run(() =>
            {
                Parallel.Invoke(
                    () => { mo.SiteList = db.Query<Sites>(querySites).ToList(); },
                    () => { mo.PractitionerList = db.Query<Practitioners>(queryPractitioners).ToList(); },
                    () => { mo.ExistingAppointments = db.Query<AvailResponseObject>(queryExistingAppts).ToList(); }
                );
            });
        }

            return mo;
        }

Then, on another class I am calling this method to set the object I am going to work with as shown below:

        RequestAppointmentDBMasterObject mo = DataAccess.GetDAMasterObject(facility, site, state).Result;

The problem is that I never get a result back when I call the API. Once I change everything back to not use the Task and Parallel libs. everything works fine. I also tried this same code in a Console App and I got data back. I am not sure why I might be missing on the code so I can use this on my API. Thank you in advance.

This code contains several bugs. Parallel.Invoke doesn't help at all. Dapper has QuerryAsync so there's no reason to call Task.Run to call Query. Worse, Parallel.Invoke` is modifying shared state — Panagiotis Kanavos
– Panagiotis Kanavos, Commented Mar 26, 2021 at 14:00
@PanagiotisKanavos, thanks for your feedback. I just fixed how Parllel.Invoke is getting called. Can you post an example on how to do this with QueryAsync — Yoismel
– Yoismel, Commented Mar 26, 2021 at 14:08
It doesn't matter how you call Parallel.Invoke. It's the wrong thing to use in the first place. It's wasting threads and prevents you from using QueryAsync — Panagiotis Kanavos
– Panagiotis Kanavos, Commented Mar 26, 2021 at 14:12
@PanagiotisKanavos, will it be valid to say this will then execute in parallel? mo.SiteList = db.QueryAsync<Sites>(querySites).Result.ToList(); mo.PractitionerList = db.QueryAsync<Practitioners>(queryPractitioners).Result.ToList(); mo.ExistingAppointments = db.QueryAsync<AvailResponseObject>(queryExistingAppts).Result.ToList(); — Yoismel
– Yoismel, Commented Mar 26, 2021 at 14:26
No. .Result blocks. You must use await to await completion without blocking. Richard's answer shows this — Panagiotis Kanavos
– Panagiotis Kanavos, Commented Mar 26, 2021 at 14:34

Richard · Accepted Answer · 2021-03-26 16:51:37Z

4

This is massively more complex than needed. Having layers of tasks wrapper others is not generally the right approach.

You would be better performing the three queries as async operations (ie. they return Task<T>, and then awaiting them in parallel:

var t1 = db.QueryAsync(...);
var t2 = db.QueryAsync(...);
var t3 = db.QueryAsync(...);

await Task.WhenAll(t1, t2, t3);

This does depend on the database connection supporting multiple concurrent connections (this is generally not true).

Use t1.Result etc. to get the returned values.

edited Mar 26, 2021 at 16:51

answered Mar 26, 2021 at 13:59

Richard

110k21 gold badges214 silver badges279 bronze badges

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

3 Comments

Yoismel Over a year ago

Thanks for your feedback @Richard. Please note that I updated the question to show how I am using Parallel.Invoke. Can you post an example of what you mean

Yoismel Over a year ago

Just applied this and worked. Already seeing better performance. Thanks a lot for your feedback.

Richard Over a year ago

@TheodorZoulias oops, forgot to fix that. Was thinking of other Parallel methods.

Stephen Cleary · Accepted Answer · 2021-03-26 14:23:28Z

4

As noted in the comments, you don't want to use Parallel at all. Instead, use asynchronous concurrency (await Task.WhenAll).

However, most database providers do not allow multiple simultaneous requests over the same database connection. So you will probably need to have three different db connections to run three queries concurrently.

internal static async Task<RequestAppointmentDBMasterObject> GetDAMasterObject(string facility, string sites, string state)
{
  RequestAppointmentDBMasterObject mo = new RequestAppointmentDBMasterObject();
  string querySites = query1;
  string queryPractitioners = query2;
  string queryExistingAppts = query3;

  using (IDbConnection db1 = new OdbcConnection(connectionString))
  using (IDbConnection db2 = new OdbcConnection(connectionString))
  using (IDbConnection db3 = new OdbcConnection(connectionString))
  {
    var task1 = db1.Query<Sites>(querySites).ToListAsync();
    var task2 = db2.Query<Practitioners>(queryPractitioners).ToListAsync();
    var task3 = db3.Query<AvailResponseObject>(queryExistingAppts).ToListAsync();
    await Task.WhenAll(task1, task2, task3);
    mo.SiteList = await task1;
    mo.PractitionerList = await task2;
    mo.ExistingAppointments = await task3;
  }
  return mo;
}

Doing multiple concurrent queries like this is possible, but it's usually better to do a single query that gets everything back all at once. That way you only need a single db connection, and all the query results are consistent with each other.

answered Mar 26, 2021 at 14:23

Stephen Cleary

460k78 gold badges721 silver badges858 bronze badges

5 Comments

Yoismel Over a year ago

thanks for your feedback. This clarifies a question I had "most database providers do not allow multiple simultaneous requests". I like the approach you are showing.

Yoismel Over a year ago

Seems this database supports concurrent connections since Richard's suggestion worked. I will have this in mind for future reference. Thanks Stephen.

Enigmativity Over a year ago

Is there a reason to do await Task.WhenAll(task1, task2, task3) followed by three lines awaiting each task again?

Stephen Cleary Over a year ago

@Enigmativity: Just preference. I prefer a single await Task.WhenAll to make the code as clear as possible.

Enigmativity Over a year ago

Fair enough. I had wondered if it were a personal coding preference. It is very clear.

Collectives™ on Stack Overflow

Using Task Parallel Library in API

2 Answers 2

3 Comments

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related