Deadlocks occur when querying table storage many times concurrently (.NET Core 1.1.0)

I’m having trouble with some deadlocks when querying table storage many times simultaneously. This happens whether I call cloudTable.ExistsAsync() and cloudTable.ExecuteQuerySegmentedAsync().

Having downloaded the source code for 8.1.1, I can see that the root of the issue seems to be in TableOperationHttpResponseParsers.cs. The problem is that instances of ODataReader are calling the Read() method when they should be taking advantage of the available ReadAsync() method.

Are these issues going to be fixed in the 8.2.0 release? My problem could be related to https://github.com/Azure/azure-storage-net/issues/431

The scenario can be reproduced with a simple Console Application (.NET Core 1.1.0) and using the following code sample:

private const string connectionString = "...";
private const string tableName = "...";

public static void Main(string[] args)
{
    Task.Run(async () =>
    {
        await MainAsync(args);
    }).GetAwaiter().GetResult();
}

static async Task MainAsync(string[] args)
{
    Console.WriteLine("Hello World!");
            
    var cloudStorageAccount = CloudStorageAccount.Parse(connectionString);
            
    // run the query many times simultaneously to produce the thread lock issue
    await Task.WhenAll(Enumerable.Range(0, 100).Select(m =>
    {
        return Query(cloudStorageAccount);
    }));

    Console.WriteLine(string.Format("We never get here..."));

    Console.ReadLine();
}

private static async Task<bool> Query(CloudStorageAccount cloudStorageAccount)
{
    var cloudTableClient = cloudStorageAccount.CreateCloudTableClient();
    var cloudTable = cloudTableClient.GetTableReference(tableName);

    Console.WriteLine(string.Format("Exists query started"));

    // threadlock occurs here when calling table.ExistsAsync() or table.ExecuteQuerySegmentedAsync()
    var result = await cloudTable.ExistsAsync();

    Console.WriteLine(string.Format("Exists query finished"));

    return result;
}

[unaizorrilla]

+1 here… any update about this issue

[benharden]

I’ve tried increasing the threadpool size by adding the following to the my project’s csproj:

<ThreadPoolMinThreads>4</ThreadPoolMinThreads>
<ThreadPoolMaxThreads>25</ThreadPoolMaxThreads>

This hasn’t made any discernible difference and still threadlocks. Am I doing this correctly?

Is there any other way I can use the azure-storage-net table library and not have it lock up on me? This is causing us a real headache on our production system and it seems ridiculous that we can’t make 100 calls to the server without it locking up.