Azure table storage query returning data from wrong partition?

Question

I'm using azure table storage, and I'm trying to iterate over the table quickish.

I must be doing it wrong, but I can't for the life of me see why;

When I specify just a single partition I end up getting back results for MULTIPLE partitions. Ie If I constrain the query by just using "pkey1", I get 1000 results back for "pkey1", and then 325 for "pkey2"

Completely confused as to how this can happen..

This is the code I'm using:

private CloudTableClient _client;
private string _tableName;
private class QueryState
{
    public CloudTableQuery<T> Ctq;
    public Action<IEnumerable<T>> Populator;
    public ManualResetEvent Mre;
    public string Pkey;

    public QueryState(CloudTableQuery<T> ctq, Action<IEnumerable<T>> populator, ManualResetEvent mre, string pkey)
    {
        Populator = populator;
        Ctq = ctq;
        Mre = mre;
        Pkey = pkey;
    }
}

    public void ParallelQueryWithClause(Action<IEnumerable<T>> populator, string[] partitionKeys)
    {
        List<ManualResetEvent> mre = new List<ManualResetEvent>();
        foreach (string pKey in partitionKeys)
        {
            //_retry.Go(tsc =>
            //    {
                    TableServiceContext tsc =  _client.GetDataServiceContext();
                    ManualResetEvent m = new ManualResetEvent(false);
                    mre.Add(m);
                    CloudTableQuery<T> query = tsc.CreateQuery<T>(_tableName).Where(e => e.PartitionKey == pKey).AsTableServiceQuery<T>();
                    Action<IAsyncResult> act  = null;
                    act = result =>
                        {
                            int retries = 0;
                            while (retries++ < 5)
                            {
                                try
                                {
                                    QueryState qsInternal = result.AsyncState as QueryState;
                                    CloudTableQuery<T> ctq = qsInternal.Ctq;
                                    ResultSegment<T> seg = ctq.EndExecuteSegmented(result);
                                    if (seg.Results.Count() > 0)
                                        populator(seg.Results);
                                    if (seg.ContinuationToken != null)
                                    {
                                        ctq.BeginExecuteSegmented(seg.ContinuationToken, iasync => act(iasync), qsInternal);
                                    }
                                    else
                                    {
                                        m.Set();
                                    }
                                    break;
                                }
                                catch(Exception ex)
                                {
                                    Logger.LogError(ex);
                                }
                            }
                        };
                    query.BeginExecuteSegmented(iasync => act(iasync), new QueryState(query, populator, m, pKey));
                //});
        }
        ManualResetEvent.WaitAll(mre.ToArray());
    }

And sample calling code:

AzureTableStorage<ProductEntity> _ats = new AzureTableStorage<ProductEntity>("Products");
string[] partitions = new string[] { "pkey1" };

    Dictionary<string, int> cntr = new Dictionary<string, int>();
    _ats.ParallelQueryWithClause(p =>
    {
        lock (cntr)
        {
            foreach (ProductEntity pe in p)
            {
                if (cntr.ContainsKey(pe.PartitionKey))
                    cntr[pe.PartitionKey]++;
                else
                    cntr.Add(pe.PartitionKey, 1);
            }
        }
    }, partitions);

Hopefully this makes sense, and someone can help!

Answer 1

You may be running into the situation where the closure value is being modified. http://marlongrech.wordpress.com/2010/06/02/closures-in-c-can-be-evil/ This would run the query with a pKey of null, effectively not filtering by the pKey and hence returning all the values in the table.

Try replacing

foreach (string pKey in partitionKeys)

with

foreach (string pKeyTmp in partitionKeys)
{
   string pKey = pKeyTmp;