查询Azure Table服务给出不同的结果

Question

I am trying to query azure table storage with Microsoft Azure Storage SDK. Sending simple request returns expected results (similar to https://docs.microsoft.com/en-us/azure/storage/storage-dotnet-how-to-use-tables#retrieve-a-range-of-entities-in-a-partition ). But when I send a more complex query like this one:

  (((((((((PartitionKey eq 'LOS-test1-20161218') or (PartitionKey eq 'SPA-test1-20161218')) or (PartitionKey eq 'LOS-test2-20161219')) or (PartitionKey eq 'SPA-test2-20161219')) or (PartitionKey eq 'SPA-test3-20161218')) or (PartitionKey eq 'LOS-test3-20161218')) or (PartitionKey eq 'SPA-test4-20161219')) or (PartitionKey eq 'LOS-test4-20161219')) or (PartitionKey eq 'SPA-test5-20161219')) or (PartitionKey eq 'LOS-test5-20161219')

I get only part of expected results (one third). Meanwhile when I run this query with Microsoft Azure Storage Explorer I get correct data.

Why does it happen?

private string GenerateFilterCondition(string s, string c, DateTime date)
{
        return TableQuery.GenerateFilterCondition(
            "PartitionKey", QueryComparisons.Equal,
            s + "-" + c + "-" + date.ToString("yyyyMMdd"));
}

private List<SiteSpecificDataTableDataSet> GetSiteSpecificDataTableDataSet()
{           
    _connection.Open();
    DateTime selectedDate = DateTime.Now.AddDays(-1);

    TableQuery<DataEntity> tq = new TableQuery<DataEntity>().Where(GenerateFilterCondition("SPA", "test5", selectedDate));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("LOS", "test5", selectedDate));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("SPA", "test4", selectedDate.AddDays(1)));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("LOS", "test4", selectedDate.AddDays(1)));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("LOS", "test3", selectedDate));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("SPA", "test3", selectedDate));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("LOS", "test2", selectedDate.AddDays(1)));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("SPA", "test2", selectedDate.AddDays(1)));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("LOS", "test1", selectedDate.AddDays(1)));
    tq = AddFilterORToTableQuery(tq.FilterString, GenerateFilterCondition("SPA", "test1", selectedDate.AddDays(1)));

    IEnumerable<DataEntity> entities = _connection.QueryData(tq);
}

And QueryData looks like:

    IEnumerable<DataEntity> res;
            List<DataEntity> queryResult = new List<DataEntity>();
            CloudTableClient tableClient = _storageAccount.CreateCloudTableClient();
            CloudTable table = tableClient.GetTableReference("data");
            TableContinuationToken tct = null;
            do
            {
                // Retrieve a segment (up to 1,000 entities).
                TableQuerySegment<DataEntity> tableQueryResult = table.ExecuteQuerySegmentedAsync(query, tct).Result;                
                if (tableQueryResult.Results.Count == 0) break;
                tct = tableQueryResult.ContinuationToken;
                foreach (var result in tableQueryResult.Results)
                {
                    queryResult.Add(result);
                }

            } while (tct != null);
            res = queryResult;

Answer 1

This condition in your code is incorrect: "if (tableQueryResult.Results.Count == 0) break", you may still need to continue querying even when the result is empty.

The correct condition should be: "if (tableQueryResult.ContinuationToken == null) break".

BTW, querying different Partition Keys in one filter string is highly not recommended since per Azure Table design it'll result in whole table scan which is terrible in performance. You need to split your query into multiple (one Partition Key at a time) queries to achieve better performance.