我有一个Azure函数应用程序,其中运行有多个函数。如果多个查询最终在同一时间运行,那么我将从DocumentQuery代码的深处得到一个InvalidCastException。如果我仅禁用一个功能(与哪个功能无关),就不会发生该错误,这使我相信存在某种共享内存问题。
我已删除所有内容,但有问题的查询除外,以试图隔离问题,从而确认问题确实源于对波斯菊Db的> 1次异步调用。
如果在评估另一个查询时调用DocumentQuery.ToString()
或await DocumentQuery.ExecuteNextAsync()
,似乎会出现此问题。具体来说,所有问题都出现在这里:
public static async Task<List<T>> ToListAsync<T>(this IQueryable<T> source, QueryCase queryCase = QueryCase.CamelCase) where T : class
{
var results = new List<T>();
IDocumentQuery<T> documentQuery = source.ChangeCase(queryCase).AsDocumentQuery();
while (documentQuery.HasMoreResults)
{
FeedResponse<T> queryResult = await documentQuery.ExecuteNextAsync<T>();
results.AddRange(queryResult);
}
return results;
}
该ChangeCase()
函数调用DocumentQuery.ToString()
,它将在那儿失败,如果我删除它,则失败将发生在ExecuteNextAsync()
堆栈跟踪:
System.InvalidCastException
HResult=0x80004002
Message=Unable to cast object of type 'Newtonsoft.Json.Converters.StringEnumConverter' to type 'Newtonsoft.Json.JsonConverter'.
Source=Microsoft.Azure.DocumentDB.Core
StackTrace:
at Microsoft.Azure.Documents.Linq.ExpressionToSql.ApplyCustomConverters(Expression left, SqlLiteralScalarExpression right)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.VisitBinary(BinaryExpression inputExpression, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.VisitScalarExpression(Expression inputExpression, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.VisitScalarLambda(Expression inputExpression, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.VisitWhere(ReadOnlyCollection`1 arguments, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.VisitMethodCall(MethodCallExpression inputExpression, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.Translate(Expression inputExpression, TranslationContext context)
at Microsoft.Azure.Documents.Linq.ExpressionToSql.TranslateQuery(Expression inputExpression)
at Microsoft.Azure.Documents.Linq.SqlTranslator.TranslateQuery(Expression inputExpression)
at Microsoft.Azure.Documents.Linq.DocumentQueryEvaluator.HandleMethodCallExpression(MethodCallExpression expression)
at Microsoft.Azure.Documents.Linq.DocumentQueryEvaluator.Evaluate(Expression expression)
at Microsoft.Azure.Documents.Linq.DocumentQuery`1.ToString()
at Api.Data.Extensions.DocumentDBLinqExtensions.ChangeCase[T](IQueryable`1 query, QueryCase queryCase) in C:\Users\jason\source\repos\HeroPlatform\HeroPlatform\Api.Data\Extensions\DocumentDbLinqExtensions.cs:line 138
编辑:
我发现从查询中使用的枚举的顶部删除[JsonConverter(typeof(StringEnumConverter))]
可以解决此错误(尽管现在我的数据正尝试按数字而不是字符串查询枚举)。
我实际上似乎没有在解决方案的任何项目中添加cosmos DB扩展版本。我正在创建一个IServiceCollection
并为其提供以下DocumentClient:
new DocumentClient(
new Uri(configuration["DocumentDb:EndpointUri"]),
configuration["DocumentDb:Key"],
serializerSettings: GlobalJson.SerializerSettings(),
connectionPolicy: new ConnectionPolicy()
{
MaxConnectionLimit = 100,
ConnectionMode = ConnectionMode.Direct,
ConnectionProtocol = Protocol.Tcp,
}))
GlobalJson.SerializerSettings()
var serializerSettings = new JsonSerializerSettings
{
ContractResolver = new CamelCaseExceptDictionaryKeysResolver(),
DateFormatHandling = DateFormatHandling.IsoDateFormat,
DateParseHandling = DateParseHandling.DateTimeOffset,
DateTimeZoneHandling = DateTimeZoneHandling.RoundtripKind
};
serializerSettings.Converters.Add(new StringEnumConverter());
return serializerSettings;
答案 0 :(得分:1)
这是回归。这是跟踪问题https://github.com/Azure/Azure-Functions/issues/1201。请pin Functions Runtime version至2.0.12382