How to get rid of deadlock in a SQL Server 2005 an

2019-03-31 03:53发布

问题:

I have a code in C# for windows service which is mainly responsible to update records in a table in my database, but I get always a lot of errors in my log, all errors are about deadlock of resource,

This is the error:

System.Data.SqlClient.SqlException (0x80131904): Transaction (Process ID 57) was deadlocked on lock resources with another process and has been chosen as the deadlock victim. Rerun the transaction. at System.Data.SqlClient.SqlConnection.OnError(SqlException exception, Boolean breakConnection, Action1 wrapCloseInAction) at System.Data.SqlClient.TdsParser.ThrowExceptionAndWarning(TdsParserStateObject stateObj, Boolean callerHasConnectionLock, Boolean asyncClose) at System.Data.SqlClient.TdsParser.TryRun(RunBehavior runBehavior, SqlCommand cmdHandler, SqlDataReader dataStream, BulkCopySimpleResultSet bulkCopyHandler, TdsParserStateObject stateObj, Boolean& dataReady) at System.Data.SqlClient.SqlCommand.FinishExecuteReader(SqlDataReader ds, RunBehavior runBehavior, String resetOptionsString) at System.Data.SqlClient.SqlCommand.RunExecuteReaderTds(CommandBehavior cmdBehavior, RunBehavior runBehavior, Boolean returnStream, Boolean async, Int32 timeout, Task& task, Boolean asyncWrite) at System.Data.SqlClient.SqlCommand.RunExecuteReader(CommandBehavior cmdBehavior, RunBehavior runBehavior, Boolean returnStream, String method, TaskCompletionSource1 completion, Int32 timeout, Task& task, Boolean asyncWrite) at System.Data.SqlClient.SqlCommand.InternalExecuteNonQuery(TaskCompletionSource`1 completion, String methodName, Boolean sendToPipe, Int32 timeout, Boolean asyncWrite) at System.Data.SqlClient.SqlCommand.ExecuteNonQuery() at WheelTrackListener.DataAccess.SQLDBA.ExecuteNQuery(SqlCommand cmd, Boolean isShowError, ConnectionStringType CountryCode, String deviceID, Int32 retry, String functionCallName) ClientConnectionId:e45e4cf1-a113-46b7-b9b5-dc5ee8111406

Now, I want to ask, can I make a try or check if resource is locked? and if locked how to wait update until it is released?

Here is my current code:

public static int updateVehicleLastPosition(string UTCDate, string UTC_Time, 
              string NS_Indicator, string Latitude, string EWIndicator, 
              string Longtitude, string Speed, string Processed, 
              string Near_ByLocation, string Near_ByLocation_AR, 
              string Gis_dataID, string address, string ar_adress, string Device_ID)
{
        SQLMethods sql = new SQLMethods();
        SqlCommand cmd = sql.cmdUpdateVehicleLastPosition(UTCDate, UTC_Time, NS_Indicator, Latitude, EWIndicator, Longtitude, Speed, Processed, Near_ByLocation, Near_ByLocation_AR, Gis_dataID, address, ar_adress, Device_ID);
        SQLDBA sqlDBA = new SQLDBA();
        return sqlDBA.ExecuteNQuery(cmd, true, ConnectionStringType.OMN, Device_ID, 10, "updateVehicleLastPosition");
}

public SqlCommand cmdUpdateVehicleLastPosition(string UTCDate, string UTC_Time, 
                      string NS_Indicator, string Latitude, string EWIndicator, 
                      string Longtitude, string Speed, string Processed, 
                      string Near_ByLocation, string Near_ByLocation_AR, 
                      string Gis_dataID, string address, string ar_adress, 
                      string Device_ID)
 {
        string sql = "UPDATE CTS_VehicleLastPosition SET [UTCDate] = @UTCDate, [UTC_Time] = @UTC_Time, [NS_Indicator] = @NS_Indicator, [Latitude] = @Latitude, [EWIndicator] = @EWIndicator, [Longtitude] = @Longtitude, [Speed] = @Speed, [Processed] = @Processed, [Near_ByLocation] = @Near_ByLocation, [Near_ByLocation_AR] = @Near_ByLocation_AR, [Gis_dataID] = @Gis_dataID, [address] = @address, [ar_adress] = @ar_adress WHERE [Device_ID] = @Device_ID";

        SqlCommand cmd = new SqlCommand(sql);
        cmd.Parameters.AddWithValue("@UTCDate", UTCDate);
        cmd.Parameters.AddWithValue("@UTC_Time", UTC_Time);
        cmd.Parameters.AddWithValue("@NS_Indicator", NS_Indicator);
        cmd.Parameters.AddWithValue("@Latitude", Latitude);
        cmd.Parameters.AddWithValue("@EWIndicator", EWIndicator);
        cmd.Parameters.AddWithValue("@Longtitude", Longtitude);
        cmd.Parameters.AddWithValue("@Speed", Speed);
        cmd.Parameters.AddWithValue("@Processed", Processed);
        cmd.Parameters.AddWithValue("@Near_ByLocation", Near_ByLocation);
        cmd.Parameters.AddWithValue("@Near_ByLocation_AR", Near_ByLocation_AR);
        cmd.Parameters.AddWithValue("@Gis_dataID", Gis_dataID);
        cmd.Parameters.AddWithValue("@address", address);
        cmd.Parameters.AddWithValue("@ar_adress", ar_adress);
        cmd.Parameters.AddWithValue("@Device_ID", Device_ID);
        return cmd;
}

public int ExecuteNQuery(SqlCommand cmd, bool isShowError, 
                    DataAccess.ConnectionStringType CountryCode, string deviceID, 
                    int retry, string functionCallName)
{
        ConnectionManager Connection = new ConnectionManager();
        try
        {
            Connection.GetConnection(CountryCode);
            if ((Connection.con == null) || (Connection.con.State != ConnectionState.Open))
            {
                if (retry <= 0) return 0;
                else return ExecuteNQuery(cmd, isShowError, CountryCode, deviceID, retry - 1, functionCallName);
            }
            int rowsAffected = 0;
            cmd.Connection = Connection.con;
            rowsAffected = cmd.ExecuteNonQuery();
            return rowsAffected;
        }
        catch (SqlException sqlexception)
        {
            if (isShowError)
                LEAMRALogger.Logger.WriteByDate("Logs\\SQLDBA\\" + functionCallName + "\\" + String.Format("{0:dd-MM-yyyy}", DateTime.Now), "SQLDBA", "SQLDBA_ERROR", "ExecuteNQuery Function: [deviceID: " + deviceID + " | retry: " + retry + "] " + sqlexception.ToString());
        }
        catch (Exception ex)
        {
            if (isShowError)
                LEAMRALogger.Logger.WriteByDate("Logs\\SQLDBA\\" + functionCallName + "\\" + String.Format("{0:dd-MM-yyyy}", DateTime.Now), "SQLDBA", "SQLDBA_ERROR", "ExecuteNQuery Function: [deviceID: " + deviceID + " | retry: " + retry + "] " + ex.ToString());
        }
        finally
        {
            if ((Connection.con != null) && (Connection.con.State == ConnectionState.Open))
            {
                Connection.con.Close();
                Connection.con.Dispose();
            }

            GC.Collect();
        }
        if (retry <= 0) return 0;
        else return ExecuteNQuery(cmd, isShowError, CountryCode, deviceID, retry - 1, functionCallName);
}

回答1:

I can't see any explicit transaction scope in your code, so I do not know what locks are already in place when you are doing your update; also it is not clear what isolation level you are using. But the commonest scenario in this type of situation is that earlier in the same transaction you have issued a select (read lock) on the same rows that you are trying to upate later. This will cause a lock escalation, and can result in a deadlock if two transactions are trying to do the same thing:

  1. Transaction A: select with read lock
  2. Transaction B: select with read lock
  3. Transaction A: update - wants to escalate its read lock to a write lock, but has to wait for transaction B to release its read lock
  4. Transaction B: update - wants to escalate its read lock to a write lock, but has to wait for transaction A to release its read lock.

Bingo! deadlock as both A and B are waiting on each other to release their existing read locks before they can do their update.

To prevent this you need an updlock hint in your select, e.g.,

select * from table with (updlock) where blah blah

This will ensure your select uses a write lock instead of a read lock, which will prevent lock escalation between concurrent transactions.



回答2:

Although it's possible to check for a lock you cannot guarantee that by the time the next statement is issued some other process hasn't taken out a lock. Possible solutions in order of preference:

1) Always reference tables in the same order within trancations.

2) @ShellShock's answer

3) Trap the deadlock error and handle it.