Develop Reference

Transactions - Entity Framework 5

I am experiencing a "Transaction timeout exceeded" issue in the code snippet below
using (var scope = new TransactionScope(TransactionScopeOption.RequiresNew,
new TransactionOptions { IsolationLevel = IsolationLevel.ReadUncommitted, Timeout = new TimeSpan(1, 0, 0) }))
{
try
{
segregationAssignment = new SegregationAssignment(dbContext).Assign(rmaU);
dbContext.SaveChanges();
scope.Complete();
scope.Dispose();
}
catch (DbUpdateException eb)
{
scope.Dispose();
return RedirectToAction("Details", details).WithErrorMessage(string.Format(Validations.not_possible_to_operation,
details.IsConfirmSegregate ? Buttons.segregate.ToLower() : Buttons.refuse.ToLower(), Models.rma_u));
}
catch (Exception ex)
{
scope.Dispose();
return RedirectToAction("Details", details).WithErrorMessage(string.Format(Validations.not_possible_to_operation,
details.IsConfirmSegregate ? Buttons.segregate.ToLower() : Buttons.refuse.ToLower(), Models.rma_u));
}
}
Inside it we have some additions of objects in context in the Assign method, only that. I do not understand why you're giving timeout in a process that takes a maximum of 10 minutes.
If you can help me, I'll be grateful. We are using version 5 of entity.

Overriding SaveChangesAsync when enclosed inside TransactionScope?

I'm fairly new to EF transactions, this is the code that is used for saving
public bool Save(TbArea area, bool isNew, out string errMsg)
{
try
{
errMsg = string.Empty;
using (var oScope = new System.Transactions.TransactionScope(TransactionScopeOption.Required, TimeSpan.FromSeconds(120)))
{
try
{
TbArea oEntity = oContext.TbArea.Where(a => a.AreaCode == area.AreaCode && a.CompanyId == MainClass.SystemCompanyId).FirstOrDefault();
if (oEntity != null)
{
if (isNew) { errMsg = Resources.ResSales.MsgRecordCodeDublicated; return false; }
oContext.TbArea.Attach(oEntity);
oContext.Entry(oEntity).CurrentValues.SetValues(area);
}
else
{
if (!isNew) { errMsg = Resources.ResSales.MsgRecordNotFoundInDB; return false; }
oContext.TbArea.Add(area);
}
oContext.SaveChangesAsync();
oScope.Complete();
return true;
}
catch (Exception ex)
{
oScope.Dispose();
errMsg = ex.Message;
return false;
}
}
I'm overriding SaveChangesAsync
so that I can save the ChangeTracker.Entries into the database.
this is a portion of the code:
dbContext.AcceptAllChanges();
logsSet.AddRange(audits);
int result = 0;
try
{
result = await dbContext.SaveChangesAsync(cancellationToken).ConfigureAwait(false);
//scope.Complete(); there was a transaction that I commented out, I thought it might overlaps the original transaction!
return result;
}
catch (Exception ex)
{
var m = ex.Message;
return result;
}
When I save an item, I get the error :
the transaction has aborted
when I remove the transaction scope the saving happens normally!

Your code is marking the transaction complete before the changes have finished saving:
oContext.SaveChangesAsync();
oScope.Complete();
You need to use await:
await oContext.SaveChangesAsync();
oScope.Complete();
If you're in a context where await can resume on a different thread, you'll probably also need to specify TransactionScopeAsyncFlowOption.Enabled.

Inexplicable NullReferenceException that only occurs in production

We have an ASP.NET website running which throws a NullReference-exception along with a stacktrace and a line number that is simply impossible. And I can't make heads nor tails from it myself.
It says:
Exception at ReportService.GetReport(String reportType) in ReportService.cs:line 1458
which is funny, because that is this line:
var exports = new List<ReportExport>();
Thanks to the (very short) stacktrace, I can see that the error is triggered in the GetReport-function and not in the "GetAllUsers" or "GetAllUsersWithFilter" functions, because I would receive a different error message in my e-mailbox or I would see it pop up in the stacktrace.
So I suspect the line number is wrong, in which case there is only one other possibility and that is this line:
foreach (var userProfile in users) {
exports.Add(CreateUserProfile(userProfile));
}
But how could users ever be null?
Full (albeit simplified) code right here:
public function IList<ReportExport> GetReport(string reportType) {
try {
IQueryable<UserProfile> users = null;
switch (reportType) {
case "abc" :
users = GetAllUsersWithFilter();
break;
case default:
users = GetAllUsers();
break;
}
var exports = new List<ReportExport>();
foreach (var userProfile in users) {
exports.Add(CreateUserProfile(userProfile));
}
} catch (Exception ex) {
SendErrorMail("GetReport has failed", ex); /* I receive this error mail */
}
function IQueryable<UserProfile> GetAllUsers() {
try {
return dbContext.Users.Where(x => x.IsRegistered == true);
} catch (Exception ex) {
SendErrorMail("GetAllUsers", ex); /* I don't receive this e-mail */
return null;
}
}
function IQueryable<UserProfile> GetAllUsersWithFilter() {
try {
return GetAllUsers().Where(x => x.ExtraFilter == true);
} catch (Exception ex) {
SendErrorMail("GetAllUsersWithFilter", ex); /* I don't receive this e-mail */
}
}
function int GetNumberOfSessions(int userId) {
try {
return dbContext.Sessions.Count(x => x.UserId == userId);
} catch (Exception ex) {
SendErrorMail("GetNumberOfSessions", ex); /* I don't receive this e-mail */
}
}
function ReportExport CreateUserExport(UserProfile user) {
try {
var cnt = GetNumberOfSessions(user.Id);
return new ReportExport() {
UserId = user.Id,
NumberOfSessions = cnt
}
} catch (Exception ex) {
SendErrorMail(("CreateUserExport", ex);
}
}

If you are in production then you might be running with optimizations switched on - therefore the line number will be wrong.
But how could users ever be null?
But you are catching the Exception then returning null. You are relying on returning data - which may not be the case in GetAllUsers.
function IQueryable<UserProfile> GetAllUsers() {
try {
return dbContext.Users.Where(x => x.IsRegistered == true);
} catch (Exception ex) {
SendErrorMail("GetAllUsers", ex); /* I don't receive this e-mail */
return null;
}
}

Returning result in nested try catch block

I am implementing some performance counters and I would like to know your opinion.
The question is should I declare response and return it outside try block or Is it OK to return it directly in the try block. Is there a difference and If so, what sample code is valid (if any).
With best regards, no9.
public string TestMethod(string document)
{
try
{
WebService ws = new WebService();
string response = null;
var startTime = PerformanceCounter.GetPerformanceCounterStartTimeHandle();
try
{
response = ws.InsertDocument(document);
}
catch (Exception ex)
{
PerformanceCounterHelper.Increment(PerformanceCounterEnum.NumberOfExternalWsCallsExceptionOnSec);
throw;
}
finally
{
PerformanceCounterHelper.IncrementPerformanceCounterByElapsedTime(PerformanceCounterEnum.DurationOfExternalCallsInSec, startTime);
PerformanceCounterHelper.Increment(PerformanceCounterEnum.NumberOfExternalCallsOnSec);
}
return response;
}
catch (Exception ex)
{
log.EventError(ex);
throw new DocumentGeneralException();
}
}
versus:
public string TestMethod(string document)
{
try
{
WebService ws = new WebService();
var startTime = PerformanceCounter.GetPerformanceCounterStartTimeHandle();
try
{
return ws.InsertDocument(document);
}
catch (Exception ex)
{
PerformanceCounterHelper.Increment(PerformanceCounterEnum.NumberOfExternalWsCallsExceptionOnSec);
throw;
}
finally
{
PerformanceCounterHelper.IncrementPerformanceCounterByElapsedTime(PerformanceCounterEnum.DurationOfExternalCallsInSec, startTime);
PerformanceCounterHelper.Increment(PerformanceCounterEnum.NumberOfExternalCallsOnSec);
}
}
catch (Exception ex)
{
log.EventError(ex);
throw new DocumentGeneralException();
}
}

As long as there isn't a difference because of not exiting (i.e. it runs additional/different code), then the code is identical. Actually, at the IL level it is illegal to ret from inside a try/catch, so one of the things the compiler does is to do exactly what you have done: introduce a local, assign the local inside the try/catch, then return that value when outside the try/catch.
Basically, go with whatever is simplest and most convenient to read. In your case, I would say "the first one".

Null values while updating DB in Parallel.Foreach

I use following script to get data from external service and store in dB. In certain rare cases less than 1% records gets updated with null values. In below code, the "re.status=fail" we see null. let us know if any thots.
public void ProcessEnquiries()
{
List<req> request = new List<req>();
var options = new ParallelOptions { MaxDegreeOfParallelism = Convert.ToInt32(System.Configuration.ConfigurationManager.AppSettings["MaxDegreeOfParallelism"]) };
try
{
Parallel.ForEach(request, options, currentRequest =>
{
ProcessedRequest processedRequest = null;
processedRequest = CommunicateToWS(currentRequest); // Here we call to webservice
});
}
catch (AggregateException exception)
{
foreach (Exception ex in exception.InnerExceptions)
{
// Handle Exception
}
}
}
public ProcessedRequest CommunicateToWS(req objReq)
{
ProcessedRequest re = new ProcessedRequest();
using (WebCall obj = new WebCall())
{
re.no = refnu;
try
{
retval = obj.getValue(inval);
objProxy.Close();
//get data
// parse and store to DB
}
catch (Exception e)
{
re.status = "fail";
//update DB that request has failed
//Handle Exception
obj.Close();
}
}
}