Michael Bell’s Blog

I had a situation where I needed to do this recently. I googled of course, and came up with an overloaded FindControl method. I implemented this in my code, and all seemed well. UNTIL… came the time to get the data posted from my control. I had controls nested in a UserControl, and that usercontrol nested within a FormView. When the postback occurred, I was able to retrieve all values from controls directly in the formview but none from within the controls inside my usercontrol. A buddy of mine helped me out on this, and believe me.. it was no fun tracking down. the values “not being posted” was actually a flaw in the FindControl method override. Please use this piece of code in the naming container that your validators reside, if they in fact need to validate against controls outside:

protected override Control FindControl(string id, int pathOffset)
{
Control c = base.FindControl(id, pathOffset);
if (c != null)
return c;
return FindControl(Page, id);
}
public Control FindControl(Control parent, string id)
{
Control recurse;
if (parent.ID == id)
{
return parent;
}
foreach (Control child in parent.Controls)
{
recurse = FindControl(child, id);
if (recurse != null)
{
return recurse;
}
}
return null;
}

I’ve been running image hosting sites for a couple of years. Through my trials I discovered a few things. Running the web server AND the sql server off the same box when you are getting a couple of million image hits a day can be trying on your system. The first version of my image hosting site wrote logs as they came in. 1 image hit actually equalled 3 queries to the DB. That was due to a lack of experience in both architecture and ASP.NET.

In July I decided to rewrite my ASP.NET 1.1 in 2.0. I have plenty of ram in my box, and i knew I wanted to do less queries to my database. What I wound up doing was an in-memory datatable that gets a record for each hit. I used the TableNewRow datatable event to check the record count with each insert. Probaboy not the best method… but I went with it. A more preferable way to do it would be to use a Timer object and firing on the Tick event. In any case, once I reach 10k records, I do a “dump” to the database of the in-memory stuff, and then clean my in-memory table.

As for the “dump”… well… the redesign could not have come at a better time. I was able to utilize the new .NET 2.0 SqlBulkCopy class. What this amounted to was a 4 second query every 10 minutes. Compare that to the alternative of doing several inserts a second… and you can start to see the advantage of in the in-memory/SqlBulkCopy dump technique.

After I released my new version of my image hosting site, I found out REAL fast that there were some problems related to sharing in-memory datatables that are accessed (read, write) constantly (writes 15-20 times a second).

It took me quite a bit of trial and error to figure out how to get it stablized. The server I was working on had dual xeon processors with hyper-threading enabled. What I found was that when this combination was being used, and I was logging hits to an in-memory datatable 15-20 times a second, different threads were trying to log to this table at the same time, causing all hell to break loose.

The fix – briefly lock the table during writes:
lock (MemoryTables.HitList)
{
    MemoryTables.HitList.Rows.Add(row);
}

The dump/rebuild procedure also requires a lock, but it’s very brief:
using (SqlConnection sourceConnection = new SqlConnection(_ConnectionString))
            {
                sourceConnection.Open();

                            // Insert into Event Log so we know logging is taking place as expected;
                            EventLog MyLog = new EventLog();
                            MyLog.Source = SourceName;
                            MyLog.WriteEntry(“Hit data has been dumped. HitList Records: ” + _HitList.Rows.Count, EventLogEntryType.Information);
                        }
                        catch (Exception ex)
                        {
                            throw;
                        }
                        finally
                        {
                            _HitList.Clear();
                        }
                    }
                    #endregion
                }
            }

I half-expected the locking to cause a delay problem in my logging abilities, but I’ve noticed no bottlenecks at all, and this method has been operating trouble free for about a month now.

I got this into a “release” state today.. and thought I’d put it out there for anyone who wants to use it. It’s a Windows looking time picker control. Have a look!

http://www.MichaelKBell.com/TimePicker

So, you’ve got the code in place to create a site in IIS via c# in one of your web apps, but you want the site ASP.NET version to default to 2.0. I had the problem and couldn’t find a simple “switch” to flip, so I wound up writing this little piece of code that goes in and manually remaps the files that need it, from the 1.1 framework to the 2.0. After that runs, your IIS ASP.NET version will read 2.0, and act accordingly

ManagementObjectSearcher searcher = new ManagementObjectSearcher(@”root\MicrosoftIISv2″,
    “SELECT * FROM IIsWebVirtualDirSetting WHERE Path = ‘” + pathToRoot.Replace(@”\”, @”\\”) + webSiteName + “‘”);