App Insights memory leak in background tasks - asp.net-web-api

I am using Simple Injector DI in .NET web api (.net 4.8 framework) and have a wrapper class(my own implementation) around the telemetry client.
public class MyAppInsights : IMyAppInsights
{
private static TelemetryClient telemetry;
public MyAppInsights ()
{
if (telemetry == null)
{
telemetry = new TelemetryClient();
}
}
//other interface method implementations
}
simple injector startup configuration looks like
container.Register<IMyAppInsights , MyAppInsights >(Lifestyle.Singleton);
I am noticing lots of Httpwebrequest objects (app insights) are being added to memory and not released for a long running background tasks. This problem appeared only recently after implementing long running background tasks which was causing out of memory exceptions.
Any pointers on how to reduce/plug the memory leak / consumption would be greatly appreciated

You can try to remove Lifestyle.Singleton to see if you still have memory leak.
Also check the site for example.
https://docs.simpleinjector.org/en/latest/quickstart.html

Related

NET Core Server Side multiple session Blazor

I'm trying to host my Blazor application on my server.
I spent all the summer on it and I just realized every time I open my website on new device it doesn't create a new session restarting from zero, but continues where I left it. The worst part is there is a login system behind it, so I feel super dumb at the moment.
I really need a big hint on how to fix this "not little" issue.
Is there a way to make server create new session every time someone open the website (without making it loose to other users)?
The solution should be use a Client Template instead, but the performance are really to slow.
UPDATE:
Accounts "user password" are:
- user user
- test test
Download project sample (requires Net Core 3.0)
[SOLUTION] itminus found the solution to my issue.
You have also to add in ConfigureServices in Startup.cs this services.AddScoped<Storage>();
public void ConfigureServices(IServiceCollection services)
{
services.AddRazorPages();
services.AddServerSideBlazor();
services.AddScoped<Storage>();
}
every time I open my website on new device it doesn't create a new session restarting from zero, but continues where I left it.
I checkout your code and find that you're using Singleton Pattern to initialize the Storage. If I understand it correctly, this Storage singleton instance will be shared across different users (also across different devices). As this instance will be used to render the Main.razor page, there will be concurrency problems that you're experiencing now .
To fix that issue, the Storage instance should be limited within some specific connection. As you're using Blazor Server Side, you could register the Storage as a Scoped Service:
In Blazor Server apps, a scoped service registration is scoped to the connection. For this reason, using scoped services is preferred for services that should be scoped to the current user, even if the current intent is to run client-side in the browser.
Firstly, remove the static singleton instance :
public class Storage
{
private static Storage instance;
private Storage()
{
}
public static Storage GetInstance()
{
if (Storage.instance == null)
Storage.instance = new Storage();
return Storage.instance;
}
public List<Items>list {get;set;} = new List<Items>();
public string password {get;set;}
}
Register this Class as a scoped service:
services.AddScoped<Storage>();
And then inject this service in your Login.razor and Main.razor :
#inject project.Storage Storage
Finally, you need change all the Storage.GetInstance(). to Storage.:
Storage.list = Order;
...
Storage.password = password;
I notice that you're also creating the Importer/Additional instance using the Singleton Pattern. I would suggest you should refactor them to use Service Injection in a similar way.

Syncfusion PdfViewerControl on Azure

I am utilizing Syncfusion's PdfViewerControl and PdfLoadedDocument classes to generate thumbnail images of a PDF. However, once I moved the project to an Azure App Service, the PdfViewerControl is throwing an exception when being initialized. I am curious if it is attempting to use system memory and Azure is blocking this. Below is the method GenerateThumbnails I've created and the exception is being thrown when creating a new PdfViewerControl. If anyone has a work around for this or has experienced something similar when moving to Azure, any assistance would be greatly appreciated.
Along with that, if someone knows of another tool to create thumbnails from a PDF in this manner that'd be very helpful as well. Thanks!
Exception:
System.AccessViolationException: 'Attempted to read or write protected memory. This is often an indication that other memory is corrupt.'
Method:
public static List<Byte[]> GenerateThumbnails(Byte[] file)
{
Int32 resizedHeight;
Int32 resizedWidth;
List<Byte[]> thumbnails = new List<Byte[]>();
using (PdfViewerControl pdfViewerControl = new PdfViewerControl())
using (PdfLoadedDocument pdfLoadedDocument = new PdfLoadedDocument(file, true))
{
// The PDF Viewer Control must load the PDF from a PdfLoadedDocument, rather than directly from the filename because
// when loaded from the filename, it is not disposed correctly and causes a file lock.
pdfViewerControl.Load(pdfLoadedDocument);
for (Int32 i = 0; i < pdfViewerControl.PageCount; ++i)
{
using (Bitmap originalBitmap = pdfViewerControl.ExportAsImage(i))
{
if (pdfViewerControl.LoadedDocument.Pages[i].Size.Width > pdfViewerControl.LoadedDocument.Pages[i].Size.Height)
{
resizedHeight = (PdfUtility.TARGET_THUMBNAIL_WIDTH_LANDSCAPE * originalBitmap.Height) / originalBitmap.Width;
resizedWidth = PdfUtility.TARGET_THUMBNAIL_WIDTH_LANDSCAPE;
}
else
{
resizedHeight = PdfUtility.TARGET_THUMBNAIL_HEIGHT_PORTRAIT;
resizedWidth = (PdfUtility.TARGET_THUMBNAIL_HEIGHT_PORTRAIT * originalBitmap.Width) / originalBitmap.Height;
}
using (Bitmap resizedBitmap = new Bitmap(originalBitmap, new Size(resizedWidth, resizedHeight)))
using (MemoryStream memoryStream = new MemoryStream())
{
resizedBitmap.Save(memoryStream, ImageFormat.Jpeg);
thumbnails.Add(memoryStream.ToArray());
}
}
}
}
return thumbnails;
}
Update
Web App for Containers on Windows is now supported. This allows you to bring your own docker container that runs outside of the sandbox, so the restrictions described below won't affect your application.
There are restrictions in the sandbox that the app is running in that prevents certain API calls.
Here is a list of frameworks and scenarios that have been found to be
not be usable due to one or more of the restrictions above. It's
conceivable that some will be supported in the future as the sandbox
evolves.
PDF generators failing due to restriction mentioned above:
Syncfusion Siberix Spire.PDF The following PDF generators are
supported:
SQL Reporting framework: requires the site to run in Basic or higher
(note that this currently does not work in Functions apps in
Consumptions mode) EVOPDF: See
http://www.evopdf.com/azure-html-to-pdf-converter.aspx for vendor
solution Telerik reporting: requires the site to run in Basic or
higher. More info here Rotativa / wkhtmltopdf: requires the site to
run in Basic or higher. NReco PdfGenerator (wkhtmltopdf): requires
subscription plan Basic or higher Known issue for all PDF generators
based on wkhtmltopdf or phantomjs: custom fonts are not rendered
(system-installed font is used instead) because of sandbox GDI API
limitations that present even in VM-based Azure Apps plans (Basic or
higher).
Other scenarios that are not supported:
PhantomJS/Selenium: tries to connect to local address, and also uses
GDI+.
https://github.com/projectkudu/kudu/wiki/Azure-Web-App-sandbox

Xamarin: 'System.Resources.MissingManifestResourceException' on WinPhone

I found an issue with localization in WinPhone app. I added couple of resources and implemented localization. It is working fine in simulator, but crashing on Phone with exception Exception thrown: 'System.Resources.MissingManifestResourceException' in mscorlib.ni.dll
Any help?
Thanks to a bit of googling and a helpful blog post I have managed to solve this issue that happened intermittently with Xamarin.Forms projects.
This is the blog post that led me to the solution:
http://blog.tpcware.com/2016/06/xamarin-forms-localization/
Basically, the way you access resources on Android and iOS with Xamarin.Forms and Windows Phone is different:
To make it short, we need to “automagically” use the ResourceLoader.GetString(…) method when running on Store apps, while continuing to use the regular ResourceManager.GetString(…) method on all other platform. And because in the Xamarin Forms solution we use a resource file of type RESX, we also have the automatically generated resource class.
The super clever idea contained in the above linked post is to “hack” the resource class injecting a derived class of ResourceManager with an overridden GetString(…) method into the resource class “resourceMan” property (for a more detailed explanation of this hack, you can read the post).
We need to create our own version of ResourceManager like so and swap it for the existing ResourceManager using reflection:
public class WinRTResourceManager : ResourceManager
{
readonly ResourceLoader _resourceLoader;
private WinRTResourceManager(string baseName, Assembly assembly) : base(baseName, assembly)
{
_resourceLoader = ResourceLoader.GetForViewIndependentUse(baseName);
}
public static void InjectIntoResxGeneratedApplicationResourcesClass(Type resxGeneratedApplicationResourcesClass)
{
resxGeneratedApplicationResourcesClass
.GetRuntimeFields()
.First(m => m.Name == "resourceMan")
.SetValue(null, new WinRTResourceManager(
resxGeneratedApplicationResourcesClass.FullName,
resxGeneratedApplicationResourcesClass.GetTypeInfo().Assembly));
}
public override string GetString(string name, CultureInfo culture)
{
return _resourceLoader.GetString(name);
}
}
All that's then left to do is call this when the app starts for the first time:
WinRTResourceManager.InjectIntoResxGeneratedApplicationResourcesClass(typeof(AppResources));
After making these changes everything should work fine now.
Of course this is absolutely a hack but I have notified the Xamarin.Forms team of the issue and they are looking into it so hopefully it will be solved soon!

ASP.NET MVC 3 Site Loading Is Extremely Slow

I really don't know where to begin with this question, but the site I'm working on at times has some really slow page loads. Especially after doing a build, but not always. I usually have to refresh the page 5-10 times before it actually comes up. I guess I am trying to see where exactly I should begin to look.
ASP.NET MVC 3
Ninject
AutoMapper
Entity Framework Code First 4.1
SQL Server 2008
Razor
UPDATES
Concerning some of the questions, it can do this long loading on every page, but after it loads its fairly quick on all the pages.
After posting this and getting your replies I started the application and it is still loading and probably won't ever load unless I click reload on the browser.
No caching, and the EF models aren't huge.
I am using Razor and Visual Studio 2010 with 6 GB of memory and an I7 processor.
I am using IIS Express and the default web server when debugging. It also does this on IIS7 on the main server.
I may look into the MVC Profiler and Glimpse to see what I can find.
Below I have some code this runs when it hits the homepage. I would say it never loads when I first start up the server. I put a break point at var model which never gets hit. If I reload the page then it does.
public ActionResult Index()
{
var model = new HomeViewModel();
model.RecentHeadlines = _headlineService.GetHeadlines(1, Config.RecentHeadlinesPageSize, string.Empty);
return View(model);
}
Below is my datacontext setup also.
public class DatabaseFactory : Disposable, IDatabaseFactory
{
private DataContext _dataContext;
public DataContext Get()
{
return _dataContext ?? (_dataContext = new DataContext());
}
protected override void DisposeCore()
{
if (_dataContext != null)
_dataContext.Dispose();
}
}
public class Disposable : IDisposable
{
private bool isDisposed;
~Disposable()
{
Dispose(false);
}
public void Dispose()
{
Dispose(true);
GC.SuppressFinalize(this);
}
private void Dispose(bool disposing)
{
if (!isDisposed && disposing)
{
DisposeCore();
}
isDisposed = true;
}
protected virtual void DisposeCore()
{
}
}
public class UnitOfWork : IUnitOfWork
{
private readonly IDatabaseFactory _databaseFactory;
private DataContext _dataContext;
public UnitOfWork(IDatabaseFactory databaseFactory)
{
_databaseFactory = databaseFactory;
}
protected DataContext DataContext
{
get { return _dataContext ?? (_dataContext = _databaseFactory.Get()); }
}
public void Commit()
{
DataContext.Commit();
}
}
I'd start by checking what the timeouts are set to in IIS for the process to be recycling itself.
I'm also a very big fan of the MVC Mini-Profiler which could show you exactly how long various parts of your page load are taking, definitely take a look at it.
Edit:
It is worth noting that the Glimpse project is also great for this task these days.
Sounds like it might be an issue with IIS AppPool recycling if you're experiencing it after builds or after periods of inactivity.
To help with AppPool timeouts you can utilize a batch file I created to help mitigate the issue.
That won't solve the problem for you after new builds because your ASP.NET MVC application needs to be JIT-compiled upon first run. If you're really eager to eliminate that issue, you can use ASP.NET precompliation.
Try Glimpse or use ASP.NET Tracing.
You could also precompile your views if you are using the Razor view engine via Razor Single File Generator for MVC.
It depends on what happened in your previous run, sometimes if you throw an error and don't clear that out then you will have issues running the application. It helps to restart the browser every time you build if there was an error.
However, this could be an issue of caching. It is possible that your database is caching due to poorly maintained context disposing. This would cause the lookups to run faster and faster as they were encountered in pages. Make sure you always call .dispose() when done with your database transactions.
funny - I've noticed something similar once with unity and mvc but the problem I believe resolved itself. You could also try ants profiler to see if the problem is outside of MVC.
If you let a single request sit there (without requesting 5+ times) what happens?
Let a single request run - is ANY of your code hit? (setup logging log4net, nlog, etc) to run application_start, etc to see if any code is getting called after the compile.

How to speed up Azure deployment from Visual Studio 2010

I have Visual Studio 2010 solution with an Azure Service and an ASP.NET MVC 3 solution that serves as a Web Role for the Azure service. No other roles attached to the service other than that.
Every deployment to the Azure staging (or production, for that matter) environment takes up to 20 minutes to complete, form the moment I click publish on Visual Studio until all instances (2) are started.
As you can imagine this makes it a PITA to publish often, or to quick-fix some bugs. Is there a way to speed the process up? Would it be faster to upload the package to de Blob storage and upgrade from there? How would I go about achieving that?
I feel on-line docs on Azure leave a lot to be desired. Particularly when it comes to troubleshooting by the way.
Thanks.
One idea for reducing the need (and frequency) for redeploying is to move static content into blob storage, external to the package. For instance, move your css and javascript to blob storage, along with images. Once this is done, you'd only have to recompile / redeploy for .NET code changes. You can upload updated css, at any time, to blob storage. If you want to test this in staging first, you could always have a staging vs. production container name for your static content and store that container name in a config setting.
This doesn't change the deployment time when you do need to redeploy, but at least you can reduce how often you go through that process...
You should enable Web Deploy in your Azure project. It works this way :
1/ Create a RDP account (don't forget, you need to upload a certificate with its private key so that Azure can decipher the password). That is hidden in the Deploy Dialog Box for your Azure deployment project.
2/ Enable Web Deployment - same place
Once you've published the app that way, right-click in the web application (not the azure deployment project) and select Publish. The pop-up has everything defined except the password, enter that as well and you'll upload your changes to Azure in a matter of seconds.
CAVEAT : this is meant for single-instance web apps, definitely not the way to go for a production upgrade strategy, and the Blob storage answer already mentioned is the best option in that case.
Pierre
My solution to this problem is only to push a new package when I am changing code in the RoleEntryPoint or with the Service Definition. In Azure 1.3 you now have the ability to use Remote Desktop Connection. Using RDC, I will compile my code locally and use copy/paste to place it on the Azure server in the appropriate directory. Once the production code is running correctly, I can then push the fully tested version to staging and then do a VIP swap. This limits the number of times I actually have to deploy a package.
You actually have quite a long window in which you can keep modifying your code in Azure before you have to publish a new package. The new package is only really needed for those cases where Azure has to shutdown/restart your role instance.
It's a nice idea to try uploading your project to blob storage first, but unfortunately this is what Visual Studio is doing for you behind the scene anyway. As has been pointed out elsewhere, most of the time in doing the deploy is not the upload itself, but the stopping and starting of all of your update domains.
If you're just running this site in a development environment, then the only way I know to speed it up is to run just one instance. If this is the live environment, then... sorry, I think you're out of luck.
So that I don't have to deploy to the cloud to test minor changes, what I've found works quite well is to engineer the site so that it works when running in local IIS just like any other MVC site.
The biggest barrier to this working are settings that you have in the cloud config. The way we get around this is to make a copy of all of the settings in your cloud config and put them in your web.config in the appSettings. Then rather than using RoleEnvironment.GetConfigurationSettingValue() create a wrapper class that you call instead. This wrapper class checks RoleEnvironment.IsAvailable to see if it is running in the Azure fabric, if it is, it calls the usual config function above, if not, it calls WebConfigurationManager.AppSettings[].
There are a few other things that you'll want to do around getting the config setting change events which hopefully you can figure out from the code below:
public class SmartConfigurationManager
{
private static bool _addConfigChangeEvents;
private static string _configName;
private static Func<string, bool> _configSetter;
public static bool AddConfigChangeEvents
{
get { return _addConfigChangeEvents; }
set
{
_addConfigChangeEvents = value;
if (value)
{
RoleEnvironment.Changing += RoleEnvironmentChanging;
}
else
{
RoleEnvironment.Changing -= RoleEnvironmentChanging;
}
}
}
public static string Setting(string configName)
{
if (RoleEnvironment.IsAvailable)
{
return RoleEnvironment.GetConfigurationSettingValue(configName);
}
return WebConfigurationManager.AppSettings[configName];
}
public static Action<string, Func<string, bool>> GetConfigurationSettingPublisher()
{
if (RoleEnvironment.IsAvailable)
{
return AzureSettingsGet;
}
return WebAppSettingsGet;
}
public static void WebAppSettingsGet(string configName, Func<string, bool> configSetter)
{
configSetter(WebConfigurationManager.AppSettings[configName]);
}
public static void AzureSettingsGet(string configName, Func<string, bool> configSetter)
{
// We have to store these to be used in the RoleEnvironment Changed handler
_configName = configName;
_configSetter = configSetter;
// Provide the configSetter with the initial value
configSetter(RoleEnvironment.GetConfigurationSettingValue(configName));
if (AddConfigChangeEvents)
{
RoleEnvironment.Changed += RoleEnvironmentChanged;
}
}
private static void RoleEnvironmentChanged(object anotherSender, RoleEnvironmentChangedEventArgs arg)
{
if ((arg.Changes.OfType<RoleEnvironmentConfigurationSettingChange>().Any(change => change.ConfigurationSettingName == _configName)))
{
if ((_configSetter(RoleEnvironment.GetConfigurationSettingValue(_configName))))
{
RoleEnvironment.RequestRecycle();
}
}
}
private static void RoleEnvironmentChanging(object sender, RoleEnvironmentChangingEventArgs e)
{
// If a configuration setting is changing
if ((e.Changes.Any(change => change is RoleEnvironmentConfigurationSettingChange)))
{
// Set e.Cancel to true to restart this role instance
e.Cancel = true;
}
}
}
The uploading itself takes a bit more than a minute most of the time. It's the starting up of the instances that take up most of the time.
What you can do is to deploy your fixes to staging first (note that it costs money so don't let it be there for too long). Swapping from staging to production only takes a couple of seconds. So while your application's still running you can upload the patched version, let your testers test it on staging and when they give the go then simply swap it to production.
I haven't tested your possible alternative approach by first uploading to blob storage first. But I think that's overhead as it doesn't speed up starting up the instances.

Resources