IIS worker process keeps hogging memmory and crashes eventually leaving un-recycled app pool to cause 500 error

[wickstargazer]

Hi,

I think the subject does explain the root cause of the problem. My implementation is

static string tpath = Path.Combine(HttpContext.Current.Server.MapPath("~/wkhtmltox/"), Guid.NewGuid().ToString(), "MYPDFAPP_WKHTMLTOX");
static object _padLock;
static IConverter _converter;
        static IConverter Converter
        {
            get
            {

                if (_converter == null)
                {
                    lock(_padlock){
                        if(_converter == null)
                             _converter = new ThreadSafeConverter(new RemotingToolset<PdfToolset>(new Win64EmbeddedDeployment(new StaticDeployment(tpath))));
                    }
                }
                return _converter;
            }
        }

I also have the loaduserprofile in app pool set to true which helps only to prolong the time from 0 to crash (eating up around 80% of Ram then crashes)

Please help, this service is absolute neccessary other wise i have to move on to another solution..

Regards

[alin-prisecaru]

Hi wickstargazer,

I've been using this plugin for over 2 years - works nicely even with 1000+ pages of PDF generated from HTML. I use these packages: package id="TuesPechkin" version="2.1.1" targetFramework="net45" package id="TuesPechkin.Wkhtmltox.Win64" version="0.12.2.1" targetFramework="net45"

Also, I had to make this fix a while ago: In Visual Studio, I opened up Tools -> Options -> Projects and Solutions -> Web Projects and I had to make sure that the option called "Use the 64 bit version of IIS Express for web sites and projects" is checked.

Implementation: Create a static class in your application:

public static class TuesPechkinInitializerService
    {
        private static string staticDeploymentPath = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "wkhtmltopdf");

        public static void CreateWkhtmltopdfPath()
        {
            if (Directory.Exists(staticDeploymentPath) == false)
            {
                Directory.CreateDirectory(staticDeploymentPath);
            }
        }

        public static IConverter converter = 
            new ThreadSafeConverter(
                new RemotingToolset<PdfToolset>(
                    new Win64EmbeddedDeployment(
                        new StaticDeployment(staticDeploymentPath)
                    )
                )
            );
    }

In GLOBAL.asax, I initialize that class on project start:

TuesPechkinInitializerService.CreateWkhtmltopdfPath();

And to use it:

HtmlToPdfDocument pdfDocument = new HtmlToPdfDocument
                {
                    GlobalSettings = new GlobalSettings(),
                    Objects =
                    {
                        new ObjectSettings
                        { 
                            ProduceLocalLinks = true,
                            ProduceForms = true,
                            HtmlText = htmlContent
                        }                    
                    }
                };

                byte[] pdfDocumentData = TuesPechkinInitializerService.converter.Convert(pdfDocument);

Hope this helps somehow. My server resources are also a bit strained when a HTML to PDF conversion is done, but in the end it cleans itself up.

By looking at your implementation, I can see that the class is initialized every time it's used - I don't think that's a good approach for this interface because it was made for desktop applications and it's purpose is to be initialized only one time per application instance.

[wickstargazer]

Hi my PDFUtil class is static ofcourse, so is the variables they are declared only once.

I tried shuffling things around, making its caller static as well or not static but what i find is this...

byte[] pdfBuf = TuesPechkinService.Converter.Convert(document);

This line .... causes the converter to eat up memory i think it is because it is converting the file inside the memory stream itself and the issue it, that memory is never returned.

Everything works fine as i said before. Its just that this memory when not returned by converter or dispose or unload whatever we want to call it. It crashes the app pool eventually.

Please help!

[wickstargazer]

*Bump

[basicvn]

Hi wickstargazer,

Did you try to unload the Toolset by calling TuesPechkinService.ToolSet.Unload() after Convert action?

I created another property in Service class to reference ToolSet (also static), so I can unload it after convert done.

[wickstargazer]

Hey, I solve the problem temporary by calling GC.Collect()

[ghost]

Hello,

Do you have any updates on this track? Any workarounds to avoid this issue?

Thanks.

[sameerAhmad9291]

I followed @alin-prisecaru approach and unload convert using TuesPechkinService.ToolSet.Unload(). I did worked for bulk generation of 1000+ pdfs. But it takes around 10 seconds to generate one pdf.

[sandeepgosavi]

I tried all the solutions stated above but, still PROD crashed. I am using pdftoolset and imagetoolset both.

[sameerAhmad9291]

i also faced issue when i loop through object list which contains pfdFile: byte[] attribute. Make sure you did clear or control your pdfFile attribute.

I solved this by nhibernate lazyloading property.