Running windows service to watch service running grow memory (leak) - c#

I have checked all posts here, but can't find a solution for me so far.
I did setup a small service that should only watch if my other services I want to monitor runs, and if not, start it again and place a message in the application eventlog.
The service itself works great, well nothing special :), but when I start the service it use around 1.6MB of RAM, and every 10 seconds it grow like 60-70k which is way to much to live with it.
I tried dispose and clear all resources. Tried work with the System.Timers instead of the actual solution, but nothing really works as I want it, memory still grows.
No difference in debug or release version and I am using it on .Net 2, don't know if it make a difference to you 3,3.5 or 4.
Any hint?!
using System;
using System.IO;
using System.Diagnostics;
using System.ServiceProcess;
using System.Threading;
using System.Timers;
namespace Watchguard
{
class WindowsService : ServiceBase
{
Thread mWorker;
AutoResetEvent mStop = new AutoResetEvent(false);
/// <summary>
/// Public Constructor for WindowsService.
/// - Put all of your Initialization code here.
/// </summary>
public WindowsService()
{
this.ServiceName = "Informer Watchguard";
this.EventLog.Source = "Informer Watchguard";
this.EventLog.Log = "Application";
// These Flags set whether or not to handle that specific
// type of event. Set to true if you need it, false otherwise.
this.CanHandlePowerEvent = false;
this.CanHandleSessionChangeEvent = false;
this.CanPauseAndContinue = false;
this.CanShutdown = false;
this.CanStop = true;
if (!EventLog.SourceExists("Informer Watchguard"))
EventLog.CreateEventSource("Informer Watchguard", "Application");
}
/// <summary>
/// The Main Thread: This is where your Service is Run.
/// </summary>
static void Main()
{
ServiceBase.Run(new WindowsService());
}
/// <summary>
/// Dispose of objects that need it here.
/// </summary>
/// <param name="disposing">Whether or not disposing is going on.</param>
protected override void Dispose(bool disposing)
{
base.Dispose(disposing);
}
/// <summary>
/// OnStart: Put startup code here
/// - Start threads, get inital data, etc.
/// </summary>
/// <param name="args"></param>
protected override void OnStart(string[] args)
{
base.OnStart(args);
MyLogEvent("Init");
mWorker = new Thread(WatchServices);
mWorker.Start();
}
/// <summary>
/// OnStop: Put your stop code here
/// - Stop threads, set final data, etc.
/// </summary>
protected override void OnStop()
{
mStop.Set();
mWorker.Join();
base.OnStop();
}
/// <summary>
/// OnSessionChange(): To handle a change event from a Terminal Server session.
/// Useful if you need to determine when a user logs in remotely or logs off,
/// or when someone logs into the console.
/// </summary>
/// <param name="changeDescription"></param>
protected override void OnSessionChange(SessionChangeDescription changeDescription)
{
base.OnSessionChange(changeDescription);
}
private void WatchServices()
{
string scName = "";
ServiceController[] scServices;
scServices = ServiceController.GetServices();
for (; ; )
{
// Run this code once every 10 seconds or stop right away if the service is stopped
if (mStop.WaitOne(10000)) return;
// Do work...
foreach (ServiceController scTemp in scServices)
{
scName = scTemp.ServiceName.ToString().ToLower();
if (scName == "InformerWatchguard") scName = ""; // don't do it for yourself
if (scName.Length > 8) scName = scName.Substring(0, 8);
if (scName == "informer")
{
ServiceController sc = new ServiceController(scTemp.ServiceName.ToString());
if (sc.Status == ServiceControllerStatus.Stopped)
{
sc.Start();
MyLogEvent("Found service " + scTemp.ServiceName.ToString() + " which has status: " + sc.Status + "\nRestarting Service...");
}
sc.Dispose();
sc = null;
}
}
}
}
private static void MyLogEvent(String Message)
{
// Create an eEventLog instance and assign its source.
EventLog myLog = new EventLog();
myLog.Source = "Informer Watchguard";
// Write an informational entry to the event log.
myLog.WriteEntry(Message);
}
}
}

Your code may throw an exceptions inside loop, but these exception are not catched. So, change the code as follows to catch exceptions:
if (scName == "informer")
{
try {
using(ServiceController sc = new ServiceController(scTemp.ServiceName.ToString())) {
if (sc.Status == ServiceControllerStatus.Stopped)
{
sc.Start();
MyLogEvent("Found service " + scTemp.ServiceName.ToString() + " which has status: " + sc.Status + "\nRestarting Service...");
}
}
} catch {
// Write debug log here
}
}
You can remove outer try/catch after investigating, leaving using statement to make sure Dispose called even if exception thrown inside.

At a minimum, you need to do this in your logging code since EventLog needs to be Dispose()d. Seems like this resource could be reused rather than new-ed on every call. You could also consider using in your main loop for the ServiceController objects, to make your code more exception-safe.
private static void MyLogEvent(String Message)
{
// Create an eEventLog instance and assign its source.
using (EventLog myLog = new EventLog())
{
myLog.Source = "Informer Watchguard";
// Write an informational entry to the event log.
myLog.WriteEntry(Message);
}
}

This should be moved into the loop, since you don't want to keep a reference to old service handles for the life of your service:
ServiceController[] scServices = ServiceController.GetServices();
You also want to dispose of your reference to EventLog and to the ServiceController instances. As Artem points out, watch out for exceptions that are preventing you from doing this.
Since memory is going up every 10 seconds, it has to be something in your loop.
If memory goes up whether or not you write to the EventLog, then that is not the main problem.
Does memory used ever come down? Ie does the garbage collector kick in after awhile? You could test the GC's effect by doing a GC.Collect() before going back to sleep (though I'd be careful of using it in production).

I am not sure I understand the problem exactly. Is the service you are going to be monitoring always the same. It would appear from your code that the answer is yes, and if that is the case then you can simply create the ServiceController class instance passing the name of the service to the constructor.
In your thread routine you want to continue looping until a stop is issued, and the WaitOne method call returns a Boolean, so a while loop seems to be appropriate. Within the while loop you can call the Refresh method on the ServiceController class instance to get the current state of the service.
The event logging should simple require a call one of the static method EventLog.WriteEntry methods, at minimum passing your message and the source 'Informer Watchguard'
The ServiceController instance can be disposed when you exit from the loop in the thread routine
All this would mean you are creating fewer objects that need to be disposed, and therefore less likely that some resource leak will exist.

Thanks to all suggestions.
Finally the service is stable now with some modifications.
#Steve: I watch many services all beginning with the same name "Informer ..." but I don't know exactly full names, that's why I go this way.

Related

Reasons for long-running WCF service dying?

I have a long-running WCF service, hosted in IIS, that handles printing without any user interaction. After about 2 hours of looping through 1000 print jobs, the WCF service just seems to die. The log file I track indicates that the last print job was sent to the printer, but it never returns success nor failure (again, according to the log file).
From the log file, it would normally say:
2015-12-17 19:00:23,673 [27] INFO Barn.WCF.SysPrsPrintServer - Sending Print Job... PrintDocumentId=168;PrintSectionId=742;CustomerId=112702;DeliveryAddressId=474984;DocumentName=/SystemProcesses/Reports/CertificateOfRegistry;PrinterLocation=HB-MFP1;PrinterPaperSource=Cassette 3
2015-12-17 19:00:32,626 [27] INFO Barn.WCF.SysPrsPrintServer - PdfPrintHandler.Print: Printer HB-MFP1 indicates the print job is complete.
But the last log file entry is:
2015-12-17 19:00:32,688 [27] INFO Barn.WCF.SysPrsPrintServer - Sending Print Job... PrintDocumentId=169;PrintSectionId=742;CustomerId=112702;DeliveryAddressId=474984;DocumentName=/SystemProcesses/Reports/CertificateOfRegistry;PrinterLocation=HB-MFP1;PrinterPaperSource=Cassette 3
Can you see how this might be possible that I get no message back at all? Is it possible the WCF service just died or IIS recycled the app pool or something like that? So without further ado, here is my class:
using System;
using System.Collections.Generic;
using System.Drawing.Printing;
using System.IO;
using System.Threading;
using log4net;
using Aspose.Pdf.Facades;
namespace Barn.API.Print.PrintHandlers
{
public class PdfPrintHandler
{
private const int LargePdfByteCount = 3000000;
private readonly ILog _log;
private readonly ManualResetEvent _resetEvent = new ManualResetEvent(false);
public PdfPrintHandler(ILog log)
{
_log = log;
var license = new Aspose.Pdf.License();
var pathLicense = AppDomain.CurrentDomain.SetupInformation.ApplicationBase + "bin\\Aspose.Pdf.lic";
license.SetLicense(pathLicense);
}
/// <summary>
/// Prints the specified stream.
/// </summary>
/// <param name="stream">The stream.</param>
/// <param name="printerSettings">The printer settings.</param>
/// <param name="pageSettings">The page settings.</param>
/// <param name="timeout">The timeout.</param>
/// <param name="errors"></param>
/// <returns>The status of the print job.</returns>
public PrintJobStatusEnum Print(Stream stream, PrinterSettings printerSettings, PageSettings pageSettings, PrintDocumentModel printJob, int timeout, out List<string> errors)
{
errors = new List<string>();
// Reset the wait handler to make sure the event is not signaled
_resetEvent.Reset();
// Set attributes for printing
var viewer = new PdfViewer
{
AutoResize = true,
AutoRotate = true,
PrintPageDialog = false,
PrintAsImage = false
};
// Add an event listener when print job sent to printer
viewer.EndPrint += ViewerOnEndPrint;
//Print document using printer and page settings
try
{
_log.InfoFormat("Sending Print Job... PrintDocumentId={0};PrintSectionId={1};CustomerId={2};DeliveryAddressId={3};DocumentName={4};PrinterLocation={5};PrinterPaperSource={6}",
printJob.PrintDocumentId,
printJob.PrintSectionId != null ? printJob.PrintSectionId.ToString() : "null",
printJob.CustomerId != null ? printJob.CustomerId.ToString() : "null",
printJob.DeliveryAddressId != null ? printJob.DeliveryAddressId.ToString() : "null",
printJob.DocumentName != null ? printJob.DocumentName : "null",
printJob.PrinterLocation != null ? printJob.PrinterLocation : "null",
pageSettings.PaperSource.SourceName != null ? pageSettings.PaperSource.SourceName : "null");
// Print
if (stream.Length <= LargePdfByteCount)
{
// Bind the stream then print
viewer.BindPdf(stream);
viewer.PrintDocumentWithSettings(pageSettings, printerSettings);
}
else
{
// Use a more efficient printing method with larger documents
viewer.PrintLargePdf(stream, pageSettings, printerSettings);
}
// Block until the event finishes or timeout reached
_resetEvent.WaitOne(TimeSpan.FromMinutes(timeout));
// Check the print status
if (viewer.PrintStatus != null)
{
// An exception was thrown
var ex = viewer.PrintStatus as Exception;
if (ex != null)
{
// Get exception message
_log.Error("PdfPrintHandler.Print: Print Error: " + ex.Message + ex.StackTrace, ex);
errors.Add(ex.Message);
}
return PrintJobStatusEnum.Error;
}
else
{
// No errors were found. Printing job has completed successfully
_log.InfoFormat("PdfPrintHandler.Print: Printer {0} indicates the print job is complete.", printerSettings.PrinterName);
return PrintJobStatusEnum.Printed;
}
}
catch (Exception e)
{
_log.Error("PdfPrintHandler.Print Exception: " + e.Message + e.StackTrace, e);
errors.Add(e.Message);
}
finally
{
viewer.Close();
}
return PrintJobStatusEnum.Error;
}
private void ViewerOnEndPrint(object sender, PrintEventArgs printEventArgs)
{
// Signal the event is finished
_resetEvent.Set();
}
}
}
WCF Service Interface:
using System.ServiceModel;
namespace Barn.WCF
{
// NOTE: You can use the "Rename" command on the "Refactor" menu to change the interface name "ISysPrsPrintServer" in both code and config file together.
[ServiceContract]
public interface ISysPrsPrintServer
{
[OperationContract]
string DoWork(string data);
}
}
I would suggest checking the operation contracts of your print methods. I have seen WCF services fail out when the isOneWay attribute is not set or is set to false on a void method. try something like this
[OperationContract(IsOneWay=true)]
public void PrintDocumentWithSettings(PageSettings settings, PrinterSettings settings);
Here it is in the Windows event log:
A worker process with process id of '3236' serving application pool
'testws.mydomain.ca' was shutdown due to inactivity. Application Pool
timeout configuration was set to 20 minutes. A new worker process
will be started when needed.
So IIS seems to have thought that there was 20 minutes of inactivity because the WCF service is called asynchronously. The process is working away, but IIS decided to recycle the application pool in the middle of the process running.
I changed the settings in IIS on the application pool according to this:
https://serverfault.com/questions/333907/what-should-i-do-to-make-sure-that-iis-does-not-recycle-my-application

How can I log LoggingChannel LogMessages to the StorageFile?

I want to add logging of exceptions to my Windows Store App. Based on an idea from here, I've started off with this code in App.xaml.cs:
sealed partial class App : Application
{
private LoggingChannel channel;
private LoggingSession session;
/// <summary>
/// Initializes the singleton application object. This is the first line of authored code
/// executed, and as such is the logical equivalent of main() or WinMain().
/// </summary>
public App()
{
this.InitializeComponent();
this.Suspending += OnSuspending;
channel = new LoggingChannel("PlatypiChannel");
session = new LoggingSession("PlatypiSession");
session.AddLoggingChannel(channel, LoggingLevel.Error);
UnhandledException += Application_UnhandledException;
}
async private void Application_UnhandledException(object sender, UnhandledExceptionEventArgs ex)
{
ex.Handled = true;
String exceptionCornucopia = String.Format("Message0 == {0}; Message1 == {1}; HResult == {2}; Inner Ex == {3}; StackTrace == {4}", ex.Message, ex.Exception.Message, ex.Exception.HResult, ex.Exception.InnerException, ex.Exception.StackTrace);
channel.LogMessage(exceptionCornucopia, LoggingLevel.Error);
// not seeing how this saves the channel's logged messages...???
StorageFile logFile = await session.SaveToFileAsync(ApplicationData.Current.LocalFolder, "CrashLog");
}
As the comment indicates, it seems to me the last line simply saves a file named "CrashLog" to the LocalFolder. But how do the LogMessages get into that file? There is obviously a key piece missing here.
I know that this question has been open for a long time, but I just want to provide an answer for anyone else finding this.
The secret here is that the LogMessages are all written into the LoggingChannel, which itself has previously been registered with the LoggingSession:
session.AddLoggingChannel(channel, LoggingLevel.Error);
When the session is then saved to a file, it obviously knows about the associated channels and where to search for pending log messages.

Find if particular thread is associated with windows process

I have a java application which on execution runs different thread. I want to see if particular thread (By thread name) is running or not. I can check this manually using JConsole. It shows me whether that thread is running in associated process or not. By inspiring from that.
I am creating a C# program to get threads associated with particular process. I need to get name of threads in particular. I have tried toString method but it does not shows thread name.
Below is the code:
static class Program
{
/// <summary>
/// The main entry point for the application.
/// </summary>
[STAThread]
static void Main()
{
try
{
Process localById = Process.GetProcessById(6440);
ProcessThreadCollection coll = localById.Threads;
foreach (ProcessThread t in coll)
{
Console.WriteLine(t.toString());
}
}
catch (Exception e) { }
}
}
Output is :
System.Diagnostics.ProcessThread
System.Diagnostics.ProcessThread
System.Diagnostics.ProcessThread
Can anyone help me with this? Need to get name of all threads running in process by process ID.

What is the most efficient way to make an application handle hundreds of simultaneous file open requests from Windows Explorer in a single instance?

When I select a bunch of files in Explorer and right click open them (or press enter) I want all the files to be passed to a single instance of my application. I've used named pipes before to pass arguments from secondary instances to an existing global instance, but it seems like doing this for hundreds of program instances simultaneously (not to mention actually loading the application hundreds of times) is far from optimal. Is there a way to get explorer to concatenate the arguments on its own?
edit: I found a copy of the Paint.net 3.36 source code and it uses a memory mapped file to communicate between instances. That seems even more bloated than named pipes though (although it's not as likely to open hundreds of images for editing).
You can do it with shell extension.
Check out Creating Shortcut Menu Handlers
and
Create Namespace Extensions for Windows Explorer with the .NET Framework
I couldn't figure out the shell extension so I went with the named pipes. It seems to perform reasonably well, especially since there seems to be a limit in Windows 7 of how many files you can open with multi-select (although this will solve that: open-more-than-15-files-at-once-on-windows-7)
It took me forever to figure out how to get the timeout behavior to work (had to use a manual signal instead of AsyncWaitHandle). Hopefully this will save somebody some time:
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//
// The Original Code is the IDPicker project.
//
// The Initial Developer of the Original Code is Matt Chambers.
//
// Copyright 2011 Vanderbilt University
//
// Contributor(s):
//
using System;
using System.IO;
using System.IO.Pipes;
using System.Text;
using System.Threading;
using System.ComponentModel;
using System.Collections.Generic;
/// <summary>
/// A handler for an application to consolidate arguments from multiple instances
/// (within a Timeout period) into a single instance.
/// </summary>
public sealed class SingleInstanceHandler
{
/// <summary>
/// Occurs when the Timeout period has elapsed: the single instance is launched with the consolidated argument list.
/// </summary>
public event EventHandler<SingleInstanceEventArgs> Launching;
/// <summary>
/// Time to wait in milliseconds for additional instances to add their arguments to the first instance's.
/// </summary>
public int Timeout { get; set; }
/// <summary>
/// Constructs a handler for an application to consolidate arguments from multiple instances
/// (within a Timeout period) into a single instance.
/// </summary>
/// <param name="uniqueID">A unique string for the application.</param>
public SingleInstanceHandler (string uniqueID)
{
var rng = new Random(uniqueID.GetHashCode());
byte[] ipcMutexGuidBytes = new byte[16];
byte[] ipcNamedPipeGuidBytes = new byte[16];
rng.NextBytes(ipcMutexGuidBytes);
rng.NextBytes(ipcNamedPipeGuidBytes);
ipcMutexGuid = new Guid(ipcMutexGuidBytes).ToString().Trim('{', '}');
ipcNamedPipeGuid = new Guid(ipcNamedPipeGuidBytes).ToString().Trim('{', '}');
Timeout = 500;
}
/// <summary>
/// Launch a new instance using 'args' or consolidate 'args' into a recent instance.
/// </summary>
public void Connect (string[] args)
{
if (Launching == null)
return; // nothing to do
// create global named mutex
using (ipcMutex = new Mutex(false, ipcMutexGuid))
{
// if the global mutex is not locked, wait for args from additional instances
if (ipcMutex.WaitOne(0))
waitForAdditionalInstances(args);
else
sendArgsToExistingInstance(args);
}
}
private void waitForAdditionalInstances (string[] args)
{
var accumulatedArgs = new List<string>(args);
while (true)
{
var signal = new ManualResetEvent(false);
using (var pipeServer = new NamedPipeServerStream(ipcNamedPipeGuid, PipeDirection.In, -1, PipeTransmissionMode.Byte, PipeOptions.Asynchronous))
{
pipeServer.BeginWaitForConnection(x =>
{
// if timed out, stop waiting for a connection
if (signal.WaitOne(0))
{
signal.Close();
return;
}
pipeServer.EndWaitForConnection(x);
signal.Set();
}, null);
// no client connected to the pipe within the Timeout period
if (!signal.WaitOne(Timeout, true))
{
signal.Set();
break;
}
using (var sr = new StreamReader(pipeServer))
{
int length = Convert.ToInt32(sr.ReadLine());
for (int i = 0; i < length; ++i)
accumulatedArgs.Add(sr.ReadLine());
}
}
// new args have been added to accumulatedArgs, continue loop to listen for another client
}
Launching(this, new SingleInstanceEventArgs(accumulatedArgs.ToArray()));
}
private void sendArgsToExistingInstance (string[] args)
{
var pipeClient = new NamedPipeClientStream(".", ipcNamedPipeGuid, PipeDirection.Out);
// try to connect to the pipe server for the Timeout period
try
{
var sb = new StringBuilder();
sb.AppendLine(args.Length.ToString());
foreach (string arg in args)
sb.AppendLine(arg);
byte[] buffer = Encoding.ASCII.GetBytes(sb.ToString());
pipeClient.Connect(Timeout);
// can this ever happen? if it does, don't handle it like a timeout exception
if (!pipeClient.IsConnected)
throw new Exception("did not throw exception");
pipeClient.Write(buffer, 0, buffer.Length);
}
catch (Exception e)
{
if (!e.Message.ToLower().Contains("time"))
throw;
// no server was running; launch a new instance
Launching(this, new SingleInstanceEventArgs(args));
}
}
private string ipcMutexGuid;
private string ipcNamedPipeGuid;
private Mutex ipcMutex;
}
/// <summary>
/// Stores the consolidated argument list from one or more instances of an application.
/// </summary>
public sealed class SingleInstanceEventArgs : EventArgs
{
public SingleInstanceEventArgs (string[] args) { Args = args; }
/// <summary>
/// The consolidated argument list from one or more instances of an application.
/// </summary>
public string[] Args { get; private set; }
}
How to use it:
static void Main (string[] args)
{
var singleInstanceHandler = new SingleInstanceHandler(Application.ExecutablePath) { Timeout = 200 };
singleInstanceHandler.Launching += (sender, e) =>
{
Application.EnableVisualStyles();
Application.SetCompatibleTextRenderingDefault(false);
Application.Run(new MainForm(e.Args));
};
singleInstanceHandler.Connect(args);
}

Execute a method in an Application in an AppPool, using DirectoryEntry and Reflection?

Good Day
I am running a small web garden with StateSession cache. The obvious problem is that when you want to clear the cache, the cache is only cleared on the workerprocess that is handling the call.
In a specific AppPool there will be at least one User Interface Application and one Web Service Application. The Web Services do the majority of the caching, so they have methods to clear their cache.
What I would like to do is create a method that will get input (the AppPool Name) and then it will iterate through the current w3wp processes and get the required pools back. That can be done and I have methods of getting that information.
Where I am getting stuck is I have the AppPool Name as well as the Applications that are running in that Worker Process, but I am lost as to how to use that information to execute the "ClearCache()" method in a specific Web Service application.
I am sure that it can be done with Reflection, but I think I am missing something obvious.
Currently I am just using a console application to get something that works. This can then be moved to a better solution in due time.
Please advise if there is a way to use to current information to execute the required method.
Below is the test application as it stands at the moment.
Thank you.
Jaco
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Diagnostics;
using System.Management;
using System.DirectoryServices;
using System.Collections;
namespace CacheCleaner
{
class Program
{
const string defaultAppPoolMetabasePath = "IIS://localhost/W3SVC/AppPools/DefaultAppPool";
static void Main(string[] args)
{
//show a list of all the processes.
//GetListOfProcesses();
//Kill all the worker processes
//KillW3WP();
//Refresh the application pool
//RefreshAppPool(defaultAppPoolMetabasePath);
//get a list of all the Applications in the DefaultAppPool
//GetApplicationPoolInformation(defaultAppPoolMetabasePath);
//get the apps in the apppool
EnumerateApplicationsInPool(defaultAppPoolMetabasePath);
Console.WriteLine("\n\nEnd of process. Press Enter.");
Console.ReadLine();
}
/// <summary>
/// Show all the processes that are currently running on the system
/// </summary>
static void GetListOfProcesses()
{
foreach (Process p in Process.GetProcesses())
{
Console.WriteLine("{0} {1}", p.Id, p.ProcessName);
}
}
/// <summary>
/// This will kill ALL the worker processes.
/// Pretty much an iisreset call.
/// Not good for Production sites :)
/// </summary>
static void KillW3WP()
{
foreach (Process p in Process.GetProcessesByName("w3wp"))
{
Console.WriteLine("Closing " + p.ProcessName);
p.Kill();
p.WaitForExit();
if (p.HasExited)
{
Console.WriteLine("Closed");
}
else
{
Console.WriteLine("NOT Terminated");
}
}
}
/// <summary>
/// Refresh a specific application pool
/// </summary>
/// <param name="metabasePath"></param>
static void RefreshAppPool(string metabasePath)
{
using (DirectoryEntry applicationPool = new DirectoryEntry(metabasePath))
{
applicationPool.Invoke("Recycle");
}
}
/// <summary>
/// Get the Name of the worker process (AppPool Name)
/// </summary>
/// <param name="metabasePath"></param>
static void GetApplicationPoolInformation(string metabasePath)
{
var scope = new ManagementScope(String.Format(#"\\{0}\root\cimv2", Environment.MachineName));
var query = new SelectQuery("SELECT * FROM Win32_Process where Name = 'w3wp.exe'");
using (var searcher = new ManagementObjectSearcher(scope, query))
{
foreach (ManagementObject process in searcher.Get())
{
//get just the name of the application
var startIndex = process["CommandLine"].ToString().IndexOf("-ap ") + 5; //remove the -ap as well as the space and the "
var endIndex = process["CommandLine"].ToString().IndexOf("-", startIndex) - 2; //remove the closing "
var appPoolName = process["CommandLine"].ToString().Substring(startIndex, endIndex - startIndex);
var pid = process["ProcessId"].ToString();
Console.WriteLine("{0} - {1}", pid, appPoolName);
}
}
}
/// <summary>
/// Get the applications in the pool
/// From http://msdn.microsoft.com/en-us/library/ms524452%28v=vs.90%29.aspx
/// </summary>
/// <param name="metabasePath"></param>
static void EnumerateApplicationsInPool(string metabasePath)
{
Console.WriteLine("\nEnumerating applications for the {0} pool:", metabasePath);
try
{
DirectoryEntry entry = new DirectoryEntry(metabasePath);
if (entry.SchemaClassName == "IIsApplicationPool")
{
object[] param;
param = (object[])entry.Invoke("EnumAppsInPool", null);
foreach (string s in param)
{
Console.WriteLine("{0}", s);
//I am sure that I should be able to use this application name
//with Reflection to be able to execute a method...
}
Console.WriteLine("Done.");
}
else
Console.WriteLine("Failed in EnumerateApplicationsInPool; {0} is not an app pool", metabasePath);
}
catch (Exception ex)
{
Console.WriteLine("Failed in EnumerateApplicationsInPool with the following exception: \n{0}", ex);
}
}
}
}
There's a few ways to approach this, some bad, some good.
Bounce each worker process. This is pretty effective and will force a cache reload but will result in a very poor experience for end users because existing requests will be stopped in their tracks (if you're updating a database without any transaction protection then you could see your data go inconsistent). You'll also see a loss of session state if you're using InProc state management.
Recycle each pool - better than bouncing the pool because existing requests will still be permitted to complete. However you'll still see a loss of session state if you're using InProc state management.
Use a Cache Dependency - many caches are populated from a SQL Database. ASP.NET supports a feature called SqlCacheDependency. You can use this to refresh your cached data should the source data in the database change.
Use a file-based Cache Dependency - if your cache data is sourced from (for example) an XML data file then you can cache that data and create a cache dependency on this file. When the file is updated the cache is invalidated and you reload the data. You could automate this with a FileSystemWatcher because the CacheDependency class itself doesn't provide any change notification events and you'd need to check CacheDependency.HasChanged on every request.
With regard to calling ClearCache() directly on the cache object in each worker process, there is no direct way to do that using reflection. You might be able to do this using the .NET Debugging API's and inject some code to clear the cache but you're talking about walking up to a process, attaching to it and pretending to be a debugger. I wouldn't really fancy writing that myself when there are other mechanisms such as Cache Dependencies available.

Categories