How can I monitor whether an application is running on Windows Server 2008?

Question

We're mostly a Linux shop, but we have a print server running Windows Server 2008, and use an application called Print Helper to print invoices. I need to find a way to check that this application is running, and automatically restart it if it's not.

On Linux I'd probably do this with a small shell script and a cron job, but I'm not sure how to accomplish this on Windows Server. I'm fairly confident I could do it in Perl using Proc::Background, but I'm reluctant to install Perl just for one script, and while I'm sure it can be done using something like PowerShell, it's not really a worthwhile use of my time to learn PowerShell for one small task. Tasklist seems to do some of what I want in that it can let you know if a specific process is running or not, but I'm not sure how I could go from there to automatically restarting the application if it's crashed.

Any help gratefully received!

Answer 1

Wow. I just answered a different question.

What you should do is "daemonize" the Print Helper process by using sc.exe or srvany.exe (which I believe is depreciated). Then you can utilize the builtin service "Recovery" options to handle failure events (including, I believe, executing a script, sending an snmp trap, sending an email, and, of course, restarting the service).

As given:

sc create printhelper binpath= "c:\program files\Print Helper\phelper.exe" start= auto depend= Spooler/lanmanserver DisplayName= "Print Helper"

This will create a service named printerhelper, with a display name of Print Helper, executing "c:\program files\Print Helper\phelper.exe" automatically, with the dependencies of the Print Spooler and SMB/CIFS Server service, running as NT AUTHORITY\SYSTEM builtin user.

Answer 2

If you're using Nagios, you can do it with NSClient++ addon.

On monitoring server, define a service:

define service{
   use                 generic-printer
   host_name           hostname
   service_description appname
   check_command       check_nt!PROCSTATE!-d SHOWALL -l appname.exe
   contact_groups      admin-sms
   event_handler       autostart_appname!hostname
   }

The autostart_appname is defined in commands.cfg:

define command { 
    command_name    autostart_appname
    command_line    $USER1$/eventhandlers/autostart_appname.sh $SERVICESTATE$ $SERVICESTATETYPE$ $SERVICEATTEMPT$ $HOSTADDRESS$
}

The event handler script autostart_appname.sh:

#!/bin/sh

HOSTADDRESS=$4

case "$1" in
OK)
    ;;
WARNING)
    ;;
UNKNOWN)
    ;;
CRITICAL)
    case "$2" in
    SOFT)
    ;;  
    HARD)
        /usr/local/nagios/libexec/check_nrpe -H $HOSTADDRESS -c autostart_appname
        ;;
    esac
    ;;
esac
exit 0

On the Windows server, define a command in NSC.ini:

[NRPE Handlers]
autostart_appname=C:\Program Files\NSClient++\scripts\autostart_appname.cmd

and batch script is simple like that:

net start "Application Name"

Answer 3

It sounds like you want to spend as little time as possible on this task. We all hate print servers equally. Using only the preferences built into Windows you can create a service, automatically recover the service, and implement rudimentary monitoring of the service.

Step 1 Create a Service

(You can skip this step if there's already a service entry in the Services control panel)

Use sc.exe to create a new service.

sc.exe create PrintHelper start=auto binPath="C:\<Print Helper Path and Flags>" DisplayName="Print Helper"

Step 2 Configure your service

In the services control panel right click on your new service select properties and choose the Recovery tab. Configure as needed.

There are many options here, what to do on the first failure, second failure, you can even run another application on failure.

Step 3 Monitor Your Service

Either configure the Event Viewer to send an email on the specific event failures for Print Helper (find the event right click on it, Select Attach Task for this event).

Or configure the Service to send an email on failure, using an external program as the recovery mechanism.

Disclaimer

Of course I would recomend using something like nagios or SCOM for a real monitoring solution. But that's an entirely new problem in itself.