This monitors external mail relays, tracks the round trip time of emails to the relay and back, and writes event log messages if the messages do not return in the configured time interval, if you want to use SCOM to alert on congestion or failures.

I'm considering using a started task to stop it and re-start it within a maintenance window, but I still need to figure out the best way to insure that it gets restarted if the server re-boots, and that only one instance of it runs at a time.

If anybody sees something that could be done better, let me know.

$check_interval = 5
$from_addr = ""
$to_addr = ""
$relays = "smtp1.relaydomain.com","smtp2.relaydomain.com"
$hubservers = "",""
$edgeservers = @{"" = "Edge1";"," = "Edge2"}
$headings= '"Timestamp","Source","Relay","Edge_Recv","Hub_Recv","Trip_Time_seconds","Msg_Subj"'
$outfile = $(get-date).tooadate().tostring() + "_mtr.csv"
ac $outfile $headings

$script_start = get-date

#Create event log source and log start event in the application event log
if (!([System.Diagnostics.EventLog]::SourceExists("Mail_Monitor"))){[system.diagnostics.eventlog]::createeventsource("Mail_Monitor","Application")}
$app_log = new-object system.diagnostics.eventlog ("Application",".")
$app_log.source = "Mail_Monitor"
$app_log.writeentry("Mail Monitor script started.","Information",10220)

$sent = @{}

function sendemail {
####   Send email
foreach ($relay in $relays){
    $from = "$from_addr"
    $to = "$to_addr"
    $subj = $subj_ts + $relay
    $body = get-date | out-string
    $SmtpClient = new-object system.net.mail.smtpClient
    $SmtpClient.Host = $relay
    $mailmessage = New-Object system.net.mail.mailmessage
    $mailmessage.from = ($from)
    $mailmessage.To.add($to)
    $mailmessage.Subject = $subj
    $mailmessage.Body = $body
   
    $smtpclient.Send($mailmessage)
    if (!($?)) {$app_log.writeentry("Error sending email $($subj) to $($relay) ","Warning",10222)}
    else {$sent.add($subj,$(get-date).tostring())}
    }
}

while ($true){
$sentby = [environment]::machinename
$subj_ts  = $(get-date).tooadate().tostring() + " Sent by " + $sentby + " To relay "
sendemail
$log_start = $(get-date).tostring()
sleep -seconds $($check_interval * 60)
$log_end = $(get-date).tostring()

$msgtrk = @()
$msg_returned = 0

foreach ($hubserver in $hubservers){
$msgtrk += get-messagetrackinglog -Sender $from_addr -Recipients $to_addr -Server $hubserver -EventID "RECEIVE" --Start $log_start -End $log_end
}

if ($msgtrk.count -ge 1){
$msgtrk = $msgtrk | sort timestamp
foreach ($msg in $msgtrk) {

if ($sent[$msg.messagesubject] -and $msg.source -eq "SMTP"){
    $msg.messagesubject -match "^(\d{5}\.\d{1,10}?)\sSent\sby\s(.+)\sTo\srelay\s(.+)" > $nul
    $recvhost = $msg.serverhostname
    $recvedge = $msg.clientip
    $sent_ts = $matches[1]
    $mailrelay = $matches[3]
    $source = $matches[2]
    $trip_time = [timespan]::fromdays($msg.timestamp.tooadate() - $sent_ts)}
    $sent_time = [datetime]::fromoadate($sent_ts)
    $sent.remove($msg.messagesubject)
   
   
    $event_text = "Email round trip time for $mailrelay at $sent_time is $($trip_time.totalseconds) seconds."

if ($trip_time.totalminutes -lt $check_interval){$app_log.writeentry($event_text,"Information",10221)}
if ($trip_time.totalminutes -ge $check_interval){$app_log.writeentry("$event_text","Warning",10221)}

$outstr = $msg.timestamp.tostring() + "," + $source + "," + $mailrelay + "," + $edgeservers[$recvedge] + "," + $recvhost + "," + $($trip_time.totalseconds)+ "," + $msg.messagesubject
ac $outfile $outstr
}
}
if ($sent.count -ne 0) {$app_log.writeentry("External email checks are overdue. `n $($sent) ","Warning",10222)}
if ($sent.count -eq 0) {$app_log.writeentry("All external mail checks have been returned. ","Information",10222)}
}

I've not reviewed the script. I thought WMI could say what was running like a script, but it appears not, at least on XP.

I don't quite understand what you're saying about making sure the script starts on reboot.

One approach for figuring out whether the script is already running is to use handle.exe from www.sysinternals.com. I can provide more details if you're stuck...

I'm running the script on one of the Hub Transoport servers (Serrver 2003 x64), so I'll check on the WMI.

I'd like for it to quit, and then restart sometime during the maintenance window to start a new event file, and free up memory.

We run SCCM, so the server may get rebooted after an unattended patch installation, and the script needs to restart after a reboot. I think I can do that with a schtask command in a machine startup script, but I don't want to get one instance started by a reboot, and another one started by the scheduled task hitting it's run window.

If it starts running multiple instance of the script, it starts logging multiple events in the appliation logs, and SCOM gets confused so I need to keep it single threaded.

Ah, yes, you will likely be doing something like this:
powershell.exe -command "& {& 'C:\Documents and Settings\user\loop.ps1'}"

In that case:
PS > gwmi win32_process|?{$_.processname -eq "powershell.exe"}|%{$_.commandline}
"C:\WINDOWS\system32\windowspowershell\v1.0\powershell.exe" -sta -noprofile
powershell.exe -command "& {& 'C:\Documents and Settings\user\loop.ps1'}"

Above, I have one powershell.exe console, and I just called powershell.exe from DOS. Notice that the full command-line shows so I can see that I have loop.ps1 running already.