Using Workflow

Trying to write a workflow to ping servers from a text file
Syntax is fine I think

Using a function it takes 30 Minutes to ‘ping’ 4000 Servers using measure-command cmdlet
using a workflow it takes only 1 second to run

Seems it is not running the foreach loop

How do I know the servers that pinged correctly and the ones that did not
since you cannot use write-host in a workflow
Powershell version is 4.0 Windows 2008 SP2
Here is the script

$Servers = Get-Content D:\PowerShell\Servers.txt
workflow WFConnection {
foreach -parallel ($Server in $Servers){
Test-Connection -ComputerName $Server -Count 1 -ErrorAction silentlycontinue

You’ve got a scope problem. $servers is created outside the workflow. While in a normal PowerShell script that would work, it doesn’t work with workflow. The WWF engine only gets what’s inside the workflow construct. You should parameterize the workflow and pass in the computer names as a parameter. As-is, $servers inside the workflow is empty, so the foreach doesn’t execute.

You’re also not really capturing your output (from Test-Connection) or doing anything with it. Not sure if that was your intent or not. If you want to keep track of which servers respond to a ping, you’d need to put that output someplace - a file maybe, but it’s up to you. I might also use the -quiet switch of Test-Connection, since that just returns a true/false. If you get true, write the server name to one file. If you get false, write to another file. Or whatever you need.

Ah OK , Thanks Don

Now I do I write to a file within the workflow
Cannot find how to do this in any of my books (I have many including yours) or on the internet

{inlinescript write-host|out-file $Server d:\up.txt -append}

but get errors

While I am here any PowerShell 4.0 books in the pipeline?

Thanks again


It’d be nice to know what the errors are ;). In general it’s fine to write to a file from a workflow except that (a) you can’t redirect Write-Host like you’re doing, that’s why we always tell people to not use Write-Host, and (b) you have to pay attention to the paths and permissions since this is bring run by WWF.

Stop using Write-Host. Always.

We’re updating PowerShell In Depth for 4.0, and I may do a DSC book with Steven Murawski, but nothing’s firm on that.

Yes I agree just read your article on write-host
I am right to assume that write-verbose would work



When you use the pipeline character, you’re only able to pass along the “main” pipeline in PowerShell.

Write-Host "This" | Out-File nothing.txt
Write-Output "That" | out-File something.txt

The second will work. Write-Verbose doesn’t place information into the main pipeline; it, -Warning, -Debug, and -Error all have their own pipelines. There’s a syntax that lets you switch info from one pipeline to another, but in this case, Write-Output is what you want. You can also just do

"Message" | Out-File works.txt

Since Write-Output is the default cmdlet; that syntax will put “Message” into works.txt.

OK Thanks Don

I will give that a Go
Happy Christmas

Writes to a file with the append switch but get the following error below even though the list of servers are in the file up.txt - throttle limit?

Microsoft.PowerShell.Utility\Write-Error : The process cannot access the file ‘D:\up.txt’ because it is being used by another process.


No, you’re dealing with simultaneous file access. Remember, you’ve got multiple copies of this all running at once. They can’t all write to the same file at the same time. That’s a basic tenet of the Windows file system. Two processes can’t both access the same file object at once, unless they take special precautions (which, for writing to a simple text file, isn’t an option).

This is, by the way, one of the more common things you run across when you start parallelizing tasks - a better approach would perhaps be to log the information to a SQL Express database, since SQL itself supports multi-user access. However, that’s a lot of extra work. That’s the thing with Workflow, though - you start to get into “a lot of extra work” pretty quickly because of the underlying complexities of parallel processing.

You could also have each workflow instance log to a separate file, wait until they’re done, and then concatenate them all into a single file. Again, more work.

Figure I’ll contribute with some actual, working code (unless I misunderstood the requirements). I had been putting off learning workflows for a good while, but started looking into it now. I started with reading Richard Siddaway’s article on Technet about the basics:

After quite a bit of experimentation, failures and learning, I cooked up this, which works:

Workflow Test-MultiConnection {
    param([string[]] $ComputerName)
    foreach -Parallel ($Computer in $ComputerName) {
        $Obj = New-Object -Type PSObject -Property @{
            ComputerName = $Computer
            Online       = Test-Connection -ComputerName $Computer -Count 1 -Quiet


$ComputerName = Get-ADComputer -Filter * | Select -ExpandProperty Name
Test-MultiConnection $ComputerName | Select -Property ComputerName, Online | Format-Table -AutoSize

In my lab env. these are the results:

PS C:\temp> .\Test-MultiConnection.ps1

ComputerName  Online
------------  ------
WIN2012R2       True
WIN8ESXI        True
SERVER2012      True
WIN8VM         False
SERVER2012RC   False
SS-WIN7         True
VMWAREWIN7     False
WINXPSSD        True
esxi           False
SERVER2008      True
XPTANKET       False
SERVER2003     False
VISTA64ESXI    False
WIN2K           True
WINXPESXI      False
WIN7ESXI        True
SIEMENS        False
SRV2003R2ESXI   True
2008R2ESXI      True
2008R2ESXI2     True

If you remove Format-Table and instead pipe to Export-Csv, you can easily create a CSV file with the results. Or whatever you want. You can also toss in a Sort-Object on ComputerName or Online value. If you assign the results to a variable and process it (twice is probably easiest, shouldn’t be much overhead) with Where-Object, you can also create one file for offline computers and one for online, by filtering on the Online attribute being “True” (or “False”) or not.

Experimenting some more, I came across a scenario when the number of computer increases from 18 to over 2000, where I got the following error:

The operation did not complete within the allotted timeout of 00:00:30. The time allotted to this operation may have
been a portion of a longer timeout.
At Test-MultiConnection:10 char:10

I searched the web and found this article:

I also found a reference to a no-longer-existing article by someone who wrote a “Foreach-Parallel” cmdlet. I think the correct way to deal with this is via the New-PSWorkflowSession cmdlet where you can customize throttle limit, concurrent sessions, etc.

I couldn’t figure out how to apply the settings to a specific workflow (maybe it affects the current PS session all together?), so I wrote a little function/cmdlet that I’ve found myself needing from time to time. This won’t perform well on large collections due to, among other things, array concatenation, but should be fine for many purposes. I probably should have figured out the correct way, but now I’m posting it anyway. :expressionless:

I called it Split-Collection and the code looks like this:

function Split-Collection {
        [Parameter(ValueFromPipeline=$true)] $Collection,
        [Parameter(Mandatory=$true)][ValidateRange(1, 247483647)][int] $ChunkSize
    begin {
        $Ctr = 0
        $Arrays = @()
        $TempArray = @()
    process {
        if (++$Ctr -eq $ChunkSize) {
            $Ctr = 0
            $Arrays += , @($TempArray + $_)
            $TempArray = @()
        $TempArray += $_
    end {
        if ($TempArray) { $Arrays += , $TempArray }

Its behavior can be demonstrated like this:

PS C:\PowerShell> 1..10 | Split-Collection -ChunkSize 2 | %{ $_ -join ', ' }
1, 2
3, 4
5, 6
7, 8
9, 10
PS C:\PowerShell> 1..10 | Split-Collection -ChunkSize 3 | %{ $_ -join ', ' }
1, 2, 3
4, 5, 6
7, 8, 9
PS C:\PowerShell> 1..10 | Split-Collection -ChunkSize 5 | %{ $_ -join ', ' }
1, 2, 3, 4, 5
6, 7, 8, 9, 10
PS C:\PowerShell>

Then I did this, processing 50 at a time, which, as intended, seems to work around the issue:

PS C:\PowerShell> $Results = $Computers[0..300] | Split-Collection -ChunkSize 50 | %{ Test-MultiConnection $_ }
PS C:\PowerShell> $Results.Count
PS C:\PowerShell> $Results | select -first 3 | ft -a computername, online

ComputerName Online
------------ ------
comp1       False
comp2       True
comp3       False

Shrug. If someone can fill in the blanks on the throttling/session limits, it’d be nice.

I can’t seem to edit the last post anymore, so I’m just throwing in a link to a tiny article about the Split-Collection cmdlet in a new comment:

This works nicely on 400 or so computer objects;
Move Test-Connection outside of creating the object and put the commands in a sequence block.

Workflow Test-MultiConnection {
param([string[]] $ComputerName)
foreach -Parallel ($Computer in $ComputerName) {
sequence {
$online = Test-Connection -ComputerName $Computer -Count 1 -Quiet
$Obj = New-Object -Type PSObject -Property @{
ComputerName = $Computer
Online = $online
Test-MultiConnection $Computers | Select -Property ComputerName, Online | Format-Table -AutoSize