please excuse my lack of knowledge. im looking for some guidance. I have numerous scripts I have made that do one thing or another… ive gotten some great knowledge re; powershell and now im looking to improve on my scripts.
the script im working with at the moment is working well. it copies files to a list of remote computers and then installs an application via a msi installer. I do test-path commands to make sure the directory has been copied correctly, installation ran correctly by making sure the new application has a path and the executable is there, and then I remove the files from the pc after its done. the whole while im sending feedback back to my console so I know its being done. what im looking to do is make a report, perhaps venture into html reports, but I need to clean up the process of having confirmation that the install ran correctly by doing if statements with test-path commands, in effect making the script better I think. so for example the following code;
foreach ($computer in $computers) {
$filepresentA = invoke-command -computer $computer -scriptblock {test-path c:\install-apphere\appname.bat}
$filepresentB = invoke-command -computer $computer -scriptblock {test-path “c:\install-appname\app.msi”}
$installapp = {c:\install-appnamehere\app.bat}
if (($filepresentA )-and ($filepresentB))
{
# some kind of output to generate report confirming install directory successfully copied on a pc by pc basis?
“$computer contains app.bat and app.msi… starting install”
invoke-command -computername $computer -scriptblock $installapp
}
else
{
# some kind of output to generate report confirming errors with install directory on a pc by pc basis?
“$computer does not contain any files… there was an error… application not installing”
}
}
how would I go about adding to this to make this possible. ive thought of creating csv files with each “computername” and then titles like “installed correctly”, and “cleanup successful”. also the creation of custom objects, but I am not able to come up with the way to do this…much less , the proper way to go about this… im lacking the logic, ive researched ways to possibly go about this but need direction.
im sorry for the confusion. let me try to clarify.
im looking to have “feedback” regarding lets say… installing an application(via msi) in a script.(keep in mind this script is running against a list of 30 computers)
I copy some files over to a remote computer to a temporary directory, then call the bat file to get the msi install started.
because the script has multiple “stages”,
I perform checks to make sure it (1)copies over correctly, (2)the msi started and ran successfully (leaving behind a log file of the install),(3)the newly installed program has now got a new directory structure that wasn’t there before), and(4) the install directory structure has been removed.
all this works by outputting info to the screen… which I know is a big no-no… but Im new at this. im getting things done…its not pretty…but more importantly. im looking to make it better, and grow my knowledge.
I realize im way over my head in terms of what my current Powershell knowledge is now, but im looking for guidance/direction on how to go about making it better. im thinking an HTML report ,formatted nicely, would be nice.
to run the script, then take feedback from my checks, and instead of simply writing output to the console that pc1,pc2,pc3,…etc all went through the 4 checks and a successful outcome has taken place, turn in to a nicely formatted report.
from the reading ive been doing… im leaning towards writing output to a hash table perhaps? I may be way off in my logic, and this may not even be possible…, or im going about it all wrong with improper logic.
im not looking for a script…, except perhaps a possible example to help push me in the right direction… im looking for guidance…ive no problem with reading about various things… ive gone from no knowledge regarding powershell… to numerous scripts that install applications such as google chrome on remote computers. again. not pretty, but its gets things done. im looking to improve. again… your thoughts are very much appreciated.
example;
pc-name check 1 check 2 check 3 check4
pc-1 yes yes yes yes
pc2 yes yes no no
pc-3 no no no no
I don’t think a hashtable would be the best approach. If your plan is to produce an HTML table, you’d want to create a custom object, and give it properties for each column you want the table to have.
ok… will look in that direction. thanks very much for your input… gonna be out on vacation for a couple weeks, so will post back then after ive messed with it a bit.
ok… ive got this working where as the first computer called from the get-content command ( of which there are 2 computers in the list at the moment) works correctly, but it doesn’t seem to loop to the next computer. what am I missing? what am I not understanding? I thought I had to make this an array. is my thinking wrong?
If you are just connecting via UNC, you might want to consider adding a Test-Connection to validate the computer is reachable before you attempt to connect via UNC path. Your “No” is supposed to indicate a path doesn’t exist, not a computer that is offline.
ok…
I will add the test-connection, thanks for the suggestion. i believe I actually have that in the beginning stage of this script where I actually install chrome… I left that portion out thinking it wasn’t part of the cause of the issue im having seeing that its working correctly.
I fixed the loop ending prematurely. (where you marked #loop ends here) by removing “}” and putting at the end, but the script was still running that for me correctly by showing me a yes or a no in that field.
I am still having the issue where the 2nd of 2 computers in the list is showing up. the first computer in my get-content call doesn’t seem to register.
It’s personal preference, but I don’t really use Add-Member unless I’m adding a column or manipulating object properties that aren’t available with this method (e.g. Alias, PSStandardMembers, etc.). Since the Add-Member requires a -Force parameter or you’ll get errors that the property exists, it feels…well forced and klunky…IMHO
thanks for the assistance… will look into this… I noticed the “+=” combo during my research into how to do this, but never tried it in this version of the script script. this was the closest I got to success in all my experiments. will also look into your other suggestions. im always open to others opinions as I do not know everything and im sure there are better ways of doing what im trying to accomplish.
thanks so much for the assistance.
ok… I got this to work… but ive noticed sometimes the (properties) columns move position. how can I define “computer name” takes the first position, and “Install Directory Deleted?” takes the second position, and so on?
If you need your script to support PowerShell 2.0, then you can use the Select-Object cmdlet to reorder the properties, as Rob pointed out. If you don’t need 2.0 compatibility, I prefer to use either the [pscustomobject]@{} or [ordered]@{} syntax:
There are a couple of options. Typically I just assign the results of a loop or function to a variable, and if the loop / function happens to produce more than one object, PowerShell automatically converts it to an array. (If you want it to be an array with zero or one elements, you can wrap the loop / function call in @() ). Example modification of the OP’s code:
Sometimes it’s more flexible to have finer control of the array that you’re building. In those cases, I like to use List or ArrayList; they perform much better at scale by keeping the number of array copies to a minimum.
$computers = @(Get-Content "C:\scripts\class-lists\test-lab-computers.txt")
$ChromeOutcome = New-Object System.Collections.ArrayList
foreach ($computer in $computers) {
$test_file = (get-item -ErrorAction SilentlyContinue "\\$computer\C$\program files\google\chrome\Application\master_preferences").LastWriteTime
$CorrectDate = "friday, april 26, 2013 10:39:56 am"
$File_Present = test-path -ErrorAction SilentlyContinue "\\$computer\C$\program files\google\chrome\Application\master_preferences"
$InstallDir = test-path -literalpath \\$computer\C$\install-chrome
$AppDir = invoke-command -computername $computer -scriptblock {test-path -path "C:\Program Files\Google\Chrome\Application\chrome.exe" }
$MasterPref = (( $File_Present ) -and ($test_File = $CorrectDate ))
if ($InstallDir -eq $true) {$variable1 = $false}else{$variable1 = $true}
if ($AppDir -eq $false) {$variable2 = $false}else{$variable2 = $true}
if ($MasterPref -eq $false) {$variable3 = $false}else{$variable3 = $true}
$object = [pscustomobject] @{
"Computer Name"=$computer
"Install Directory Deleted?"=$InstallDir
"Application Directory Present?"=$AppDir
"Correct Master_Preference File is Present?"=$MasterPref
}
$null = $ChromeOutcome.Add($object)
}
# In most cases you can just output $ChromeOutcome here; PowerShell will enumerate its elements just like an array. If you really
# need to convert it back to an Object[] type, use $ChromeOutcome.ToArray()
$ChromeOutcome
thank you gentlemen for the comments. im trying to learn as much as possible while creating and using scripts that make my life easier. but also… keep in mind… my end plan is to turn this into an HTML report. so being easy to read is important so the column headers were made to look like " Computer Name" as opposed to “ComputerName” and so forth. Version 2 compatibility is a good thing due to most of the PC’s this may run against may still be version 2. we are reimaging computers to bring them up to ver 4.
so with HTML reporting in mind, are any of your suggestions geared better towards that? and also. considering this may run against 30 computers at a time, are there any performance benefits one way or the other?
Once again, thank you for the suggestions. it gives me multiple directions to research.
You can change the column names at any point before you “present” the data. For instance:
#Use a calculated expression
$processes = Get-Process | Select ProcessName, @{Label="The name of the process that is running!";Expression={$_.ProcessName}}
#Add a alias to the property
$processes | Add-Member -MemberType AliasProperty -Name "Another Really Long Property" -Value ProcessName
#Say you want to generate a report and are going to filter the results to
#only show computers that only have WINWORD running (or Install Dir deleted)
#I'm just saying (personally), I rather NOT do ...
$processes | Where{$_."Another Really Long Property" -eq "WINWORD"}
#versus this:
$processes | Where{$_.ProcessName -eq "WINWORD"}
#For instance, if I wanted a HTML fragement:
$frag = $processes |Where{$_.ProcessName -eq "WINWORD"} | Select @{Label="The name of the process that is running!";Expression={$_.ProcessName}} | ConvertTo-HTML -Fragment
HTML reporting is fine, but Powershell is going to gather all of the data. You need to ensure you have all the data you need from the computers first before reporting it with HTML. If you are going to be running anything against multiple computers, you should leverage Powershell Jobs. Although 30 computers is not a lot, you might see some performance gain running it as a job.
BTW, reimaging a computer just to put Powershell V4 on computers…you guys should look at hiring a application packager.
Last notes. This does nothing:
if ($InstallDir -eq $true) {$variable1 = $false}else{$variable1 = $true}
if ($AppDir -eq $false) {$variable2 = $false}else{$variable2 = $true}
if ($MasterPref -eq $false) {$variable3 = $false}else{$variable3 = $true}
In the example I provided, the variable (e.g. $variable1) was referenced when building the object which you are not doing, so it can be removed. Second, why are you testing a path with UNC on the first line but using Invoke-Command to test another path? You should consider updating the second check to UNC or making the paths local ( “C:\Program Files\Google\Chrome\Application\chrome.exe” ) and then running your script with: Invoke-Command -ComputerName $computer -File “C:\Script\Get-ChromeStats.PS1”
"BTW, reimaging a computer just to put Powershell V4 on computers…you guys should look at hiring a application packager. "
we are a school district… reimaging is almost a yearly occurrence due to how messed up they get over the course of the year. so its a chance to make updates, tune them up, etc.
the “Install-chrome” directory structure is a folder I copy over to each pc and run the install local from there. once the install is complete… I want to clean up the directory and not leave it there… so the check is to confirm I have indeed successfully cleaned up the directory after the install…i have since put in firthe checks to make sure it isn’t being deleted until the install has proven successful.
