Archive

Tech

It’s been far too long and way too much has been going on. I took a new position back in February with our corporate office with the lofty goal of migrating 15,000 endpoints into a centralized Jamf Pro instance. We worked with Jamf and with our internal resources to determine where to place the infrastructure and finally settled on AWS. I’ll go into more detail in other posts because for now, I wanted to talk about sideloading packages in S3.

Most of us have dev or test environments. Ours is in AWS alongside our production environment. With the release of Jamf Pro 10 this past week, it meant it was time to upgrade our dev instances to 10. Not only do we host the servers in AWS, but we also host the distribution point for our dev instance on S3. As you may be aware, each time you create a new cloud distribution point with Jamf Pro a new S3 bucket is spun up. That’s fine, but what if you have a bunch of test packages already in a dev S3 bucket that you want to use? That’s where I found myself this evening, and I figured out how to move those into the new S3 bucket that Jamf Pro created.

After spinning up your new Jamf Pro servers and creating your S3 bucket in Jamf Pro, head over to the AWS Console to manage your S3 buckets. Locate your original bucket and select all of the files in the bucket.

Next go under the More button and choose Copy.

Now go find the new bucket that was created by Jamf Pro and go under the More button in the bucket and choose Paste.

After getting the files over, we now have to tell Jamf Pro that the packages are there. Using the AWS CLI from your computer, grab a listing of your bucket:

Now that you’ve got a list of all packages in that S3 bucket, you’ll want to get rid of all of the extra data so that you only have the package filename. I used Excel to delete columns and combine columns if there were spaces (use the Open menu in Excel and choose “Delimited” and use Spaces as your delimiter). Once you have a clean list, save that out of Excel and then open it in your favorite text editor. TextMate is my go to for this. You’ll want to save the text out as a CSV with LF only.

We now have a clean list that we can send through a handy API script to stub out the packages in Jamf Pro. We will use the following script to import this data into the Jamf Pro server. Be sure to edit the JSS address in the script before running.

Let’s open up terminal and run our script, feeding it our cleaned up text file.

That’s it. As that script runs it will add the packages to the Jamf Pro server so you don’t have to. This will help speed up the process of creating new dev instances each time.

In some of my next posts I’ll discuss how we are using the Server app on machines as distribution points, and utilizing simple scripts with LaunchDaemons to keep them in sync.

At our recent Dallas area Casper User Group meeting, we got into a discussion around collecting data during a Casper recon. Specifically we were discussing the use of Extension Attributes to collect information about virtual machines.

Extension Attributes are a way to capture information from your systems. You can use scripts to pull information or drop downs or text boxes to store static information in the database. In the instance of collecting info about virtual machines, a script would be run on the systems during the recon to gather the information. Running a script on the system each time a recon happens can be processor heavy, depending on the data that is being gathered. For example, gathering home folder size by running “du” each time a recon happens can be taxing.

Rather than run the script each recon, you can use a policy to run the script once a week, once a month, or just one time, to gather the information you need and place it in a plist file somewhere. During the standard recon period, you can then use an Extension Attribute to read the information in that plist file. This is much less taxing on the systems than running the script during a recon.

Stash The Data

For our example, rather than run through grabbing info about virtual machines, let’s work on grabbing the home folder size for the logged on user. We will store the info in a plist file that we will stash in /Library/IT_Data.

First we need to find the logged in user name. There are plenty of ways to do this, but we’ll use the “Apple approved” method, using a Python one liner. Okay, it’s not really a one liner, it’s just built like one.

Now that we have our logged in user, we just need to find the user’s home folder and use du to grab the data. We’ll use dscl to grab the home folder location and then du to get the home folder size.

The next thing we need to do is to store this information into our plist file. Using the defaults command, we can write as much data as we want into the plist file, We can use different keys to store whatever data you want, and then recall it during recon by asking for those specific keys.

Retrieve The Data

Now that we have the data stashed away, we just need to grab it during the recon process. To do this, we’ll use the defaults command again, to grab the data. We’ll use some variables for the folder path and the plist name, that way we can re-use this code fairly easily. We also want to make sure the file actually exists before trying to read data from it, hence the If statement.

Once we’ve read the data, all that’s left is to echo it out into the EA.

That’s All Folks

That’s pretty much all there is. Now that we know how save data to a plist and then read it back, this method could be used for any data we only need to gather once, or gather at infrequent times.

The full script to write the data and to then read the data are below.

A few years ago I was searching for a way to easily create bookmarks in Microsoft Remote Desktop 8 on the Mac. Prior to version 8 you could drop an .RDP file on a machine and that was really all you needed to do to give your users the ability to connect to servers. Granted, you can still use this method, it’s just a bit sloppier, in my opinion.

So I went searching for a way to script the bookmarks, and that led me to my good friend Ben Toms’ (@macmuleblog) blog. I found his post, “HOW TO: CREATE A MICROSOFT REMOTE DESKTOP 8 CONNECTION” and started experimenting. After some trial and error, I discovered that using PlistBuddy to create the bookmarks just wasn’t being consistent. So I looked into using the defaults command instead. I finally was able to settle on the following script:

You can find that code in my GitHub repo here.

RDC URI Attribute Support

I had posted that script up on JAMF Nation back in June 2014 when someone had asked about deploying connections. Recently user @gmarnin posted to that thread asking if anyone knew how to add an alternate shell key to the script. After no response, he reached out to me on the Twitter (I’m @stevewood_tx in case you care). So, I dusted off my script, fired up my Mac VM, and started experimenting.

The RDC GUI does not allow for a place to add these URI Attributes. I read through that web page and Marnin forwarded me this one as well. Marnin explained that he was able to get it to work when he exported the bookmark as an .RDP file and then used a text editor to add the necessary “alternate shell:s:” information. Armed with this knowledge, I went to the VM and started testing.

First I created a bookmark in a fresh installation of RDC. I had no bookmarks at all. After creating a bookmark I jumped into Terminal and did a read of the plist file and came up with this:

Now that we had a baseline, I exported the bookmark to the desktop of the VM, edited it to add the “alternate shell” bits, and then re-imported it into RDC as a new bookmark. I then tested to make sure it would work as advertised. After some trial and error, I was able to get the exact syntax for the “alternate shell” entry to work. Now I just needed to see what changes were made in the plist file. A quick read showed me the following:

The key is the line that has “remoteProgram” as part of the entry. You have to get the full path on the Windows machine to the application you want to run on connection to the server. Once you know that path, you can adjust your bookmark script however you need.

The script I posted above, and is linked in my GitHub repo, contains the line to add that Remote Program (alternate shell). If you do not need it, just comment it out of the script.

 

I’m a fanboy. There, I said it and I’m proud of it. I’m a fanboy of JAMF Software’s Casper Suite. I’m also a fanboy of Code42 and their CrashPlan software. Put them together and it’s like when the two teens discovered peanut butter and chocolate as an amazing combination.

I am all about trying to minimize the amount of time my users need to be interrupted due to IT needs. That’s a large part of the reason we use Casper, so that my users do not have to be inconvenienced. Let’s face it, the more time I take performing IT tasks on their computer that cause them to not be able to work, the less money they are making for our agency. It’s one of my primary tenets of customer support: make every reasonable effort to not disturb the end user, period. So when I discovered several of my end user machines were not backing up via CrashPlan, I needed to find a way to deploy CrashPlan with as little interruption as possible. In steps Casper and CrashPlan together.

Our original setup of CrashPlan that has been running for several years, was setup using local logins. At the time when we first deployed, we were not on a single LDAP implementation, so I didn’t want to deploy an LDAP integrated CrashPlan. Fast forward to now, and we have a single LDAP (AD) and I want to take advantage of that implementation to provide “same password” logins for my users.

Fortunately JAMF has a technical paper outlining how to do this, titled Administering CrashPlan PROe with The Casper Suite. This paper was written back when CrashPlan PROe was still a thing. With the release of version 5 of CrashPlan, it has now become simply Code42 CrashPlan. This document still works for the newer version of the software.

Get The Template

The first step is to get ahold of CrashPlan custom template for the installer. Following the paper, you can download the custom template by navigating to this URL:

http://YourServerAddress:4280/download/CrashPlanPROe_Custom.zip

NOTE: If you are deploying version 5 or higher of CrashPlan, you can use this URL to download a newer version of the kit:

http://YourServerAddress:4280/download/Code42CrashPlan_Custom.zip

While there are two different URLs, you can use either one to customize your install.

Edit Away

After downloading and expanding the zip file, you will need to edit the userInfo.sh file to set some settings. First of which is to hide the application from your users during installation. Simply set the following line:

The next thing you will want to edit are the user variables. CrashPlan uses these variables to grab the user’s short name and their home folder location. An assumption is made when it comes to the user’s home folder, and that is the assumption that the home folder lives in /Users. If your home folders do not live there, or you want to script the generation using dscl, you can. I’m lazy and so I simply went with the /Users setting.

Also, the method to grab the user short name is based on the user that is logged in currently. Now, we didn’t discuss before how you were deploying this via Casper (login, logout, Self Service, etc), but suffice it to say, it is preferable to deploy this when a user is logged in to the computer. There have been many discussions on JAMF Nation about CrashPlan and how to grab the user, I used the information found in this post to grab the info I needed:

Now that you have the edits done, keep going through the technical paper, running the custom.sh script next to build the Custom folder we will need in a minute, and to download the installers. The custom.sh script will download the installers for Windows, Mac, and Linux, and slipstream the Custom folder into the installer package for us. In our case, since we are only concerned with the Mac installer, it places a hidden .Custom folder at the root of the DMG. We want that folder. So follow along in the tech paper to mount the Mac installer DMG and copy the .Custom folder out somewhere.

Package It All Up

We are going to need to deploy these custom settings alongside the CrashPlan installer. The tech paper has you using Composer (no surprise since it is their product), but I personally like to use Packages for my packaging fun. I’m not going to get into a discussion about what the best packaging method is, because that’s like debating which Star Trek movie was the best.

Using your method of packaging, create a package that drops that Custom folder (notice we are dropping the period so it is not hidden) into the following location:

/Library/Application Support/CrashPlan

Now that we’ve got our custom settings, we can move over to the JSS to work on our deployment. I’m going to skip discussing how to do this via Self Service, and instead stick with either a Login trigger or Recurring Check-In trigger. But first things first, go ahead and upload that custom settings package you just created into the JSS. Once it’s uploaded set the priority to something low, like 8:

 

Screen Shot 2016-02-12 at 12.05.44 PM

Create Your Policies

The tech paper discusses uploading the CrashPlan installer along with the custom properties, but I like the method that is discussed in this JAMF Nation post. It’s towards the bottom, and basically it uses curl to download the installer from the CrashPlan server. This method insures you have the latest version for your server. Of course, if you are trying to deploy to end users around the globe that may not have curl access to your CrashPlan server, uploading the installer to Casper may be your only option. For me, however, it was not.

First step is to create a new script in the JSS (or upload a script if your scripts are not stored in the database). The script itself is nothing special, it checks for the presence of the CrashPlan launch daemon, and if it is there unloads it and removes CrashPlan. Then the script continues on to install the custom properties (via a second policy) and finally installs CrashPlan:

Screen Shot 2016-02-12 at 12.14.24 PM.png

Screen_Shot_2016-02-12_at_12_28_13_PM.png

As you can see, I’m using a second policy to install the custom properties. You could do everything with one policy and two scripts, or one policy and curl the custom properties from another location. The key point is that if you are removing an existing installation (like I was), you cannot install the custom properties until you are done removing the existing. Make sense?

Now that we have all of our pieces and parts up there, you will create your two policies, one to install the custom properties and the other to run the script.

To Trigger Or Not To Trigger

With your policies created, you now need to determine how you want to trigger these policies. Obviously you will need to trigger one from within the script, but what about the main policy that kicks it all off? Well, I would probably do this via a recurring check-in trigger. It keeps the user from having to wait for the policy to complete before their login completes.

Of course, you could use the login trigger and throw up a nice notification using jamfHelper, Notification Center, or CocoaDialog. That sounds like a nice post for another day.

I Didn’t Do It

I cannot take the credit for this process. It was people like Bob Gendler and Kevin Cecil on JAMF Nation, along with the folks at JAMF and Code42, that did the heavy lifting. I just put it all into one location for me to remember later.

Since I tend to forget things the day after I do them, I’ve decided I’m going to write this one down. While I can manage my way around a LINUX install, especially Ubuntu flavors of Linux, I’m no system administrator or whiz kid by any stretch of the imagination. I tend to plunk around and can find some articles online (usually on Ask Ubuntu forums) to get done what I need. Rather than forget where I found this info, this time I decided I’m going to write it up.

The 5.1.2 upgrade to CrashPlan Pro requires Oracle Java 8 as indicated in this article. Since I run all of my Ubuntu boxes headless (they’re all VMs) I do everything the way any good UNIX head does: via the terminal. Of course, as I stated above, I’m no super user in LINUX so I take the easy way out and use package managers to do the installation, apt-get in particular since I am on Ubuntu. Unfortunately, apt does not have Oracle Java 8 as an available package. So in order for me to be able to use apt-get, I needed to first find a repository that had it.

A quick Google search led me to this article. After following the steps in the article, I had Java 8 installed and ready for the upgrade to version 5.1.2 of CrashPlan.

The next step was to make sure I had a fresh database dump from CrashPlan. Into the GUI for the CPP server to generate a dump. It’s a good idea to copy that dump file off to another server/system.

Now that we’ve backed up and we have Java installed, go ahead and download the update files. You can find the files, along with the official CrashPlan documentation here.

Finally, once you have the server upgraded, you’ll want to upgrade your client devices. You can find the official documentation here.

Recently on JAMF Nation there was a discussion about the Adobe Flash Player Distribution site going away. This site is where admins can go to get a copy of Flash that can then be legally distributed to their fleet of machines. The discussion started out to be about the change Adobe recently made to the URL for this site, but quickly turned to a discussion around distribution of Flash via Casper.

While I have signed up for the Adobe distribution site, I currently utilize a PKG file that comes from AutoPKGr (I replaced my Jenkins install with AutoPKGr last year sometime). Utilizing AutoPKGr makes my life easier, because I do not have to do anything except update my policy to replace the actual PKG file. I’m not going to go into setting up AutoPKGr for use with Casper, there have been plenty of discussions on that, but rather I am going to list out my procedures for processing Flash upgrades.

It’s Upgrade Day

I typically find out that there is a Flash upgrade from JAMF Nation. Someone typically posts that there is a Flash update almost immediately upon release. Once I’ve verified that the update has been uploaded to my JSS by AutoPKGr, I will go update my policy, changing out the PKG file.

As you will see, the policy is set to trigger on “Recurring Check-In” because I don’t care if a web browser is open or not. Flash can be installed while browsers are open, the users just have to restart the browsers that are open after the update. We’ll handle letting them know via a CocoaDialog script.

There are a few pre-requisite items we need to have in place for this process to work. First, we need to have a way to grab the Flash version off of the machines in our fleet. Second, we need to have a Smart Group that will capture all of the machines that are out of spec. This will allow us to scope our policy to those machines.

Grab the Version

I utilize an Extension Attribute to grab the version of Flash and store it in the database. While it can be argued that utilizing an EA to grab the version is not efficient, since the EA will run every time a Recon runs, there really isn’t another reliable method for grabbing the version.

So, setup an EA to grab the version of Flash. My EA is named “AdobeFlashVersion” and utilizes the following BASH script:

That’s pretty straight forward. Now that we have the version, we can build our Smart Group.

Screen Shot 2015-12-18 at 5.38.04 PM.png

As you can see, just pick your EA name out of the list of criteria to search for, and enter the version you are searching for using the “is not” operator.

Policy Time

Now that we’ve got our Smart Group collecting machines that are out of date, we can build our policy to install the update. We will name our policy “Update Flash Player” and place it in whichever category makes sense to your deployment of Casper.

I have my update policy set to run at “Recurring Check-in”, which means that machines will update as soon as they contact the JSS. The frequency is set to “Once per computer”, since we only need it to run one time.

General

We’ll click on Packages next so that we can add our Flash package. Click on Configure to get a list of all packages in the JSS:

 

Screen Shot 2015-12-18 at 5.32.24 PM.png

We should now have a list of all packages that the JSS knows about. Locate our latest Flash Player package and click Add to add it to the policy:

Screen_Shot_2015-12-18_at_5_32_29_PM.png

I utilize a script that runs after Flash has been installed to notify end users to restart any open web browsers. My script uses CocoaDialog to make these notifications, but you can use the built in notification process that Casper has. The script I utilize is below:

Now that we’ve added that script to our policy, we will add a line to the Files & Processes tab to set Flash to not auto update.

FilesAndProcesses.png

That line of information in the Execute Command box simply adds a line to a file called mms.cfg to tell Flash Player to not try to auto update. The line is:

The final thing for us to do is to add our Scope. Just click on the Scope tab at the top and add our Update Flash Smart Group:

Screen Shot 2015-12-18 at 6.01.26 PM.png

 

That’s all there is. Now that we have our update policy in place, each time there’s a new version we just have a few simple steps to update our end users:

  1. Get the new Flash package into the JSS
  2. Change our Smart Group to look for the new version number
  3. Change our policy to remove the old version and add the new version
  4. Finally, Flush All on the policy logs so everyone in the Smart Group gets the update.

 

I have been utilizing this method for updating Flash for well over a year now, and I have not had any troubles at all.

I hope this quick article has helped you out.

 

Wells Fargo BuildingEvery year as October gets closer, I get anxious. I know that at some point, usually toward the end of the month, I will be traveling up to Minneapolis for the JAMF Nation User Conference, JNUC. The conference is located at the Guthrie Theater in downtown, in the Mill district of Minneapolis, right on the Mississippi river. It’s one of my favorite locations to go to.

The JNUC is one of my favorite conferences to attend. Not just for the great content, but for the relationships that get formed and strengthened there. There are friends at JNUC that I’ve known for well over 10 years now, just from attending different conferences in the past. It’s great to catch up with these friends.IMG_0839

This year was also special because I was presenting there. I had the opportunity to present on imaging in a session titled “Unwrap the Imaging Enigma”.  It was a wonderful experience, and one that I will repeat again. Giving back to the community by presenting is important for any admin. If you’re interested in the slides, you can find them on my GitHub repository.

Now that I’ve called more attention to this blog, and to myself, I will try to post more relevant content regularly. If there is a topic you’d like to see, just post it in the comments and I’ll see if I can come up with something. Or reach out to me on Twitter: @stevewood_tx