{"id":2402,"date":"2015-11-29T15:00:00","date_gmt":"2015-11-29T20:00:00","guid":{"rendered":"https:\/\/www.tmurgent.com\/TmBlog\/?p=2402"},"modified":"2015-12-08T15:21:23","modified_gmt":"2015-12-08T20:21:23","slug":"im-converted-using-the-converter-from-app-v-4-x-to-5-1","status":"publish","type":"post","link":"https:\/\/www.tmurgent.com\/TmBlog\/?p=2402","title":{"rendered":"I’m Converted: How I Convert Packages from App-V 4.x to 5.1"},"content":{"rendered":"

\"Convert\"This has to rank among the top posts that I never thought I would write in my entire life.\u00a0 I mean, it is\u00a0right up there with “Why I’ll never play golf again”.\u00a0 If we had\u00a0subtitles, the subtitle of this article would be “The 5.1 Converter Doesn’t Suck<\/i><\/b><\/i>“.\u00a0 High praise indeed!<\/p>\n

Abstract<\/strong><\/p>\n

This post details out how to organize an efficient conversion project to move packages from Microsoft App-V 4.x to 5.x.\u00a0\u00a0The example project uses App-V 5.1 and\u00a0the free AppV_Manage\u00a0tool (5.1.3) to efficiently manage the process.<\/p>\n

The Back Story<\/strong><\/p>\n

When Microsoft re-wrote App-V in version 5, they made such major changes to how application virtualization worked, modernizing it for today’s platforms and apps, they had to change for format of the packaged applications in an incompatible way. To recognize how dramatic that is, the last time there was a major breaking change in the format was back in 2003.<\/p>\n

So as part of the 5.0 release Microsoft included a package conversion utility. They told us how wonderful it was. We (the App-V MVPs) tried it. It wasn’t. We generally told customers that it probably wasn’t worth the bother. Particularly galling was that it worked better on packages that were originally sequenced to the VFS instead of into the root folder. Microsoft continued recommending root folder sequencing right up to the end of 4.x (even though I started training people to use VFS installs years before). So a very large portion of converted apps wouldn’t work for Microsoft customers that I didn’t train.<\/p>\n

So Microsoft improved it in 5.0 SP2. We tested it. It wasn’t that improved.<\/p>\n

Then they improved it in 5.0 SP3. Maybe others tested it, I didn’t bother.<\/p>\n

Then they improved in in 5.1. Being jaded, I barely even noticed it in the ReadMe for the release. But it was a side discussion that I had at the MVP Summit with the development team that lead me to give it a shot. While the long version of the results are in the post below, the bottom line is the 5.1 converter doesn’t suck<\/em><\/strong>. They have taken a lot of feedback from customers that tried conversions in the past and made changes so that those packages will now convert. It doesn’t mean they all work, but for those that haven’t made the move over to App-V 5 yet, it is probably worth the effort.<\/p>\n

Overview of\u00a0my Conversion Process<\/strong><\/p>\n

Moving from 4.x to 5.x\u00a0should be treated\u00a0a project.\u00a0 If you use the App-V Management Server, you’ll be replacing the servers.\u00a0 You’ll be installing new App-V Clients — maybe in parallel or maybe as a replacement.\u00a0 And you need to convert or repackage the apps.<\/p>\n

That last sentence, convert or repackage, shouldn’t be taken lightly and should be run as a project all on it’s own.\u00a0 Just as it would be to certify the apps on Windows 10 (which you should do as part of this project anyway), you’ll need to test, test, test.\u00a0 The image below is an overview of that process:<\/p>\n

\"Convert\"<\/p>\n

The work-flow has two distinct stages, conversion and testing.\u00a0 Each of the red-tinged boxes are tasks which may result in a package being removed as a candidate for conversion.\u00a0 Note how happy the user looks at the end of the project!<\/p>\n

Stage 1: Conversion<\/strong><\/p>\n

It starts by getting a list of all of the existing software packages.<\/p>\n

Because the\u00a0work is automated and difference in effort to attempt conversion of an extra package is quite\u00a0minor, I prefer to\u00a0attempt the conversion on every package, so I don’t prefilter.<\/p>\n

In parallel,\u00a0you will want to filter out some of the packages. Often, this effort is actually performed by someone other than the hands-on person doing the rest of the work in this\u00a0stage.<\/p>\n

This might involve running reports from the existing management server to determine apps that really aren’t being used.\u00a0 You should probably involve the business units to review the list and at least give them an opportunity to say they don’t need some of them.\u00a0 You can do this filtering up-front, but I depict this as happening in parallel during stage 1, as I find it better to do the first part, which is quite automated, against all of the packages in case some packages get filtered out and down the line someone pops up with a need for a filtered out app.<\/p>\n

As part of filtering, you will also want to develop a prioritization of the list.\u00a0 This will dictate the order in which packages are verified and, if necessary, remediated.\u00a0 You should also consider whether you want a policy to attempt remediation, or just mark for repackaging, any issues found.<\/p>\n

The converter requires the 4.x packages\u00a0to be up to 4.5 to be converted.\u00a0 I suppose you could open and save older packages in a 4.6 based sequencer first, but maybe those should be repackaged anyway.<\/p>\n

Stage 1 Details: VM Prep<\/strong><\/p>\n

You basically use a VM with the latest version of the sequencer on it.\u00a0 If there are any apps that were sequenced on an x64 OS, you’ll want this VM to be x64 also.\u00a0 There are two parts to the conversion, which is done via a script in an elevated (RunAs Administrator) 32-bit PowerShell window.<\/p>\n

The conversion doesn’t care what is on the VM, so it doesn’t need snapshots of clean machines – there is no\u00a0runtime monitoring involved, just an asset conversion.<\/p>\n

You will want\u00a0a folder that has all of the existing packages under it.\u00a0 It doesn’t matter what the folder structure is, and it can be a network share.\u00a0 You can point it to the current content store of your production packages as no files are opened for write access.<\/p>\n

You will want an empty\u00a0local folder that will store the converted packages.<\/p>\n

You will want a PowerShell script to perform the conversion.\u00a0Here is a link to a zip file containing my script<\/a>.\u00a0 The script has a bunch of special stuff to take care of weird things like multiple copies of the same package and it provides some statistics.\u00a0 The only thing you need to do is change the first two non-comment lines, where you reference the source folder (or share) of the content source and the top level destination folder.<\/p>\n

Stage 1: Testing and Converting<\/strong><\/p>\n

The script will test each package to see if it is likely to convert.\u00a0 If there are no errors recorded, it will then run the conversion.\u00a0 The conversion also runs the test, but there are situations where the conversion fails to display a warning because of a secondary issue that only is found during conversion — and I like to know about all of these when I review the output.<\/p>\n

You just start the script and come back later on to review the results. The script provides on-screen display as it is working, and each output folder has a copy of the test\/conversion results.\u00a0 I did find a couple of situations where the on-screen display reported an error that didn’t appear in the output — so you’ll want to scan the on-screen display for red text.<\/p>\n

On my system with my packages, it took between 5 seconds and 15 minutes to attempt to convert each package.\u00a0 The larger times are for very large packages, like AutoCAD and ArcGIS.<\/p>\n

In my sample of 170 packages,\u00a0I averaged 1.25 minutes each, but as you can see in the graph below, most packages convert quickly.<\/p>\n

\"ConvertTimes\"<\/p>\n

Stage 1: Analyzing Conversion Output<\/strong><\/p>\n

The packages that I ran were not necessarily production quality packages. I simply took a subset of packages left over from some of our old App-V 4.x training classes, created by people that were learning and weren’t necessarily great packages. So some of the problems that I found were simply bad 4.x packages to start with and you probably won’t run into these yourself.<\/p>\n

Here is a sample output file, ConvertedTestOutput.txt, produced by the script:<\/p>\n

Monday, November 9, 2015 8:57:39 AM\r\nTEST CONVERSION:\r\nSource\u00a0\u00a0\u00a0\u00a0\u00a0 : \\\\tmuhost\\Content\\FB_APPS\\IBM_WMBToolkit_7.0\\5_ProdRelease\\App-V\r\nErrors\u00a0\u00a0\u00a0\u00a0\u00a0 : {\r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 The package converter detected one or more applications in your package that is targeted for an \r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 unsupported operating system.\u00a0 This package cannot be converted unless the target operating system \r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 restriction is removed from the .osd file.\r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 }\r\nWarnings\u00a0\u00a0\u00a0 : {}\r\nInformation : {}\r\nTarget\u00a0\u00a0\u00a0\u00a0\u00a0 :\r\nACTUAL CONVERSION:\r\nSource\u00a0\u00a0\u00a0\u00a0\u00a0 : \\\\tmuhost\\Content\\FB_APPS\\IBM_WMBToolkit_7.0\\5_ProdRelease\\App-V\r\nErrors\u00a0\u00a0\u00a0\u00a0\u00a0 : {\r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 The package converter detected one or more applications in your package that is targeted for an \r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 unsupported operating system.\u00a0 This package cannot be converted unless the target operating system \r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 restriction is removed from the .osd file.\r\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 }\r\nWarnings\u00a0\u00a0\u00a0 : {}\r\nInformation : {}\r\nTarget\u00a0\u00a0\u00a0\u00a0\u00a0 :\r\n<\/pre>\n
The issues that the test\/conversion found in my packages included the following lists:<\/p>\n
Errors<\/strong><\/p>\n