The Curious Case of VMware Snapshots with MCS in Citrix Cloud

While working on a project, I came across what seems to be an undocumented issue with Citrix Cloud and on-premises VMware. So, I thought it would be helpful to document and share it along in case you, or someone you know, experiences it.

The scenario

The virtual machines users use for resources are hosted on (a.k.a. Virtual Desktop Agents (VDAs)) Citrix Cloud, with an on-premises VMware vCenter host connection.

Citrix Machine Creation Services (MCS) is used to create and update machine catalogs from a master image virtual machine.

Though not yet visible, the problem begins when using VMware’s default naming convention for snapshots on your master image. In this scenario, VMware names it “VM Snapshot,” along with a date, time, and time zone format as part of the name (as shown below):

The problem

Interestingly, when you create or update a Citrix machine catalog for the first time (using the scenario described above), everything works fine. But the second time through, the problem appears.

On the second attempt to update a machine catalog with MCS in Citrix Cloud, you simply don’t see the latest available VMware snapshot.

Here’s how the Citrix Cloud MCS update process looks. Note that we don’t see the latest available 3:34:14 PM VMware snapshot:

Here’s how the same master image VMware snapshot looks (this time) via Citrix on-premises. Note that we can see the latest available 3:34:14 PM snapshot:

Back in Citrix Cloud, if you hover the mouse over the icon (next to the last-shown snapshot) you receive an error stating it cannot find the path because it does not exist:

But with Citrix on-premises, hovering over the same icon on the same snapshot does not give you the error and instead, you’re able to see the latest snapshot’s details:

Here’s why

This occurs only with Citrix Cloud because using VMware’s default naming convention for snapshots causes the “/” (forward slashes) to be encoded as back slashes, and seen as paths instead—so it errors out and never goes past seeing the next available snapshot.

Here’s a closer look.

When you create a VMware snapshot using its default naming convention, it names it “VM Snapshot [MM/DD/YYYY] [time] [time-zone]” as shown below:

If you look at the snapshot details in VMware, you’ll note the “/” (forward slashes) are encoded to either %2f (single encoding) as shown below:

Or %252f (double encoding), as was shown in the above examples.

The workaround

Because things in the cloud are built differently, they interact differently. So, the best workaround for this would be to not use VMware’s default naming convention for snapshots. Instead, incorporate your own naming convention for VMware snapshots (with no slashes). One that has worked great for me is MM.DD.YY.[letter] – For example, assuming today is March 19, 2020:

• If 1st snapshot for the day, name it: 03.19.20.A
• If 2nd snapshot for same day, name it: 03.19.20.B
• If 3rd snapshot for next day, name it: 03.20.20.A

And so on…I also enter a brief summary in the description field.

Problem Solved! I hope that helped. For additional tips and tricks, be sure to check out our blog. And to learn more about how Anexinet can help your organization transform its infrastructure—in addition to fixing your VMware snapshots—please don’t hesitate to reach out to us. We’d love to help.