Microsoft has wonderful products, but half the world is still using Citrix Presentation Servers for their terminal servers. The latest version is no longer called Presenation Server but is called XenApp. After version number 4 they change not only the name, but a lot of functionality, engine's as well as looks en feels. Personally I think the XenApp 5 product is nothing more than a Presentation server 4 with different screens, but XenApp version 6 is growing out to be a very good product, which will be used a lot, worldwide. But, as all products, I ran into some nasty and persistent problem lately, but this error is a bit stupid. As always, whenever there is work in process, mistakes can happen... and this was one of them. Thanks to our wonderful monitoring tool System Center Operations Manager, the damage was rather low.


I was on my way to the office when our standby engineer called me and told me none of the servers in a particular farm did not come up, and wouldn't perform a PXE boot. Wow!!, so I stepped on the peddle and 10 minutes later I found the problem. Every XenApp server had a message on the console, saying that it wanted to perform a PXE boot, but it had an error doing that, mentioning that the Vdisk is locked.

So I checked what happend the day before, and a colleague of mine, demonstrated this particular farm for a small audience the night before.

Aha, thats an easy solution!

Quickly opened the Provisioning Console and yes, there it was. The new copied disk, got a higher number and Provisioning services automatically added this one as the new "Golden Image".  This was a copy of the normal "live" disk but unfortunately this image was still in "Private" mode. And new provisioning services can only boot from read-only disks. Solved it rather quickly by connecting the original "Read-Only" previous disk, and resetted the virtual machines. No harm done, as I'm always one of the first people in the office, and customers hardly log in at those hours.


I thought it might be handy to create an article on this, as perhaps other people with less knowledge might run into the same situation. 
What the customers experience is also nice, and below is a little screenshot of it.
On launching the desktop they while only see the text "An error occured while making the requested connection".   

Thats logical as there are no servers available to provide a session when all servers are down.


Now how can we prevent this from happining again? I added the error event in a picture below.










This simple event 11, from StreamingProcess can be read with your monitoring tools. It clearly says that provisioning services is trying to get a read-only lock for this disk and that it failed. This can be monitored with any reasonable monitoring tool you want to use, and when you are using event forwarding it will even become more easy to set an alert for it! All alarm bells should go off when this is happening. 

By the way, these pictures are taken from my test-environment ofcourse :-).

So, we learned a lesson, to check things better after a demonstration,
and I can only hope you do not make a mistake like this.....