0

I've recently spun up several new Windows 2012 R2 cloud servers on Rackspace, using their base Windows 2012 R2 image (without SQL server), and in each case the servers suffer from random/unexpected reboots. There are certain actions that can consistently trigger the reboots (more on that in a bit), but unfortunately there is not much useful information being captured in the event log, nor is there any crash dump being generated.

In the case of these particular servers, they are intended primarily for running the Jenkins build/CI server. Whenever I run our main build job in Jenkins (which usually takes around 7-8 minutes), the servers will typically reboot a few minutes in, prior to completion of the job. Browsing the web seems to also trigger a reboot, though not immediately (Please spare me the comments about running a browser on a server. We're just using it to make the process of setting the build server up a little easier).

As for the event log, the only event that consistently shows up is a Critical "Kernel-Power" event (Event ID 41) with the message The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly. When looking at the details of the event, the "Bugcheck" properties are always set to 0.

I've tried building new servers multiple times now, each time with the same result. The servers I'm trying to build are 2GB General Purpose servers, using local storage for the boot volume. I've even been able to reproduce the issue without making any changes to the server, and simply running Internet Explorer and browsing for a few minutes.

I've been in touch with Rackspace support, asked if there's potentially an issue with the image itself, and also requested they try moving the VM to another host. They've tried their best to help so far, but have been unable to provide any type of resolution.

We've built plenty of Windows servers on Rackspace in the past, and have never experienced this problem.

I can even build new servers from an image we have of a Windows server that was built several years ago. The only reason we can't use this image is because the size of the server it was taken from is larger than what we need for our build server, and we do not want to pay the extra $$ for resources we don't need.

Has anyone else experienced, and hopefully solved, a similar problem?

Justin Holzer
  • 101
  • 1
  • Not sure what you build, but 2g is TINY for anything in windows, paricularly a build server. Checking my build servers and each image is a LOT bigger. Also 2012R2 is quite old - in terms of: any reason not to use a current OS image for build servers? The vm integration is a little better, if provided. – TomTom Jun 06 '19 at 14:29
  • 2 GB is all we need. The reason we're using 2012 R2 is b/c that's the most recent version of Windows offered by Rackspace. I don't see the need for a debate about the server size, as its unrelated to my question. We've been running a 2GB build server for several years now, and have had no performance issues. Unfortunately, we did have a non-performance related issue with our old server, which is why we're building out a new one. And FWIW, I have been able to reproduce the issue with larger sized servers, so it doesn't appear to be related to using a 2GB instance. – Justin Holzer Jun 06 '19 at 14:34

1 Answers1

0

The problem turned out being the virualization drivers (they use Citrix Xen hypervisors) that were installed with Rackspace's base Windows 2012 R2 image.

Rackspace eventually provided a link to download the updated drivers. Once I installed them, the unexpected reboots stopped.

Justin Holzer
  • 101
  • 1