Unstable machine and lost connections to Linux VM (Ubuntu default Image)

Question

I'm testing the Linux VM functionality for a few days now. After a while I suddenly lost the connection and the VM did not recover. The load, I/O, swapping, etc were very low at that specific moment.

Wondering what happened?

Answer 1

Hey Djon,

Has the Linux Virtual Machine since come back up or are you unable to connect to it even now, one hour later? What state is listed next to the Virtual Machine in the portal?  

Based upon the symptoms you are calling out, it is possible the hardware hosting the VM went bad, indicating the lower I/O and swapping numbers. Then, we may have detected it and migrated your virtual machine to new hardware. However, I would expect it to be back up by now...

If it is still unable to be connected, can you please send the following information to [e-mail]:

• Subscription ID
• Live ID
• Deployment ID of the VM
• Name of the VM
• Approximate time when the VM start showing bad symptoms and then time when it died.

We would love to investigate further if this issue is still plaguing you.

In the meantime, if you were using that disk and had saved a bunch of content, you could always copy the disk and bring it back in as another Virtual Machine. I explain how to do this in this blog post: http://social.msdn.microsoft.com/Forums/en-US/WAVirtualMachinesforWindows/thread/94b52c79-5487-453d-81b7-016baec62d28

Corey

Answer 2

Hi Corey,

The VM got back online after a reboot, although the issues are still here. This Linux VM is very unstable. Tens of times a day the machine is inaccessible for 5 to 20 minutes and comes back again. This behavior has been repeating during the last couple of days.

Sometimes the machine seem to have rebooted (looking at the uptime), but not every time. 

I'll drop you an email with the information required for further investigation.

Thanks, Djon

Answer 3

After investigation by Azure operations, the issues are reported to be caused by faulty memory modules in the host server. They took the host out of the production pool, causing this VM to be transfered to another host.

This solved all the issues.

For future reference: if you experience the same issue, you may delete the VM and redeploy the same VHDs again. Most probably this solves the issue since there is only a slight change the new VM instance would land on the same host server.