2

I have a Linux VM running on Azure. It's funded via a MSDN account, though I don't think that is relevant to my question.

Since creating it ~1 week ago, I've had four separate times where I've found it's state is suddenly "Stopped - Deallocating". If I look in resource health, I see:

"This virtual machine is stopping and deallocating as requested by an authorized user or process"

This is not true. I am the only person with access to my account, and I have no automation or anything like that which might cause a command to be sent.

I'm a bit baffled. If I look inside the VM, all I see in the log entry in syslog that looks like this:

Jan 14 11:02:08 HOSTNAME kernel: [81679.063027] hv_utils: Shutdown request received - graceful shutdown initiated

Any ideas what could be triggering this? My intent is to have the VM up 24x7. If I go into the portal and click Start, it fires right up and works great, but obviously this isn't what I was hoping for.

Thanks!

  • Googling hv_utils: Shutdown request received gives many social.msdn.microsoft.com results of people with similar problems.

    It seems that the VMs can be reboot for a multitude of reasons, and some may be out of your control. One way to avoid this is configuring them for high availability.

    – MC10 Jan 14 '19 at 21:32
  • 4 times has become 12 in ~10 days. I can't imagine Azure is engineered for this. Additionally, in the last two years, I've experienced 1 reboot across six other VMs, so I'm thinking that 12 in 10 days for one indicates an issue. I will try to redeploy. – raindog308 Jan 16 '19 at 21:12
  • Redeploy made no difference, and I've had another 48 (wow!) stop/alloc in the last 10 days. Microsoft is clueless so I think it's time to just find another platform. Azure may be ready for prime time, but neither its interface nor its support is because neither can diagnose this elementary issue. Alternatively, dozens of failures a week is normal performance. – raindog308 Jan 26 '19 at 19:57
  • 2
    And just to wrap up, I did finally locate the issue - I think. There is an auto shutdown option in Azure (amazing, MSFT support knew nothing about it, even though it's in the panel). It's designed to shut down VMs at a particular time (e.g., 1700 UTC every day or something), but apparently it malfunctions as I saw many shutdowns per day. Turning it off has, over the last 24 hours, resolved the issue. If you search in the (current, Jan 2019 version of the) Azure panel for auto shutdown, you'll find the option. – raindog308 Jan 27 '19 at 15:44
  • I had the same problem. Very unexpected feature, especially to be on by default. – Lee Jensen May 09 '19 at 20:00
  • yep https://azure.microsoft.com/en-us/blog/announcing-auto-shutdown-for-vms-using-azure-resource-manager/ Auto shutdown enabled by default? WTF Microsoft? – Toolkit Aug 22 '19 at 19:44

1 Answers1

2

azure auto shutdown

sometimes I'm wondering if this is a bad joke working with azure...

And the best is that Microsoft is telling you YOUR machine is performing the shutdown (which is technically reasonable, as the underlying Hyper-V triggers the Hyper-V Tools and these are shutting down the VM)... But there MUST (to be conform with RFC) at least a notice that auto-shutdown has been enabled (by default \w/)

Issue occured with Debian 10 Buster VM as of today.

To enable other searchers to find this thread: azure instance shutdown unexpectedly

The diagnosis by azure:

We ran diagnostics on your resource and found an issue

We identified that your VM became unavailable at 2020-03-09 19:07:19 (UTC). This expected occurrence was caused by a user initiated shutdown action.

The shutdown was triggered by an authorized user or process from either the Azure Portal or from Azure Resource Manager interfaces. As a result, your VM was shut down and remained in this state until user action was taken to restart it.

To troubleshoot similar issues in the future, you can use Azure Resource Health, which provides insights into the current and past health of your resources.

T.J.L.
  • 1,305