0

Dell PowerEdge T110 II Tower Server |
Windows Server 2012 R2 | Hyper-v VM manager | Windows 7 Pro (Guest OS)

The Hyper-V guest OS (win7 pro) serves as the database server (MySQL) for storing customer data. Power outage occurred, network switch connecting the tower server to the outside world was at the moment (for some reasons) not connected to power backup systems therefore was turned off along with other devices except the DELL server.

I wasn't 'on-site' when this happened so local personnel (NO login access to the DELL server tower's host OS), thinking it was just a simple matter of power cycle, powered up the network switch but after having been informed by the front-line tellers that the SQL database server (installed on guest OS) within the tower server's host OS is still not accessible, pressed the server's POWER BUTTON to initiate a soft shutdown (again thinking it was just a simple power-cycle feat).

Database still not accessible, I was called and turned right up to find out- Guest OS is in a "SAVED" state with a few parent and child snapshots (checkpoints). The attempt to turn it ON pulled up errors after errors that hindered me from even seeing the guest OS' boot screen. I selected the option to "Delete Saved State" and start but to no avail. So I went ahead to just create a NEW virtual machine using the same VHD file that the troubled VM was using. Creation successful, logged in to the Gust OS, check all installed programs especially the SQL database and other supporting applications. Everything looks perfectly fine even from the front-line teller's workstation's end. Problem fixed so I thought I'd call it a day.

Then the most dreadful thing happened - all daily inputs of data from FOUR MONTHS AGO went missing! Now the tellers are breathing down my neck. I had no choice but to resurrect the poor old VM. Fortunately, I was able to pull it off. How did that happen? Isn't data supposed to be stored in the actual guest OS and NOT the snapshots associated with it?

enter image description here

David Makogon
  • 2,768
  • 1
  • 20
  • 29
fink
  • 21
  • 1
  • 1
  • 4
  • 1
    Welcome to serverfault. I realize you're a bit agitated but... for future questions, please attempt to leave out the color commentary - as written, this is quite a lengthy question (plus it didn't help to have it written as one giant paragraph, which I broke into multiple paragraphs to make it somewhat readable). – David Makogon Apr 14 '16 at 03:31

1 Answers1

2

Umm... no. The original VHD(X) is put into a READ only state and all new writes are written to the snapshot disk. By using the original VHD(X) to get the VM going again you essentially reverted the data back to the point it was in before the first snapshot was taken. You should probably bone up on your snapshot knowledge.

Additionally, why are you creating snapshots for a production VM and why are you running on the snapshots so long? That's counter to best practice. Snapshots are meant to be used as a way to roll back to a known good point in time when you're testing updates, installations or other changes to your VM.

http://www.virtualizationadmin.com/articles-tutorials/microsoft-hyper-v-articles/backup-recovery/managing-hyper-v-snapshots-part1.html

I hope you don't think that snapshots are backups, because they aren't. I also hope you're taking regular backups of the VM and/or the data... but it sounds like you probably aren't.

joeqwerty
  • 109,901
  • 6
  • 81
  • 172