0

I just followed creating Azure ScaleSet tutorial,and selected Windows scale set. As Parameters, I provided admin username as "admin" for testing. When I start the deployment it started creating storage accounts, public IP etc. and suddenly it failed complaining that the username is not strong enough.

But when I check my Azure resources, I found those storage accounts and other resources being created. Isn't there a roll back process to rollback these resources?

enter image description here

Dhanuka777
  • 8,331
  • 7
  • 70
  • 126

1 Answers1

0

The concept behind ARM is that everything is managed at the Resource Group level. That doesn't matter whether you're talking about a single storage account or a thousand VMs. If a deployment doesn't work the way to rollback is to simply delete the resource group and everything contained in it will be deleted as well.

from the Azure Resource Manager overview

All of the resources in your group should share the same lifecycle. You will deploy, update and delete them together. If one resource, such as a database server, needs to exist on a different deployment cycle it should be in another resource group.

Like most other things in IT the important part is deciding where the lines of division should go. Personally I try to keep a resource group as a discreet service, so a DB, a scale set, front end web servers. That way each individual unit can be swapped out without affecting the other parts.

Michael B
  • 11,887
  • 6
  • 38
  • 74
  • @Micheal - Resource Group is the high level group which includes more that the Scale Sets. In my case there are other resources in the resource group which is not related to above operation. Therefore deleting resource group is not an option for me. – Dhanuka777 Apr 07 '16 at 06:27
  • In that case, your only solution is to manually go through them and delete. and next time you deploy have a separate RG to make all this stuff easier to manage – Michael B Apr 07 '16 at 06:30
  • Thanks for the suggestion. I was expecting a auto roll_back functionality and surprised to see half of the results committed. Isn't there anything we can do at script level or programming level for this? – Dhanuka777 Apr 07 '16 at 06:33
  • Like I say this is the functionality of the RG, it is the container for the deployment if it doesn't work you blow it away and start again. I can't think of an easy way you could script it either, I think you'll really need to pick them out by hand – Michael B Apr 07 '16 at 06:36