CODEWITHSUNDEEP
azure
Mustafa Hussien
Mustafa Hussien

Reputation: 1

Cloud service restored to the last production deployment

I've update production deployment yesterday morning then I've made changes to service files using remote connection add and update files and everything was OK. today morning all the changes I've done after deployment was undone and customers use the old version and this cost us hundreds of thousand of pounds i need to know what's happen nothing appeared in operations log

Upvotes: 0

Views: 30

Answers (3)

Simon W
Simon W

Reputation: 5496

As others have said, PaaS Web Roles are stateless. If you're making manual configuration changes to your deployed solution package after it has been auto-deployed then any re-deployment by the Azure fabric will simply deploy the package minus your manual changes. To solve this issue you could use startup tasks to apply your manual changes using a PowerShell script or similar (depending on what you're changing). See http://msdn.microsoft.com/en-us/library/jj129544.aspx.

Note that startup tasks don't just run when a machine gets re-imaged or rebooted.

Upvotes: 0

kwill
kwill

Reputation: 10998

PaaS services are stateless, which means the VMs running your service can be destroyed and recreated at any time, at which point the VM will be recreated with the content from your original .cspkg.

For more information see http://blogs.msdn.com/b/kwill/archive/2012/09/19/role-instance-restarts-due-to-os-upgrades.aspx and http://blogs.msdn.com/b/kwill/archive/2012/10/05/windows-azure-disk-partition-preservation.aspx.

Upvotes: 0

Lukos
Lukos

Reputation: 1852

Probably what has happened is that Microsoft has updated your servers at the Cloud Centre and re-deployed your application from the original deployment. This is in their terms and conditions, you should not make any important manual changes to the deployment after it is deployed unless they are stored in the portal (environment settings etc.), otherwise they might be lost during updates or reboots.

I learned this the hard way too. I had a cache role with only one instance (I thought it only made sense with one instance) and while updates happened, my whole site went down several times over several days!

Upvotes: 1

Related Questions