Cancel RollingUpgrades before deleting VMSS resources

[harshavmb]

RollingUpgrades when setup on VM ScaleSets cause lot of problems during destroy. This gets worse when extensions are run on these scalesets.

At the moment, we use destroy provisioners, to cancel rolling upgrades & then destroy resources such as extensions & VMSS.

This tiny boolean flag cancel_rolling_upgrades_before_deletion when set to true will cancel rolling upgrades if they are running and then move forward to delete the VMSS. This way, we avoid writing destroy provisioners.

In addition to this, the logic in helpers.go file, future.Response().StatusCode == http.StatusConflict && strings.Contains(err.Error(), "There is no ongoing Rolling Upgrade to cancel.") is to handle a scenario on a newly provisioned VMSS resource where rolling upgrades are set but no extensions, auto-scaling are run.

The error parsing logic is to handle the error thrown by Azure API when no Rolling Upgrades are ongoing.
Get-Latest API URL ::
https://management.azure.com/subscriptions/{subId}/resourceGroups/{resourceGroup}/providers/Microsoft.Compute/virtualMachineScaleSets/{vmSSName}/rollingUpgrades/latest?api-version=2021-07-01

In other cases, we cancel, wait until the cancel is complete & move ahead to delete the resource associated with VMSS.

Have compiled & tested locally, it works as expected.

Hope this helps!

[harshavmb]

RollingUpgrades when setup on VM ScaleSets cause lot of problems during destroy. This gets worse when extensions are run on these scalesets.

At the moment, we use destroy provisioners, to cancel rolling upgrades & then destroy resources such as extensions & VMSS.

This tiny boolean flag cancel_rolling_upgrades_before_deletion when set to true will cancel rolling upgrades if they are running and then move forward to delete the VMSS. This way, we avoid writing destroy provisioners.

In addition to this, the logic in helpers.go file, future.Response().StatusCode == http.StatusConflict && strings.Contains(err.Error(), "There is no ongoing Rolling Upgrade to cancel.") is to handle a scenario on a newly provisioned VMSS resource where rolling upgrades are set but no extensions, auto-scaling are run.

The error parsing logic is to handle the error thrown by Azure API when no Rolling Upgrades are ongoing. Get-Latest API URL :: https://management.azure.com/subscriptions/{subId}/resourceGroups/{resourceGroup}/providers/Microsoft.Compute/virtualMachineScaleSets/{vmSSName}/rollingUpgrades/latest?api-version=2021-07-01

In other cases, we cancel, wait until the cancel is complete & move ahead to delete the resource associated with VMSS.

Have compiled & tested locally, it works as expected.

Hope this helps!

Finally apologies for these commits originated from my main fork as there is another PR waiting for approval. Next time, I create separate branches for each PR so that commits look clean.

[harshavmb]

Hi @tombuildsstuff @katbyte @jackofallops ,

Could you please provider some update on this PR? Waiting in the review queue for quite sometime.

[evaroquaux]

Hello,

This PR will fix a known and real limitation when working with Azure VM scale sets in Terraform. Meanwhile, it is blocking the development of our Terraform-based automation.

Is there any chance to have it considered? It was submitted a long time ago and there has not been any movement for a while.

[katbyte]

closing in favour of #18973

[github-actions]

I'm going to lock this pull request because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active contributions.
If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.