Closed TTTouper closed 2 days ago
Hello @TTTouper. I wasn't able to reproduce the issue, in my case subnet deletion goes into 'still destroying...' loop. Are you using some kind of module? Which region are you launching this config on?
Will it remain in the 'still destroying...' loop? What will the final state be? It’s in the OTC pre-prod environment
This issue occurred during resource destroy after the problem in Issue 2712 https://github.com/opentelekomcloud/terraform-provider-opentelekomcloud/issues/2712
The subnet destruction loop occurs when a resource tries to delete itself but cannot due to other related resources still being in use. In this case, an active ECS instance prevents the subnet from being deleted. While unbinding the ECS during deletion might resolve the issue, this isn't the correct approach when using Terraform. Resources created through Terraform configuration should not be manually modified via the console. Binding an ECS to a Terraform-managed subnet can lead to loops or inconsistencies with the Terraform configuration. All Terraform resources are based on documentation from the official portal. If the API contains bugs, they may be inherited by the Terraform provider. Some of these issues can be resolved programmatically on the Terraform side, while others cannot. The PreProd environment requires debugging and investigation efforts from our side and will likely uncover numerous discrepancies across different PreProd services. This requires significant effort and testing, involves many resources, and the Terraform development team lacks the bandwidth for such extensive testing. Therefore, please only open issues if you have encountered them in the production environment.
Hi @artem-lifshits ,
please stop saying "don't open tickets regarding PreProd environment".
The PREPROD environment is the quality gate of the OTC ecosystem, which ensures that our customers will receive high quality services. If you are willing to investigate only those issues which are coming from the production system, this would jeopardize the business of our customers and the OTC brand in the long run which may lead to serious financial consequences.
The strategy, what determined by the OTC Leadership, is clear: we must deliver top quality products to our customers. This is the highest priority, which can only be achieved if we perform proper testing in the PREPROD environment in order to eliminate as many problems as possible. Releasing products to customers, which contain several issues/problems, can lead to penalty and loss of reputation.
I hope, now you understand how important is this and why it is crucial to investigate and fix such issues, which occur in the PREPROD environment.
I expect full cooperation and support regarding the investigation of this issue from your side. I know that every team is heavily overloaded, but this cannot be the reason for ignoring such issues. Please reopen this ticket and help to figure out the root cause of this problem.
@mrumpler3 issue happens because of subnet management API, feel free to explore why it's not working on preprod.
Hi @artem-lifshits ,
so you says that the received log entry about the successfully completed subnet destruction (which did not happen in the reality) is coming from a subnet related API and Terraform has nothing to do with it.
Am I understand your statement correctly?
Hi @mrumpler3, could you please provide a more detailed description and a reproducible test case, including what you’ve done and what you’re aiming to achieve? Also, I noticed an issue where the Terraform state may be disrupted if changes are made manually in the cloud. It’s important to manage all resources through Terraform and avoid making manual adjustments.
Hi @anton-sidelnikov ,
you can find all the necessary information in the following bug ticket: [RFS][PREPROD] Auto-rollback does not operate properly in some edge cases
There are 2 problems:
The problem is not that either a resource creation or a destruction fails, this was the aim of the test and the expected outcome. The problem is that the result of the creation and the destruction was incorrectly identified.
Hi @mrumpler3, could you create account for me in preProd? Also good to have a call next week to discuss this RFS thing, could you arrange one?
Hi @anton-sidelnikov ,
of course, I will arrange a call for next week. I also sent you the account information you requested in Zulip.
Hi @mrumpler3 @TTTouper, i've added test results in JIRA ticket, this is not tf issue but RFS itself.
Hi @anton-sidelnikov , thank you very much for the support!
Terraform provider version
1.36.15
Affected Resource(s)
opentelekomcloud_vpc_v1 opentelekomcloud_vpc_subnet_v1
Terraform Configuration Files
Debug Output/Panic Output
Steps to Reproduce
terraform init
terraform apply
terraform destroy
Expected Behavior
destroy failed
Actual Behavior
but resource still exist
Important Factoids
References