Open hasanaburayyan opened 6 months ago
Even if we run nuke scripts, we probably want wing test
to handle sigterm as this could happen even outside of our release process.
I'll just add a list @tsuf239 created at some point:
I currently trying to do a manual clean-up every 2 weeks, and I have scripts that can be adjusted and used daily (excluding the current day each time)
Hi,
This issue hasn't seen activity in 90 days. Therefore, we are marking this issue as stale for now. It will be closed after 7 days. Feel free to re-open this issue when there's an update or relevant information to be added. Thanks!
I already made one action for cleaning left out azure resources :) #6666 I'll make one for aws as well (and for gcp when will be included in the spec tests)
When the SDK Spec tests run they use Terraform to deploy infra, then once the tests complete run a Terraform destroy to clean up those resources.
In our current release workflow, we run the SDK spec tests as part of the release quality checks. However, since the workflow's behavior for consecutive releases is to cancel running jobs and start a new release workflow (bundling the releases together) this can result in terraform destroy commands never being run. Which leaves behind orphaned resources in our AWS account.
As well the state files used for these spec tests are using local backends so once the action is gone we cant even recover the state file to perform destroys manually, which creates a cumbersome cleanup process.
Possible Solutions
wing test
to handle sigterm for cleanupSlack thread: https://winglang.slack.com/archives/C04AUK4P5N2/p1710264284126939?thread_ts=1710264251.032359&cid=C04AUK4P5N2