K8SPXC-1650 delete resources in azure #3646
Conversation
nmarukovich
requested review from
eleo007,
hors,
jvpasinatto and
valmiranogueira
as code owners
…rcona-Lab/jenkins-pipelines into K8SPXC-1650_delete_resources_in_azure
cloud/azure/README.md
Outdated
| @@ -0,0 +1,7 @@ | |||
| # To redeploy function run in cmd folder: | |||
There was a problem hiding this comment.
Can we add to README where this script will be executed, where to check its logs, how to run it.
From DM: this is the only way to redeploy the functions.
It would be to have here smth like:
- The code is deployed and executed at (link)
- Logs of execution are (link/how to check if not obvious from the 1st)
- The script can not be redeployed/changed through interface. To redeploy download the script (how?) and run below (where?)
| @@ -0,0 +1,169 @@ | |||
| # Remove expired AKS clusters (Azure, cluster-only) | |||
There was a problem hiding this comment.
Nice to have for future:
- There is a case where the cloud tag is present but the TTL is not correct (creatio time is missing/incorrect format). It would be good to receive a notification about such clusters to recheck why this happens and maybe update scripts corrspondingly.
| @@ -0,0 +1,14 @@ | |||
| 1. In the Azure portal, search for **Function App** (make sure the subscription is set to **eng-cloud-dev**). | |||
There was a problem hiding this comment.
Can we add some intro, like: The cleanup scrips runs in Azure, to check it: do and then the steps
| resource_groups_client = ResourceManagementClient(credential, subscription_id) | ||
| aks_client = ContainerServiceClient(credential, subscription_id) | ||
|
|
||
| logging.info("Searching for AKS clusters to remove.") |
There was a problem hiding this comment.
Just to ensure consistency, we are using 3 terms for the same thing (remove, delete, terminate) can we stick to just one and apply in the whole file?
| logging.info("[DRY-RUN] Would delete cluster %s/%s", rg_name, cluster_name) | ||
| return | ||
|
|
||
| aks_client.managed_clusters.begin_delete(rg_name, cluster_name) |
There was a problem hiding this comment.
it seems that this method returns https://learn.microsoft.com/en-us/python/api/azure-core/azure.core.polling.lropoller?view=azure-python
Any specific reason to have wait_for_cluster_delete and not use this?
aks_client.managed_clusters.begin_delete(rg_name, cluster_name).result(timeout=300) or
aks_client.managed_clusters.begin_delete(rg_name, cluster_name).wait(timeout=300)
jvpasinatto
changed the title
PR provides the automatic deletion aks resources based on tags.
To use this script for your aks cluster please add tags during cluster creation:
--tags team=cloud delete-cluster-after-hours=1 creation-time=$(date -u +%s)