Un-deleted cluster resources may prevent future deployments.

[willgraf]

As noted in several other issues, some cloud resources may not be properly deleted, or if cluster deletion fails for some reason, they may be stranded.

The following resource types have prevented clusters from being created, and the resources had to be manually deleted from the Cloud console.

See here for a suggestion as to how this could be prevented or maintained, though it is an open problem.

• GKE LoadBalancers and Target Pools (LoadBalancers and Target Pools are not being deleted properly #162)

• GKE Service Accounts (Service Account Deletion Bug #113)

• GKE Persistent Disks (Persistent Disks are not deleted during cluster deletion #140)

• AWS VPCs (AWS too many VPCs error #67)

• AWS Launch configurations, Network Interfaces (Deleting instances in AWS #69)

[willgraf]

I want to relate this to #174 as the some of the resources (Persistent Disks especially) can cost money long after the cluster is deleted.

[willgraf]

I believe this has been largely resolved with more robust cluster destruction, specifically #199.

However, I may want to leave this issue up as a warning for future users.

[dylanbannon]

I think we're actually in a position to deal with the case where cluster creation or destruction fails and leaves resources stranded, which appears to be the only remaining edge case, so let's move this discussion to #274.