https://kubernetes.io/docs/concepts/extend-kubernetes/operator/
You might be exploring a tool and find a section to "Run this as an operator in
your Kubernetes cluster". That sounds swell, you run one magic command, some
make install and now you can define some config to just make things work.
On paper that sounds good, but there are few unspoken considerations.
Hello :megaphone:
dev-merit has a new CA certificate, please update locally running $ make ca-certs-install from inside kubernetes-manifests/ (must be on PA VPN/Wiresteward)
Verify new expiry date:
$ make ca-certs-dates | grep -A1 dev-merit
dev-merit
Not After : Jan 22 00:35:00 2026 GMT
We have an s3 bucket with an inline resource-based policy granting full access to an iam user and an iam role. Both principals (user and role) have no identity-based policies and are bounded by a policy allowing only sts:GetCallerIdentity. Since resource-based policies are not affected by permissions boundaries, both principals should be able to get objects from the bucket. However, only the user can, the role gets "Access Denied". If we remove the permission boundary from the role, then it can get the object without issue. This is the behaviour we find confusing
We are assuming the role via aws cli, so I don't think session policies are getting in the way (plus removing the boundary from the role fixes the problem, so the boundary seems to be the entity blocking access).
Below are the details of the scenario. Let me know if you need anything else, either by email or slack. Thanks!
Details:
Bucket policy:
| #!/bin/bash | |
| ## usage: ./rs-ns <kube-context> | sort -h -k 2 -r | |
| export env=${1} | |
| total() { | |
| ns=${1} | |
| echo "${ns} $(kubectl --context=${env} -n${ns} top pod | tail -n +2 | awk '{print $3}' | cut -d'M' -f1 | paste -s -d+ - | bc) " | |
| } |