I deployed prometheus-community/kube-prometheus-stack which contains grafana.
I looked at the ConfigMap responsible for grafana.ini.
I set a new value in the grafana / grafana.ini section of the helm chart and deployed.
I expected the value in the ConfigMap to be updated, but it turned out to be old. At the same time, the value on the pod was new. After reconnecting to the cluster in Lens, the problem went away.