Bunnyshell Kubernetes Cluster
Introduction
In order to test out the platform, Bunnyshell provides you with a Kubernetes cluster as a ready-made, no-effort playground cluster.
Each organization has the cluster already connected, with the name Bunnyshell cluster.
The Bunnyshell managed cluster should not be used for production workloads, as it has a number of limitations, explained below.
Pre-requisites
Please see how resources can be specified for Docker Compose Components.
In order to understand the limitations properly, the Pod concept from Kubernetes must be understood. The Pod is the smallest deployable unit in Kubernetes, and a Pod can have one or multiple containers.
Limits and defaults
Memory
If resources are not specified, the default memory reservation for a container (not for the whole Pod) is 500MiB per container, with the limit set at 750MiB.
For any given Pod, at most 3GiB of memory can be used, if requested. The maximum memory which can be used by a namespace is 10GiB.
CPU
If resources are not specified, the default CPU reservation is 0.25, with the limit set at 0.3.
For any given Pod, at most 1 CPU can be used, if requested. The maximum CPU which can be used by a namespace is 5.
Storage
The storage limit per organization is 10GiB, across all environments deployed on the Bunnyshell cluster.
Pods and namespaces
A maximum number of 10 Pods can be deployed on the Bunnyshell cluster for any given organization.
A maximum of 5 namespaces can be running simultaneously, meaning maximum number of 5 environments can be deployed and in Running
status, as each environment is deployed in its own namespace.
Deployments
At most 100 successful deployments can be made during 30 days on the Bunnyshell cluster. There is a rolling window, with day resolution.
There is a maximum TTL of 2h since the last deployment was made on an environment deployed on the Bunnyshell cluster. After this time passes since the last deployment, Bunnyshell will stop the environment.
Helm Charts and Kubernetes manifests are restricted (for now) from running on the Bunnyshell cluster, so you will need to connect your own cluster to deploy Helm Charts or Kubernetes manifests.
Updated 10 months ago