leaseweb knowledge base logo - white color on the orange background
leaseweb knowledge base logo - white color on the orange background

Limits & Known Issues

< Back

Limits & Known Issues

Known limits

Cluster

The following limits apply per cluster by default:

  • Worker nodes: 10 by default spread across node pools. You can request up to 100 by opening a support ticket
  • Pods per node: 110
  • Namespaces: 10,000
  • Services per namespace: 5,000
  • Services per cluster: 10,000

Networking

Each cluster includes 3 public IP addresses free of charge. The subnet is selected at order time. If you need to change your IP allocation or subnet after the cluster has been created, please contact our sales team by opening a ticket through the Customer Portal.

The total outbound bandwidth available to a cluster is 1 Gbps shared across all nodes. This is a current limitation of the CloudStack vRouter when using isolated networks.

Private networking is not yet available. As a result, traffic to and from Object Storage uses the public network and is therefore subject to standard bandwidth rates. Integration with Leaseweb File Storage and Block Storage is also not possible until private networking is in place.

There are known issues when exposing UDP-based traffic. Workloads relying on UDP — such as VOIP, media streaming, and live media — may experience instability. Contact support before deploying UDP-dependent workloads to production.

Storage & persistent volumes

The following limits apply to persistent volumes provisioned through the CSI driver:

Max volume size2 TB per volume
Max volumes per worker node32
Default storage quota per cluster3.5 TB — upgradable on request

The CSI driver does not support instant snapshots or volume encryption at rest. If encryption is a compliance requirement, it must be handled at the application level.

Volumes can only be mounted by a single node at a time (ReadWriteOnce). ReadWriteMany is not supported — a volume cannot be shared across multiple nodes or pods simultaneously.

Ecosystem & integrations

Leaseweb Object Storage and Leaseweb CDN are available. As noted above, Object Storage traffic is billed at public network rates until private networking is available.

Cloud Connect (powered by Megaport) is not currently available. This integration depends on private networking, which is on the roadmap.

Scheduling & node affinity

Worker nodes are deployed with soft anti-affinity by default. This means the scheduler attempts to spread worker nodes across different physical hypervisors. However, this is not guaranteed — if more worker nodes exist than available hypervisors, or if hypervisors are fully occupied, the scheduler may place multiple workers on the same physical host. Hypervisor maintenance and CloudStack Distributed Resource Scheduler (DRS) rebalancing may also move nodes during the cluster lifecycle.

Control plane components are deployed with hard anti-affinity. At least three copies of the etcd database are always placed on different physical hypervisors — two etcd members cannot share the same physical host.

Back to the top