Data Disk Space Allocation

Allocating Data Disk Space

When creating a node, configure data disks for the node. You can also click Expand and customize the data disk space allocation for the node.

Figure 1 Allocating data disk space

• Space Allocation for Container Engines
  • Specified disk space: CCE divides the data disk space for two parts by default. One part is used to store the Docker/containerd working directories, container images, and image metadata. The other is reserved for kubelet and emptyDir volumes. The available container engine space affects image pulls and container startup and running.
    • Container engine and container image space (90% by default): stores the container runtime working directories, container image data, and image metadata.
    • kubelet and emptyDir space (10% by default): stores pod configuration files, secrets, and mounted storage such as emptyDir volumes.
  • Shared disk space: In clusters of v1.21.10-r0, v1.23.8-r0, v1.25.3-r0, or later versions, CCE allows a container engine (Docker/containerd) and kubelet components to share data disk space.
• Space Allocation for Pods: indicates the basesize of a pod. You can set an upper limit for the disk space occupied by each workload pod (including the space occupied by container images). This setting prevents the pods from taking all the disk space available, which may cause service exceptions. It is recommended that the value is less than or equal to 80% of the container engine space. This parameter is related to the node OS and container storage rootfs and is not supported in some scenarios. For details, see Mapping Between OS and Container Storage Rootfs.
• Write Mode
  • Linear: A linear logical volume integrates one or more physical volumes. Data is written to the next physical volume when the previous one is used up.
  • Striped: available only if there are at least two data disks. A striped logical volume stripes data into blocks of the same size and stores them in multiple physical volumes in sequence. This allows data to be concurrently read and written. A storage pool consisting of striped volumes cannot be scaled-out.

Space Allocation for Container Engines

For a node using a non-shared data disk (100 GiB for example), the division of the disk space varies depending on the container storage Rootfs type Device Mapper or OverlayFS. For details about the container storage Rootfs corresponding to different OSs, see Mapping Between OS and Container Storage Rootfs.

• Rootfs (Device Mapper)
  By default, the container engine and image space, occupying 90% of the data disk, can be divided into the following two parts:

  • The /var/lib/docker directory is used as the Docker working directory and occupies 20% of the container engine and container image space by default. (Space size of the /var/lib/docker directory = Data disk space x 90% x 20%)
  • The thin pool is used to store container image data, image metadata, and container data, and occupies 80% of the container engine and container image space by default. (Thin pool space = Data disk space x 90% x 80%)

    The thin pool is dynamically mounted. You can view it by running the lsblk command on a node, but not the df -h command.

  Figure 2 Space allocation for container engines of Device Mapper
  

• Rootfs (OverlayFS)

  No separate thin pool. The entire container engine and container image space (90% of the data disk by default) are in the /var/lib/docker directory.

  Figure 3 Space allocation for container engines of OverlayFS
  

Space Allocation for Pods

The customized pod container space (basesize) is related to the node OS and container storage Rootfs. For details about the container storage Rootfs, see Mapping Between OS and Container Storage Rootfs.

• Device Mapper supports custom pod basesize. The default value is 10 GiB.
• In OverlayFS mode, the pod container space is not limited by default.

When configuring basesize, consider the maximum number of pods on a node. The container engine space should be greater than the total disk space used by containers. Formula: the container engine space and container image space (90% by default) > Number of containers x basesize. Otherwise, the container engine space allocated to the node may be insufficient and the container cannot be started.

Figure 4 Maximum number of pods

For nodes that support basesize, when Device Mapper is used, although you can limit the size of the /home directory of a single container (to 10 GB by default), all containers on the node still share the thin pool of the node for storage. They are not completely isolated. When the sum of the thin pool space used by certain containers reaches the upper limit, other containers cannot run properly.

In addition, after a file is deleted in the /home directory of the container, the thin pool space occupied by the file is not released immediately. Therefore, even if basesize is set to 10 GB, the thin pool space occupied by files keeps increasing until 10 GB when files are created in the container. The space released after file deletion will be reused but after a while. If the number of containers on the node multiplied by basesize is greater than the thin pool space size of the node, there is a possibility that the thin pool space has been used up.

Mapping Between OS and Container Storage Rootfs

Garbage Collection Policies for Container Images

When the container engine space is insufficient, image garbage collection is triggered.

The policy for garbage collecting images takes two factors into consideration: HighThresholdPercent and LowThresholdPercent. Disk usage exceeding the high threshold (default: 80%) will trigger garbage collection. The garbage collection will delete least recently used images until the low threshold (default: 70%) is met.

Recommended Configuration for the Container Engine Space

• The container engine space should be greater than the total disk space used by containers. Formula: Container engine space > Number of containers x basesize
• You are advised to create and delete files of containerized services in local storage volumes (such as emptyDir and hostPath volumes) or cloud storage directories mounted to the containers. In this way, the thin pool space is not occupied. emptyDir volumes occupy the kubelet space. Therefore, properly plan the size of the kubelet space.
• You can deploy services on nodes that use the OverlayFS (for details, see Mapping Between OS and Container Storage Rootfs) so that the disk space occupied by files created or deleted in containers can be released immediately.

Data Disk Shared Between a Container Engine and kubelet Components

Docker/containerd and kubelet components share the space of a data disk.

• This function is available only to clusters of v1.21.10-r0, v1.23.8-r0, v1.25.3-r0, or later versions.
• If Rootfs is set to OverlayFS, shared data disks are supported. If Rootfs is set to Device Mapper, shared data disks are not supported.
• If you have installed an NPD add-on in the cluster, upgrade the add-on to v1.18.10 or later. Otherwise, false alarms will be generated.
• If you have installed a log-agent add-on in the cluster, upgrade the add-on to v1.3.0 or later. Otherwise, log collection will be affected.
• If you have installed ICAgent in the cluster, upgrade it to v5.12.140 or later. Otherwise, log collection will be affected.

Figure 5 Configuration for sharing disk space

For nodes using a shared data disk, the container storage Rootfs is of the OverlayFS type. After such a node is created, the data disk space (for example, 100 GiB) will not be divided for the container engines, container images, and kubelet components. The data disk is mounted to /mnt/paas, and the storage space is divided using two file systems.

• dockersys: /mnt/paas/runtime
• Kubernetes: /mnt/paas/kubernetes/kubelet

Figure 6 Allocating the storage space of a shared data disk