Hardware and Software Specifications
OS Drive Partitions
Table 1 lists the recommended OS drive partitions.
Partition |
Description |
Size |
Boot Flag |
|---|---|---|---|
/boot |
Boot partition |
500 MB |
on |
/var |
Partition for storing data generated during software running, such as logs and caches |
> 300 GB |
off |
/var/lib/docker |
Partition for storing Docker images and containers NOTE:
Docker images and containers are stored in the /var/lib/docker partition by default. If the usage of the /var/lib/docker partition is greater than 85%, Kubernetes automatically triggers the resource eviction mechanism. Ensure that the usage of the /var/lib/docker partition is less than 85%. |
> 300 GB |
off |
/etc/mindx-dl |
Partition for storing imported certificates and KubeConfig files. Adjust the partition as required, and 100 MB is recommended. |
100 MB |
off |
/ |
Primary partition |
> 300 GB |
off |
Hardware Specifications
Resource |
Requirements |
|---|---|
CPU |
Management node CPU > 32 cores |
Memory |
Management node memory > 64 GB |
Drive space |
> 1 TB For details about the drive space plan, see Table 1. |
Network |
|
Resource Configuration Requirements of Cluster Scheduling Components
The resource configuration of cluster scheduling components must meet the following requirements.
Component |
< 100 Nodes |
< 500 Nodes |
< 1000 Nodes |
|||
|---|---|---|---|---|---|---|
CPU (Core) |
Memory Size (GB) |
CPU (Core) |
Memory Size (GB) |
CPU (Core) |
Memory Size (GB) |
|
Volcano Scheduler |
2.5 |
2.5 |
4 |
5 |
5.5 |
8 |
Volcano Controller |
2 |
2.5 |
2 |
3 |
2.5 |
4 |
Ascend Operator |
2 |
2.5 |
2 |
3 |
2.5 |
4 |
ClusterD |
1 |
1 |
2 |
2 |
4 |
8 |
Component |
CPU (Core) |
Memory Size (GB) |
|---|---|---|
Ascend Device Plugin |
0.5 |
0.5 |
NodeD |
0.5 |
0.3 |
NPU Exporter |
1 |
1 |
Ascend Docker Runtime |
Docker service plugin, which does not require separate CPU resources and memory space. |
|