Modifying the ephemeral storage limit

The default ephemeral storage limit for a new PX Runtime instance is 10 GB. If your cluster has additional storage available, you can increase or remove the limit.

To plan the optimal size of the ephemeral storage for a compute pod, you can consider the amount and type of data that is processed during DataStage jobs. The size of ephemeral storage depends on the types of jobs that run, the amount of data that the jobs are processing and the concurrency of the jobs on the same instance.

To check your ephemeral storage usage, use the following command: 
$ oc get --raw "/api/v1/nodes/worker0.tahoetest882.cp.fyre.ibm.com/proxy/stats/summary"
See the following example output:
... 
  "ephemeral-storage": {
    "time": "2025-03-06T21:48:21Z",
    "availableBytes": 71087955968,
    "capacityBytes": 267830407168,
    "usedBytes": 408403968,
    "inodesFree": 125135426,
    "inodes": 130809280,
    "inodesUsed": 3288
   },

Ephemeral storage comes from the cluster worker node that the pod is running on. If the limit is raised higher than the amount of ephemeral storage available on the worker nodes, a worker node may run out of storage and cause stability issues. To avoid stability issues, you can add local storage. For more information, see Adding local scratch devices to DataStage pods.

To increase the ephemeral storage limit to 20 GB, use the following command. 
oc -n ${PROJECT_CPD_INST_OPERANDS} patch pxruntime <cr-name> --patch '{"spec":{"ephemeralStorageLimit": "20Gi"}}' --type=merge
To remove the ephemeral storage limit, use the following command. 

oc -n ${PROJECT_CPD_INST_OPERANDS} patch pxruntime <cr-name> --type='json' -p='[{"op": "remove", "path": "/spec/ephemeralStorageLimit" }]'