using labels or pvc #218
-
Why should I use labels instead of the kubernetic way pvc. Is the sole reason simplicity? |
Beta Was this translation helpful? Give feedback.
Answered by
srikumar003
Feb 8, 2023
Replies: 1 comment 2 replies
-
Hi, we introduced labels for two reasons: 1) It makes pod declarations simpler; and 2) It enables different uses of datasets e.g. as environment variables, instead of mounting alone. You can use the PVC, if you prefer. But, in a future update, datashim may not generate PVCs with names same as datasets to cater for different use-cases e.g. a dataset that encompasses multiple data sources. |
Beta Was this translation helpful? Give feedback.
2 replies
Answer selected by
5cat
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi, we introduced labels for two reasons: 1) It makes pod declarations simpler; and 2) It enables different uses of datasets e.g. as environment variables, instead of mounting alone. You can use the PVC, if you prefer. But, in a future update, datashim may not generate PVCs with names same as datasets to cater for different use-cases e.g. a dataset that encompasses multiple data sources.