I think there is no other way. 

In any case, here is how I usually configure my (all-purpose and jobs compute) clusters to access secrets via environment variables so that you don't have to update all references if some similar issue arises again. The environment variables act as an abstraction between secrets and consumer code.

KR