I have a Meteor app that uses Apollo and I am using SyncedCron (https://github.com/percolatestudio/meteor-synced-cron) to schedule an update operation on the database every 2 hours.
Every two hours, fetch X data from some external API and store in Y collection of the database. Rinse and repeat.
My question is this: should this update operation be implemented as a graphQL mutation, and called server-side from within the cron job (if so, how do you do it?!), or should this be implemented as a normal JS function since it will only ever execute on the server?
All the mutation examples I can find online are invoked from the client by wrapping your View component in a graphql-enabled HOC.
Perhaps I am just misunderstanding the scope of mutations, and the larger question here is whether or not mutations CAN be invoked from the server, or if they are client-side only.
Hoping to find some clarification here on mutation best practices. Thanks!