Protected datasets

natejgardner

It would be helpful if I could protect a dataset to prevent it from being overwritten without first unprotecting it. This would allow me to effectively lock a dataset while recipes are running against it or let collaborators know its current state is important and should not be overwritten right now. This would also prevent automations from overwriting a dataset if a team member needs to keep it static temporarily without needing to manually pick automations to disable. Implementation could be as simple as a lock button in the dataset flyout and a modal on unlock indicating which user locked it and when. Then, any time a user tries to execute an operation that would modify that dataset, an error would be triggered indicating the locked datasets and notifying the user they need to unlock them before trying again. Users could even be warned of locked datasets before executing pipelines and given the option to build only from the locked dataset down to prevent wasted time on lengthy builds.

Ashley

Thanks for your idea, @natejgardner

Your idea meets the criteria for submission, we'll reach out should we require more information.

If you’re reading this and would like to see lockable or protectable discussions in DSS, be sure to kudos the original post!

Take care,

Ashley

robersond

This kind of feature would be a helpful flag on the rebuild behavior settings for a dataset. I often have cross-database "soft" schema problems (as when writing to Oracle date fields or when going from decimal to float) and would much prefer if my flow would break when it feels these are intractable instead of just dropping a table that I have marked 'append only'. I lose data and lose the constraints and indexes on tables I had never intended Dataiku to be the owner of.