MongoDB likely still has a side where it's better than PostgreSQL: clustering.
Running a multi-node cluster is a built-in feature (no patroni bullshit around) and you can ask on a per-query basis for consistent (acknowledged by a majority of the nodes) or non-consistent (do not require data to have been acknowledged by a majority of the nodes) values. Same for transaction (eg: ask for a transaction to return as completed only after 2 out of 3 nodes in the replicaset have acknowledged it).
It's sad to see PostgreSQL team ignoring this aspect for so long, it's really the missing piece to make PostgreSQL the definitive database.
I still dream of some built-in multi-master postgresql solution.
Running a multi-node cluster is a built-in feature (no patroni bullshit around) and you can ask on a per-query basis for consistent (acknowledged by a majority of the nodes) or non-consistent (do not require data to have been acknowledged by a majority of the nodes) values. Same for transaction (eg: ask for a transaction to return as completed only after 2 out of 3 nodes in the replicaset have acknowledged it).
It's sad to see PostgreSQL team ignoring this aspect for so long, it's really the missing piece to make PostgreSQL the definitive database.
I still dream of some built-in multi-master postgresql solution.