Open framitdavid opened 8 months ago
Since we already using a posgress database we can use advisory locks to create our distributed locking mechanism implementation.
Currently not being able to scale Designer is a fundamental issue in our solution. I suggest adding this to sprint 51 or 52.
We can start with creating abstractions and implementations both for Postgres (distributed) and in-memory (non-distributed) and see how it behaves.
@nkylstad @framitdavid
We need to implement support for locking files that create or manipulate files. This is to avoid multiple endpoints working on the same files simultaneously.
Note: Can we implement our own locking mechanism initially without considering whether the locking is local or distributed? So we can add distributed caching in the next step once we figure out where to store the distributed cache?