N replications across all nodes, always

I recently came across a use case where I think cockroach may be a good fit.
I got a bit lost in the persuit of answers to the following questions, and I hope you guys can help me figure out whether cockroachdb is a good fit for me.

I have a case of N desktops that would like to work on the same db, all in the same LAN
Running a single DB on any of them (like mysql) is not a good fit, as those are eventually laptops all of them belong to employees that goes in and out of the LAN.
They are all working on a desktop app, that, in the best case scenario, is always writing to a local cockroach db, that is part of the cluster of the LAN and the data is always replicated to everyone else to see the data instantly.

Laptops comes and goes, and I would want data to be fully replicated to all nodes always, such that even if a single node remains, he is the leader and is able to operate alone.
If other laptops join the LAN, they join the cluster and get the replicated data from the single previous instance so that he can go offline and the data survives via the other nodes.

Is that conceivable with cockroachdb?
If I follow the docs for simple cluster of 3 nodes, then If I want to increase it to 5 nodes, I need to increase the replication factor to 5 since otherwise two nodes going down will cause an unavailable range.
While I still dont entirely understand what a range is, I do understand that its not a healthy cluster :slight_smile:

Hey @SillyPosition

In order for Cockroach DB to be available, a majority of the nodes in the cluster need to be able to communicate with one another at all times. So in the scenario you mentioned, where a single instance is alone, the data would not be available as the cluster is no longer accessible. A range is essentially a portion of the data, and part of the process of Cockroach DB is to spread the ranges across the cluster, on multiple nodes. These copies are called replicas so that in case one of the nodes that have the range is missing, there are other copies available.

However, without the majority of the nodes, the cluster is unable to provide consensus, and so the data would be unavailable.

Let me know if you have any other questions regarding your scenario in question.

Cheers,
Ricardo