Different storage in different nodes

[PandaUppercut]

I am going to build a cluster with three different machines. Do all three machines need the exact same amount of storage or can they be diffferent?
IE
node 1 = 1 tb
node2 = 500 gb
node3 = 1 tb

 

[Dunuin]

Hard to tell without any information on what storage will be used etc.

 

[UdoB]

Technically it will work; "same size" is irrelevant as is the technology (NVMe/SSD/rotating rust).

Just make sure your main storage has exactly the same name. (By default and with ZFS it is "local" and "local-zfs" for example.)

Of course migration from one node to another can only work as long as all required ressources are "free" on the target node.

Have fun - and welcome to the club.

 

[carolyn.nardi]

If your cluster is used for distributed storage or processing (e.g., Hadoop, distributed databases, or storage clusters), having different storage capacities on each node can affect data distribution and data redundancy strategies. You need to plan accordingly to ensure data can be distributed and replicated effectively across nodes.
Different machines with varying storage capacities may also have different workloads or roles within the cluster. For example, a machine with more storage might be used for data storage, while a machine with less storage might be used for compute tasks. This can be a valid configuration if your workload supports it.