Error, not enough memory - cluster with different instance types

ScyllaDB
,
1

Hi team,
Iam working with AWS cloud, deployed a cluster.
One node configured with like 2CPU and 8GB RAM (m6g.large-instance type) and another node configured with lower 1CPU and 4GB RAM (m6g.medium-instance type).

Error:
After deploying lower configured node that node is not up in the status error it is showing shard distribution error. That screenshots attached below.

Team, can you help on this,
It will work or not otherwise we need to keep same configuration on both the nodes.
please suggest.
Thanks,

systemctl status
systemctl status800×347 66.7 KB

2

First, running production clusters with different instance types is possible, but:

  • hard to manage
  • the cluster will work as fast as the slowest node
    (so not recommended)

Second, the error code clearly state you do not have enouth RAM per core. You need to either increase the memory, or reduce the number of core.

3

Thanks @tzach.
But Iam using for development cluster for data standby.
it is ok?

4

Sure, if it’s non-production, and you understand the risk.
Almost all the tests we are doing are for clusters with symmetric nodes. Asymmetric nodes with different numbers of cores, should work, but they are less tested.