I have 10 node scylla-cluster,5 of these nodes are newly joined scylla-cluster ,but 5new node High load,5 old node load is normal
Please share more details, things like:
- Platform (physical/VM/cloud instance type/docker):
- Hardware: sockets= cores= hyperthreading= memory=
- Disks: (SSD/HDD, count)
- Scylla version (or git commit hash):
- Cluster size:
- OS (RHEL/CentOS/Ubuntu/AWS AMI):
- The data model
- What was the motivation for adding the additional nodes?
Platform:AWS-i3.4xlarge
Hardware:16C 122G
Disks:raid0 3.8T
Scylla-version:5.2.2
OS:centos7.5
RF:3
In the beginning have 5 node ,and than add 5new node in scylla-cluster, but only new node have high load and always play compaction very frequent.
just like:
avg ops 12k , max 30k
@huang can you clarify what is the graph it’s operation per sec ?
which clients are used ? java / python / rust ? are those scylla ones ? what are there versions ?
something is unbalanced, and for understand the cause we’ll need much more information about the load, we’ll need the schema, and the few example of the typical queries being done.
I think ScyllaDB University | Basic & Advanced Data Modeling goes into some details of how the schema should be built
More details are needed to understand what’s going on, things like:
- Schema used
- Installation details (ScyllaDB version, cluster size, OS)
- Hardware details: Platform (physical/VM/cloud instance type/docker), sockets= cores= hyperthreading= memory=, Disks (SSD/HDD, count)