Closed clincha closed 11 months ago
This looks useful for testing out the disks: https://access.redhat.com/solutions/4885641
Ceph HDD
Ceph SSD
Local NVME
After setting the MTU to 9000 I'm seeing inconsistent results with the SSD pool.
First run
INFO: 99th percentile of fsync is 371195904 ns
WARN: 99th percentile of the fsync is greater than the recommended value which is 371195904 ns > 20 ms, faster disks are recommended to host etcd for better performance
Second run
INFO: 99th percentile of fsync is 299892736 ns
WARN: 99th percentile of the fsync is greater than the recommended value which is 299892736 ns > 20 ms, faster disks are recommended to host etcd for better performance
Third run
INFO: 99th percentile of fsync is 19267584 ns
INFO: 99th percentile of the fsync is within the recommended threshold: - 20 ms, the disk can be used to host etcd
Running on local-lvm for now. That space is easily expandable and quick but makes the guests sticky to a host. This is fine with Kubernetes managing applications
I'm hitting issues that I believe are caused by high disk latency. This needs to be resolved so that the Kubernetes cluster is performing well for the applications that sit on it.