How can I resolve the error "P99 write latency disk speed must be less than 10ms. A higher performance disk is required" when running preflight checks?

Issue Copied

When you run preflight checks, you may see an error indicating that P99 write latency exceeds the required threshold (10 ms):

sudo ./itrs-analytics install run-preflights --license license.yaml

This error indicates that disk I/O latency is too high. To confirm the problem on the node, check the k0scontroller logs:

journalctl -u k0scontroller --no-pager | grep "too long"

Example output:

Jan 21 09:26:57 myhostname k0s[5768]: time="2025-01-21 09:26:57" level=info msg="{\"level\":\"warn\",\"ts\":\"2025-01-21T09:26:57.025981Z\",\"caller\":\"etcdserver/util.go:170\",\"msg\":\"apply request took too long\",\"took\":\"3.742334451s\",\"expected-duration\":\"100ms\",\"prefix\":\"read-only range \",\"request\":\"key:\\\"/registry/leases/kube-node-lease/k0s-ctrl-myhostname\\\" \",\"response\":\"range_response_count:1 size:511\"}" component=etcd stream=stderr

In this example, the etcd apply request took 3.74 s, well above the expected duration of 100 ms or less.

You can also generate a support bundle. It includes disk latency metrics that help confirm the diagnosis.

Resolution Copied

ITRS Analytics requires SSD (Solid State Drive) storage for optimal performance. Ensure that SSD disks are present and that the required disk space is met. See Resource and hardware requirements for more information.