Symptoms

The following alert is being observed:

Disk cache settings are not optimal

Disk <disk_name> (CS#<cs_id> on node <hostname> has cache settings different from other disks of the same tier.

Cause

This alert is aimed at notifying the user that a disk (CS) has different caching (journalling) settings than other disks in its tier. For example, one disk doesn’t have any caching, while all other disks in the tier use caching on SSD. It impacts on cluster performance, because CS without cache operates slower than CS with SSD cache. This alert can be considered as a performance warning. We recommend using the same caching configuration for all disks in a tier to reach the best performance.

A typical configuration where this error could occur is a cluster in which some nodes have SSD disks that are used for caching and some nodes don't have SSD. The best way is to install SSD disks on all nodes and recreate all CSs with caching on SSD.

If performance optimization is not required for your infrastructure or cluster is in middle of cache reconfiguration, you can ignore this alert.

Solution

1. To review the disks caching settings refer to the following documentation: Administration Guide - Monitoring node disks.

2. Gracefully release the issued disk mentioned in alert via WebCP and assign it again with correct caching settings (same as for other disks in the tier), alert is expected to be cleared out automatically in a while.