Discrepancy between the values of "BUFFER_CACHE_HIT_RATIO" SQL server metric collected by both SQLSERVER and NTPERF probes.

Products

• DX Unified Infrastructure Management (Nimsoft / UIM)

Issue/Introduction

In a scenario where both SQLSERVER and NTPERF probes were configured to collect and compare the values for BUFFER_CACHE_HIT_RATIO" SQL metric, you can see a big discrepancy between the values:

Cause

This is caused by the query used by SQLSERVER probe to fetch the values from SQL server.

Environment

sqlserver probe 5.42 and earlier

Resolution

Please contact the Broadcom Support Team, and open a support case to request the sqlserver_5.42T4 probe version. 

After deploying this probe version, the metric values will appear very closer on each other:




ATTENTION:  If you have configured any one of the following variables in the buf_cachehit_ratio checkpoint message, they will not be resolved:



This is because this new probe version uses a new query to fetch the values, from where these variables were removed. So, to avoid this problem, you can the sqlserver.cfg and remove them (take a backup first), as below:

from: 

msg_variables = $check.x;$profile.x;$instance.x;$hit_ratio.n;$page_lookups.i;$page_reads.i;$page_writes.i 




to: 

msg_variables = $check.x;$profile.x;$instance.x;$hit_ratio.n



After that, you must deactivate/activate the probe to make the changes have effect.