Configurable verbosity in HAProxy status script

Description

In clusters with a high number of PXC instances, the HAProxy status script generates a substantial volume of log entries. This can lead to increased costs when using certain log storage solutions. 

To help with this I propose implementing a VERBOSE environment variable that controls the logging behavior. When set, logs will only be generated if the cluster is in an unhealthy state. This feature will be disabled by default.

Ref: https://forums.percona.com/t/is-it-possible-to-set-haproxy-log-level/23782

Environment

None

AFFECTED CS IDs

CS0034102

Activity

Show:

Giovanni Tirloni April 18, 2024 at 7:33 PM

These messages are coming from /usr/local/bin/check_pxc.sh and the ability to turn them was merged in this PR.

Docker image percona/haproxy:2.8.5 contains the fix.

Aaditya Dubey November 30, 2023 at 10:34 AM

Hi ,

Sending the concern to engineering for further review and updates.

Ian Driver November 21, 2023 at 1:17 PM

@aaditya.dubey as far as I can see those existing variables do not have any effect on the HAProxy PXC check script.

Aaditya Dubey November 21, 2023 at 8:15 AM

Hi ,

Thank you for the report and feedback. Please check below report.

  • : Configuring the log structuring and leveling is now supported using the LOG_STRUCTURED and LOG_LEVEL environment variables. This reduces the information overload in logs, still leaving the possibility of getting more details when needed, for example, for debugging

Done

Details

Assignee

Reporter

Needs QA

Yes

Components

Fix versions

Priority

Smart Checklist

Created November 1, 2023 at 9:36 AM
Updated September 23, 2024 at 9:11 AM
Resolved April 18, 2024 at 7:41 PM