PMM 2 - pmm-admin allow disable collection of metrics for mysqld_exporter

Description

Currently there is no way to disable the collection of a specific metrics for the mysqld_exporter.

In my case when client logging is enabled, the log get's spammed by entries like:

2019-11-14 15:05:46 root[root] @ localhost [127.0.0.1] ERROR 1286: Unknown storage engine 'TOKUDB' : SHOW ENGINE TOKUDB STATUS 2019-11-14 15:06:46 root[root] @ localhost [127.0.0.1] ERROR 1054: Unknown column 'FILE_FORMAT' in 'field list' : SELECT SPACE, NAME, ifnull(FILE_FORMAT, 'NONE') as FILE_FORMAT, ifnull(ROW_FORMAT, 'NONE') as ROW_FORMAT, ifnull(SPACE_TYPE, 'NONE') as SPACE_TYPE, FILE_SIZE, ALLOCATED_SIZE FROM information_schema.innodb_sys_tablespaces 2019-11-14 15:06:46 root[root] @ localhost [127.0.0.1] ERROR 1146: Table 'heartbeat.heartbeat' doesn't exist : SELECT UNIX_TIMESTAMP(ts), UNIX_TIMESTAMP(NOW(6)), server_id from `heartbeat`.`heartbeat` 2019-11-14 15:06:46 root[root] @ localhost [127.0.0.1] ERROR 1286: Unknown storage engine 'TOKUDB' : SHOW ENGINE TOKUDB STATUS

It would be great to have the possibility to disable some metrics via pmm-admin command.

 

Regards,

Patrick

How to test

None

How to document

None

AFFECTED CS IDs

275474, 275980, 276166, CS0024820

Smart Checklist

Activity

Show:

Nurlan Moldomurov July 22, 2024 at 2:14 PM

duygu.aksoy February 10, 2023 at 7:02 AM

Hey Thank you for bringing this to our attention. It was not previously on our radar. I will make plans for it.

Roma Novikov April 19, 2021 at 1:53 PM

For cases when it's known what's making the noise we added https://perconadev.atlassian.net/browse/PMM-7306#icft=PMM-7306
However, it's not a replacement for the full automatic collector disabling based on the parameters of the service PMM is monitoring

Soumya K August 6, 2020 at 4:50 PM

We're affected by this too after upgrading to PMM2.

With PMM1, it was easy to disable collection of certain metrics by passing them to pmm-admin. For example to disable performance_schema related metrics collection (since we don't use it in some environments), we used:

sudo pmm-admin add mysql:metrics --user <username> --password <password> --socket <socket> -- --collect.perf_schema.tablelocks=false --collect.perf_schema.tableiowaits=false --collect.perf_schema.file_events=false --collect.info_schema.query_response_time=false --collect.perf_schema.eventswaits=false;

 

Echo'ing Michael's comment - logging to /var/log/pmm-metrics.log, as with PMM1, kept the logs separate and tidy.

 

Mike Shield August 4, 2020 at 9:23 AM

I have the same issue - it 's an error as I know already which engines I've chosen to use. Also, while the Centos 6 agent logs to a separate file, Centos 7 logs to messages, which just adds to the existing swirl. It would be nice to get that back

Done

Details

Assignee

Reporter

Priority

Needs QA

Needs Doc

Affects versions

Smart Checklist

Created November 14, 2019 at 3:10 PM
Updated November 13, 2024 at 9:45 AM
Resolved July 22, 2024 at 2:14 PM

Flag notifications