How to Reclaim Disk Space on PMM 2 Docker Deployment (RHEL)

Hi Team,

We are currently using PMM Server 2 deployed on Docker in a RHEL Linux environment.

Recently, we noticed that the disk usage on the PMM server has been growing continuously and is starting to consume a large amount of storage space.

We would like to get recommendations and best practices on how to safely reclaim disk space and manage PMM storage growth.

Environment details:

  • PMM Version: PMM 2

  • Deployment Type: Docker container

  • Operating System: RHEL Linux

  • PMM is monitoring multiple database servers

We would appreciate guidance on the following:

  • How to identify which PMM components are consuming the most disk space

  • How to check Docker volumes used by PMM

  • Whether old metrics, logs, or monitoring data can be cleaned up manually

  • Recommended retention settings for Prometheus / VictoriaMetrics

  • Safe procedures to reclaim space without impacting monitoring

  • Any recommended Docker cleanup or PMM maintenance steps

  • Best practices for long-term storage management in PMM 2

If there are official documents or commands that can help investigate storage usage, please share them as well.

Thank you.

@Ly_Kimmeng

Let me provide some manuals and links that offer insight into data retention policy and ways to prune old metrics using VM downsampling.

Data retention:

You can limit or disable tablestats if not really needed:

Victoria Metrics Downsampling/pruning:

Well, you can also check this blog post to validate or remove any orphan containers or volumes that exist somehow.

Since PMM 2.41.0, ClickHouse was upgraded from version 21.3.20.1 to 23.8.2.7, introducing changes in how logs are written and retained. If you are using PMM ≥ 2.41.0, clickhouse records are inserted into log tables every second, which can cause rapid table growth.

Can you please confirm your exact PMM version?

Dear @anil.joshi ,

Thank you for your respond.

Here is our PMM server Version: 2.44.0.

If you require more information please let me know.

Best regards,

Kimmeng