Prometheus -> 9 Host = 28.5g ? PMM v2


I used PMM 2 Appliance and /usr/sbin/prometheus use 28.5g of RAM for 9 host (9 pmm-clients mysql).

Can you tell my where is the problem?

is there any option with version 2 i missing?

Best Regards.

My config is :

CentOS Linux release 7.8.2003 (Core)

Linux dc1vm00084 3.10.0-1127.13.1.el7.x86_64 #1 SMP Tue Jun 23 15:46:38 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

PMM Version => Current version: 2.11.1 (October 19, 2020)

All Hots (Mysql) using :

PMM Server: Version: 2.10.1

PMM Client:

   Connected       : true

   Time drift      : 1.534497485s

   Latency         : 2.146778ms

   pmm-admin version: 2.11.1

   pmm-agent version: 2.11.1


   /agent_id/67e4d73c-e668-4971-b40b-ab859c775501 mysql_perfschema_agent Running

   /agent_id/bee0c1a4-450c-42c7-921f-36271139cbc0 node_exporter Running

   /agent_id/d9ac2bff-5398-4a34-93bc-eb6abe76e637 mysqld_exporter Running

TOP Serveur PMM

TOP pmmv2.JPG

Hi @mramirez ,

Here is a post about used RAM by Prometheus.

Could you provide a screenshot of amount of time series in your installation? It’s shown on the dashboard “Prometheus”, row “Time Series Information”.

Amount of time series directly depends on used metrics resolution. It can be checked or changed on the tab “PMM Settings”.

Hi @adivinho

thanks for you feedback.

Here is my screenshot of Setting Metrics.


And here is Time Series Information

Thank you for the additional information.

Big amount of time series and active dashboard viewing could impact RAM usage.

Prometheus will be replaced by VictoriaMetrics for metrics storing in the next pmm-server release. It will significantly reduce memory usage.

pmm-server 2.12 is planed to be released on the next week.

Thanks you for information

I look forward to install next released on the next week.

Best regards

How do you see what Prometheus takes so much memory ?

Prometheus often reports very high VM usage, sometimes 10x or more compared to the actual memory it really uses due to its internal design

Hi Peter.

Here is my PMM process Memory.

I restarted server PMM last friday and Memory seems reasonable now, maybe memory leak bugs?