How to Enable Additional OS Metrics Collectors in PMM?

Hi Everyone,

I could see the pmm-agent status, Only few collectors taken from node exporter. How can I enable additional collectors like cpu_info ? And also how can I list the collectors that pmm-agent has taken by default?

● pmm-agent.service - pmm-agent
     Loaded: loaded (/lib/systemd/system/pmm-agent.service; enabled; vendor preset: enabled)
     Active: active (running) since Wed 2023-01-04 04:58:50 UTC; 18min ago
   Main PID: 10946 (pmm-agent)
      Tasks: 23 (limit: 1143)
     Memory: 46.1M
        CPU: 11.596s
     CGroup: /system.slice/pmm-agent.service
             ├─10946 /usr/sbin/pmm-agent --config-file=/usr/local/percona/pmm2/config/pmm-agent.yaml
             ├─10952 /usr/local/percona/pmm2/exporters/node_exporter --collector.bonding --collector.buddyinfo --collector.cpu --collector.diskstats --collec
1 Like

@Asuwini_P If you use ps instead of service, you’ll see that all collectors are enabled by default for the node_exporter. Please look at PMM UI to see if any are missing.

1 Like

ps is the most common tool for looking at running processes. Try ps -Af | grep node_exporter

1 Like

Thanks, @matthewb
Now I am able to get the collectors are enabled by default in PMM

How we can enable additional collectors in PMM?

1 Like

Which collectors are you interested in turning on? Have you verified using ps that the collector you want is not enabled?
You should also be aware that even if you enable another collector, there might not be a dashboard within PMM UI for the additional information. You would need to import or create a new dashboard to see the info.

1 Like

I am creating a custom dashboard in PMM that will display the CPU model, and it’s related metrics. --collector.cpu.info collector will get this information. I have checked that cpu.info is not enabled by default in PMM. In this case, how can we add the cpu.info collector to get this information?

1 Like

Hi @Asuwini_P

Sad to say but now there is no functionality to disable/enable collectors.
Additional metrics can be generated by already installed node_exporter if it would be run as an external service.

/usr/local/percona/pmm2/exporters/node_exporter --collector.disable-defaults --collector.cpu --collector.cpu.info 2>&1 &

pmm-admin add external --group=cpu  --metrics-path=metrics --listen-port=9100

Please notice that some cpu metrics would be duplicated and can be filter out by parameters agent_type=“external-exporter” or external_group=“cpu”

2 Likes

Thank you, @Vadim_Yalovets, I will definitely try it out

1 Like