PMM2 push pull and scrape duration


I am a little confused on how the client work, I used to think it pushes data to the server, but I watched the talk by " Monitor & Manage your Databases with Percona Monitoring & Management PMM - Puneet Kala" and he said it can either push the data to the server or the server can pull the data from the remote database. So how to configure either of them.

As for the scrape interval, why do we have three number, low, medium high? how do they impact each client and do we have three of them not just a single scrape interval?


when you are adding service you can select which mode to use using --metrics-mode=pull or --metrics-mode=push. By default it’s push.

scrape intervals are used to collect different collectors(metrics) with different intervals, because some of them might be resource-intensive, so we collect them less often then others. it was done to not decrease performance on services.

1 Like

Thank you for your answer, so let me re-iterate what I understood:
1- not all metrics are pushed together (you guys seem to have split the metrics into three categories and they are pushed with the different scrape intervals).

2- if we use the push approach, the pmm-client (pulls the configuration from the pmm-server) and pushes the data according to the scrape intervals.

3- if we use the pull approach, the server itself reaches out to the nodes (according to the scrape intervals) and pulls the data from the pmm-client.

Did I get it right?
Thx again

Yes, that’s correct.

1 Like