How do I delete instances that are no longer used?

How do I delete instances that are no longer used, such as those marked in the red box in the picture?
These instances, I have removed on the client, using the “pmm-admin remove” command, but still on the page

Hi peng19832

There used to be an instruction in the documentation that said “The Dashboards tab lists all available dashboards. To deactivate any dashboard, select the Trash Bin button.” but I admit that I can’t see that now. Does this apply to you though?

Meanwhile, I will ask the team about it…

Hi again … can I check which version of PMM client you are using? I found this post which suggests there might be an issue with purge… https://www.percona.com/forums/questions-discussions/percona-monitoring-and-management/52807-how-can-i-delete-the-dashboard-of-the-hosts-that-are-no-longer-monitored

pmm-client:1.9.1
server:1.16.0

Hi peng19832

Once you remove the client packages and confirm the exporters are stopped, PMM Server will no longer be able to collect metric series from the host. Removing services does not purge the data from PMM Server - rather , this data will automatically expire as time goes on. If you don’t want to see these hosts in your dropdown Host lists, you’ll need to do two things: [LIST=1]
[]Select a time range that is AFTER the time you removed the services.
[
]Force the browser to refresh.
[/LIST] The reasons are that PMM Server will display metric series during a given interval if that metric series exists - so this is why you need to identify a time range after you removed the services. And our dashboards are set up to evaluate Host list dropdowns ONLY on dashboard reload (i.e. page reload). Hope this helps!

ok, I will try

@Michael CoburnSorry, after I followed the 2-step operation you said, the instance in the drop-down list that has been removed by pmm-admin still exists.

Hi peng19832

If you select say a 5 minute range, and force the browser to do a full refresh (important), you still see the instances in the dropdown list?

It isn’t clear to me why the nodes would then still show up, we use a Prometheus query to generate the drop-down list. This list is evaluated on dashboard load (browser refresh).

[IMG2=JSON]{“data-align”:“none”,“data-size”:“full”,“src”:“https://www.percona.com/forums/image/gif;base64,R0lGODlhAQABAPABAP///wAAACH5BAEKAAAALAAAAAABAAEAAAICRAEAOw==”}[/IMG2]​
You can also inspect Prometheus directly to determine if the data remains in the database. For example this query will tell me how many instances of MongoDB I have per cluster. In my case I have only one Cluster defined, and 11 nodes as part of it.

Can you run these and let us know what you see?

query-opts.PNG

Oh, thank you, this question has disappeared.

peng19832 do you mean it’s fixed for you now?

Yes, after waiting for 30 days (default), those display exceptions instances was removed from the drop-down list.