erick dagenais <[email protected]> 2:34 AM (10 hours ago) to *Brian* Hey Brian appreciate the answer,
**Copying over all message from direct message** ‘First, find some label or combination of labels which matches the process memory to the associated node memory - in this case it might just be the "instance" label.’ This is the exact problem I have though. Since I’m mixing metrics from different exporters (in this case ‘node-exporter’ and ‘process-exporter’, they have different output metrics, none which overlap in all of them. Specifically, process-exporter doesn’t have a job or instance label. This is where the dummy_label workaround came from. Since they are two metrics with no intersection, I created a fake intersection, I guess in your case a fake “instance” label named “dummy_label”. The creation of the dummy label is exactly what I’m trying to avoid here. On 03/10/2020 10:34, erick dagenais wrote: > Specifically, process-exporter doesn’t have a job or instance label. It *must* have both a job and an instance label, since prometheus adds these labels itself. https://prometheus.io/docs/concepts/jobs_instances/#automatically-generated-labels-and-time-series **New Message** Hey Brian, You're right I should have looked into these labels more, looks like I was summing without these labels to get rid of them. https://www.robustperception.io/controlling-the-instance-label The relabel config solution in this post is one solution I was looking for. The other one would be to make node-exporter and process-exporter be scraped under the same job, so that they have the same job label. I'll have to look into which solution is preferred, if either. I guess the real question I should have asked first, is why don't they have any labels in common in the first place. Thanks Brian for the help! On Sat, Oct 3, 2020 at 2:13 AM Brian Candler <[email protected]> wrote: > One other thing: to do this readily, you need to use "meaningful instance > labels": > https://www.robustperception.io/controlling-the-instance-label > > That is, rather than > > node_info{instance="foo:9100"} > process_info{instance="foo:9256"} > > you need to record > > node_info{instance="foo"} > process_info{instance="foo"} > > This is easy to arrange using relabelling at scrape time. Set the > instance label to what you want, and then set __address__ to the > address:port. Example: > > - job_name: node > scrape_interval: 1m > scrape_timeout: 50s > file_sd_configs: > - files: > - /etc/prometheus/targets.d/node_targets.yml # don't include the > port number in here > metrics_path: /metrics > relabel_configs: > - source_labels: [__address__] > target_label: instance > - source_labels: [__address__] > target_label: __address__ > replacement: '${1}:9100' > > -- > You received this message because you are subscribed to the Google Groups > "Prometheus Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/prometheus-users/74c30936-b7c8-4d4e-825b-c320db295707o%40googlegroups.com > <https://groups.google.com/d/msgid/prometheus-users/74c30936-b7c8-4d4e-825b-c320db295707o%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/CAFTp1BwwiTrY14nuyM6V_3%3DOj4b60HiYfSML3PMVQhQrM7qAnA%40mail.gmail.com.
