How to name foreach pipelines?

Grafana Alloy
1

I’m trying to use the experimental foreach block to read many files from S3 buckets. Each file will be used as the targets for a different Prometheus scrape component. But I can’t find a way to name each iterated component. In this example, they are all named remote.s3.default so I’m unable to access them.

foreach "s3" {
  var = "pipeline_name"
  collection = ["snmp_christie4415rgb", "qsys_core", "ping",]
  template {
    remote.s3 "default" {
      path = string.format("s3://mybucket/%s.yml", pipeline_name)
      poll_frequency = "1m"
    }
  }
}

prometheus.scrape "christie4415rgb" {
  forward_to      = [prometheus.relabel.christie4415rgb.receiver]
  targets         = encoding.from_yaml(remote.s3.christie4415rgb.content)
  scrape_interval = "1m"
}

I also tried putting the scrape component inside the foreach block but then I would need to insert the variable in the targets list, and I don’t think it’s possible to construct capsule values. I tried like this:

    prometheus.scrape "christie4415rgb" {
      forward_to      = [string.format("prometheus.relabel.%s.receiver", pipeline_name)]
      targets         = encoding.from_yaml(remote.s3.default.content)
    }

And got this error:

string.format(“prometheus.relabel.%s.receiver”, pipeline_name) should be capsule, got string

2

I tried a different approach, and found that the foreach variable also doesn’t work when used as a key for json_path. With this config:

local.file "targets" {
  filename = "targets.json"
}

foreach "default" {
  var = "each"
  collection = json_path(local.file.targets.content, each)
  template {
    prometheus.scrape "default" {
      forward_to = [prometheus.remote_write.local.receiver]
      targets    = encoding.from_json(json_path(local.file.targets.content, string.format("$.%s[*]", each)))
    }
  }
}

It shows error:

Error: test.alloy:7:54: identifier “each” does not exist

3

Hello, I think that in your case you would need both the scrape and the relabel component in the foreach and have the relabel component to point to a remote_write component outside of the foreach like this:

foreach "s3" {
  var = "pipeline_name"
  collection = ["snmp_christie4415rgb", "qsys_core", "ping",]
  template {
    remote.s3 "default" {
      path = string.format("s3://mybucket/%s.yml", pipeline_name)
      poll_frequency = "1m"
    }
    prometheus.scrape "default" {
      forward_to      = [prometheus.relabel.default.receiver]
      targets         = encoding.from_yaml(remote.s3.default.content)
      scrape_interval = "1m"
    }
    prometheus.relabel "default" {
      forward_to = [prometheus.remote_write.common.receiver]
    }
  }
}

prometheus.remote_write "common" {
  endpoint {
    url = "http://localhost:9009/api/v1/push"
  }
}
4

Thanks for the suggestion! That won’t work for me because each scrape needs a different set of relabel rules.
I’ve worked out a way to handle this logic in my own script instead, so I poll S3 and download new files if their Etag has changed, write them to local files, which Alloy sees with several discovery.file components.