Metrics endpoint not working

[ars3nicum]

I am trying to get metrics according to this instruction and getting error:
{"message":"Database name cannot be undefined.","status":500}

Budibase version 2.5.0

What am I doing wrong?

[andz-bb]

hey @ars3nicum

are you on cloud or a self hosted setup? If it's the latter, could you give some more details about your installation?

I've tested this on both cloud and self-hosted on v2.5.2 and the endpoint works ok on my instances. I believe this metrics endpoint is looking for your global-db on couch so could you check that it exists and is named properly? You can use our docs on accessing your CouchDB here for reference.

[ars3nicum]

I updated budibase (self-hosted) to version 2.5.2 and endpoint is working now, but if I try to curl it with the api-key header I get the following:

curl --request GET \
  --url https:/<mybudi>/api/public/v1/metrics \
  --header 'x-budibase-api-key: <myapi>'
Warning: Binary output can mess up your terminal. Use "--output -" to tell
Warning: curl to output it to your terminal anyway, or consider "--output
Warning: <FILE>" to save to a file.

From prometheus I also get an error: server returned HTTP status 403 Forbidden

[andz-bb]

@ars3nicum glad to hear you're a step closer with it.

For the curl you need to add the --compressed flag to it so that it decompresses the response. I saw the same on my end with the curl example in the docs so I will update it now to reflect this.

I'm not sure if there's a way to include the x-budibase-api-key header out of the box on the prometheus scrape_configs so that's maybe why you are getting a 403 in this case. I'm not an expert on prometheus config so I can't help you further on this one, although it's likely someone else online has a workaround for this by using a reverse proxy or similar to include the header on the request.

[ars3nicum]

Thank you! Curl also works.
As for the Prometheus there is example config, but for me not working.

[odcheck]

I guess it ain't working cause the http basic auth part, this is a working snippet for the scrape job tha I use:

  - job_name: 'budibase'
    scrape_interval: 30s
    scheme: http
    metrics_path: /api/public/v1/metrics
    static_configs:
      - targets: ['IP-ADDRESS:10000']
    authorization:
      type: "x-budibase-api-key"
      credentials: "API-KEY"

Of course as you can see I missuse type so that I will be used as header