Cp rest proxy metrics

Hello forum,
I am once again reching out :slight_smile:

I have set up a rest proxy deployment on kubernetes which ia working as expected. Now i would like to retrieve some JMX metrics, so that i can feed them in to elasticsearch.
However, i have been unable to dechipher how this is achieved.
Basically i just want to enable the metrics, but i am unsure if the are actually enabled by default, or if i have to set some parameter in my deployment.
And once they are enabled, do i then need to either mount in a jolokia jvm agent, or a prometheus jmx exporter?

Best regards
Oelsner

hi @Oelsner

welcome back :wink:
just to be sure:
you like to get onl yrest proxy metrics or metrics from all your services deployed?
how did you deploy your environment?
using confluent helm charts or something else?

best,
michael

Hi @mmuehlbeyer
I would like to get just the metrics from rest proxy as described here Monitoring REST Proxy | Confluent Documentation. At least for now. Later on i would probably like to get metrics from other deployments.
I created “read stole from left and right ideas”/inspired by this one https://gitlab.com/atkozhuharov/kafka-rest-proxy-kubernetes/raw/master/kafka-proxy-deployment.yaml

The rest proxt is working just fine after a few minor changes to suit my needs. But i cant for the life of me figure out how to fetch those metrics. One suggestion i came across was this one https://gitlab.com/atkozhuharov/kafka-rest-proxy-kubernetes/raw/master/kafka-proxy-deployment.yaml
I therefore tried to add

env:
 - name: JMX_PORT
    value: "9011"

to the deployment in hopes that it would expose the metrics on that port as i had been informed that the afore mentioned metrics are exposed by default.

Any help on this is greatly appreaciated :slight_smile:

Best regards
Oelsner

hi @Oelsner

ok I see.
did you consider using the confluent rest proxy helm chart?

whole jmx stuff is already in place there
maybe you could use it as a template and adapt your conf :wink:

best,
michael

Hi @mmuehlbeyer
I actually did not even consider it, but i just checked the link you gave me.
It appears that it has a dependency to a running and accissible zookeeper ensemble, which i cant honor.
This rest proxy i ment to work with confluent cloud, where is dont have any url’s to any zokeepers.
I will try and see if i can find inspiration from the chart though and see if there are some things i can incorporate.

Cheers for now :slight_smile:
Oelsner

hi @Oelsner

ah i see.
according to your provides examples it maybe sufficient to add the prometheus exporter part and the port.

will test by myself and keep you posted :slight_smile:

best,
michael

hi @Oelsner

short update.
was able to connect my local test stack (docker-based) to my confluent cloud trial.
converted the compose file with kompose to some kubernetes usable files.
need to sanitize them :wink: could provide them afterwards, guess it might help

best,
michael

Hi @mmuehlbeyer
That sounds awesome!
I would be very interested in seing them once sanitized :slight_smile:

And thanks so much for taking your time to help with this.

Regards
Oelsner

Hi @Oelsner

you’re welcome

have a look at:

I guess not perfect maybe a good point to start with :slight_smile:

best,
michael

Hi @mmuehlbeyer
Thanks so much for that. It looks interesting, and very different as to how i have created my own deployment :slight_smile: , i cant wait to try it out.
I will report back here tomorrow in the evening with my findings… Now it is time to sleep :slight_smile:

Once again for your time, i will keep you posted!

Best regards
Oelsner

1 Like

Hi @mmuehlbeyer
Sorry about the latency.
I was delegated to other tasks :confused:
Your provided solution deploys just fine, but i have been unable to connect to jmx.
I will dig more into this d.uring the week. This week is easter holiday/playtime, so i can commit some time to it :slight_smile:

I will let you know in a day or two how it goes!

Again, thanks so much for taking the time to help with this.

/Oelsner

@mmuehlbeyer
I once again have to appologize for the latency.
Tasks piled up, and product owner asked me to shift my attention to an immidiate delivery.
However, your solution totally applies, but i still have to verify metrics availbility and such. Mind if i keep you posted?
I have a deleviry deadline on wednessday for a service, which will keep me tied up until then.
After that, i will revive this convo :slight_smile:
Again, your help is always much appreaciated!

Best regards
Oelsner

no worries :slight_smile:

just keep me posted, I’m quite interested how everything is going