Hey guys, great work! We are running ClickHouse in our K8s cluster already using the ClickHouse operator and are looking to run SigNoz as well. What's the recommended approach? 1. Install a ClickHouse instance through the SigNoz helm chart without installing the ClickHouse operator (which we have installed already) - is this possible? That feels like it would be ideal? 2. Run both operators in parallel - not sure if that has any other consequences? 3. Set up a separate ClickHouse cluster for SigNoz (trying to replicate the settings from the clickhouse/values.yaml and following the external clickhouse guide) and setting the following values in the SigNoz values.yaml when installing the helm chart:

clickhouse:
    enabled: false # use an "external" clickhouse instance instead
  externalClickhouse:
    host: name_of_the_new_signoz_clickhouse_instance

@Prashant Shahi can you help here?

Hi @Robert Winter 👋

What's the recommended approach?

If possible, it would be best to keep SigNoz and ClickHouse isolated in a separate K8s cluster.

without installing the ClickHouse operator (which we have installed already) - is this possible? That feels like it would be ideal?

yes, it is possible. However, there are multiple changes that you would need to do to your existing cluster as stated in the external clickhouse guide. You will also need to version match the operator and the clickhouse server.

Run both operators in parallel - not sure if that has any other consequences?

Your existing clickhouse operator might be using

ClusterRole

and

ClusterRoleBinding

, you will need to use

Role

and

RoleBinding

instead.

Hey thanks a lot!

not sure how it could affect our other workloads that uses the other clickhouse cluster though.

As stated in previous message, you will need to use

Role

and

RoleBinding

instead for your clickhouse operator. That will keep the scope of your clickhouse operator to its own namespace. Otherwise, your existing clickhouse operator might make multiple periodic attempts to communicate to SigNoz-managed CHI pods.

Thanks @Prashant Shahi, I'll give it a go. Just a follow-up question on your answer below to clarify if option (1) is possible:

Install a ClickHouse instance through the SigNoz helm chart without installing the ClickHouse operator (which we have installed already) - is this possible?

yes, it is possible. However, there are multiple changes that you would need to do to your existing cluster as stated in the external clickhouse guide. You will also need to version match the operator and the clickhouse server.

Is there a setting in the signoz or clickhouse values.yaml files allows me to disable installing the clickhouse operator but still install ClickHouse using my existing clickhouse-operator? I see I can disable installing clickhouse but I didn't find a similar setting for skipping installing the operator?

Setting

clickhouse.enabled

to

false

should disable the clickhouse dependency chart i.e both operator and clickhouse installation.

Ok thanks! Then option (1) is out the window I guess as I can't use the signoz chart to install just the clickhouse cluster but not the operator

Based on this comment would it be enough you think to set my existing ClickHouse operator to only look in my existing namespaces?

``` # Concurrently running operators should watch on different namespaces.

namespaces: ["myexistingnamespacewithclickhouse"]```

seems like it. However, I cannot say for certain without testing out the complete set up with two different operators and respective CHIs.