support sidecar containers and provide cloudsqlproxy example

[mnichols]

This PR extends the helm templates to support sidecar containers. An example for adding a Google cloud sql proxy sidecar is provided.

[underrun]

Thanks for the PR!

My initial thoughts are that it would be really nice if users could apply values files without editing. Maybe just an example in the docs would be useful?

[mnichols]

Hi @underrun! Helm isn't really my wheelhouse and I wasn't sure of a way to apply a sidecar to the existing templates without the approach taken here.
How does one extend the deployment without doing the -f <valuesfile> approach used so far in this repo?
Or were you just saying to not include the cloudsqlproxy values file and make it purely a form of documentation?

Personally I don't mind the approach in the values.postgres.yaml and others but could see how they should be maintained outside. Only question I'd have is how best to keep it maintained in sync with the helm templates they are extending.

[underrun]

I meant the documentation angle... But let me think about this a bit and chat with some folks next week.

Thanks again for working on this!

[mnichols]

Sure thing..lemme know what y'all need. I was just pleased to be able to simply support cloudsqlproxy for our prod stuff in GCP. Your work here made it a snap to extend...nice job!

[jpearll]

Hi,
I am new with Helm. I am trying to deploy temporal on GCP (GKE).

I have tried to use the PR for getting Cloud Proxy support, with the following command:
helm install -f ./values/values.cloudsqlproxy.yaml workflow . --timeout 15m

But I have systematically a status 'Does not have minimum availability' on GCP for the temporal-frontend, temporal-worker, temporal-history, temporal-matching and temporal-worker workloads. Even if I tried to add nodes (6 nodes, auto-scaling).

I am not sure about how to tackle this. Also, what is the recommended GKE configuration for the cluster/node pool/machine types?

Thanks !

[mnichols]

Hi,
I am new with Helm. I am trying to deploy temporal on GCP (GKE).

I have tried to use the PR for getting Cloud Proxy support, with the following command:
helm install -f ./values/values.cloudsqlproxy.yaml workflow . --timeout 15m

But I have systematically a status 'Does not have minimum availability' on GCP for the temporal-frontend, temporal-worker, temporal-history, temporal-matching and temporal-worker workloads. Even if I tried to add nodes (6 nodes, auto-scaling).

I am not sure about how to tackle this. Also, what is the recommended GKE configuration for the cluster/node pool/machine types?

Thanks !

@jpearll I cant comment on the recommended GKE config you need, but it sounds like if your pods arent starting up with this cloudsqlproxy config in place your attempt to connect to your db might be hanging. This is a sidecar and I've found they tend to fail badly (eg silently). You might want to check a direct db connection to your cloudsql instance first before trying with this proxy in place?

[underrun]

My apologies for the much-longer-than-a-week it took me to come back to this. We'd like to try and merge this in with a little cleanup. Thanks again for putting this together!

[jasonzoladz]

@underrun, what's the ETA for merging this? We need this as well. :)

[underrun]

@mnichols would you like to make the requested change? If not I can open a parallel PR to get this merged and close this one? Either way is totally fine. Thanks again for your work on this!

[CLAassistant]

All committers have signed the CLA.

[mnichols]

@underrun I made the changes I think you were after and squashed it after rebasing on latest master

[JDWardle]

@underrun Is there anything blocking this from being merged in? This would be very helpful to have not just for cloudsql but support for adding in any other sidecars as well.

[hazcod]

@mnichols @underrun any news here?

[mnichols]

🤷
@underrun ?