GCP docs: Suggest setting up gke cluster with 3 nodes

[vinayvinay]

What this PR does / why we need it:

I'm updating docs to get a smooth initial setup. This first example won't run without at least 3 nodes:

https://github.com/SeldonIO/seldon-core-launcher/blob/master/seldon-core/getting_started/deployment/deploy.ipynb

[seldondev]

Hi @vinayvinay. Thanks for your PR.

I'm waiting for a SeldonIO or todo member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the jenkins-x/lighthouse repository.

[axsaucedo]

NIce one - thanks for the PR
/cc @edshee

[edshee]

Good spot, thanks for fixing

[seldondev]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: edshee
To complete the pull request process, please assign rafalskolasinski
You can assign the PR to them by writing /assign @rafalskolasinski in a comment when ready.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[edshee]

/assign @RafalSkolasinski

[RafalSkolasinski]

Not sure if 3 nodes is a strict recommendation, especially here. Could just be a general note that with this flag one can adjust number of nodes.

[edshee]

Good point

[vinayvinay]

hi @RafalSkolasinski, well, the Create GKE Cluster link suggests keeping num-nodes to 1, so if someone like me blindly follows those docs, they might not be able to complete https://docs.seldon.io/projects/seldon-core/en/latest/install/gcp.html#install-seldon-core (seldon pods remain in Pending state, from my experience). giving this heads-up helps those who are not familiar enough with kubernetes to troubleshoot a pending deployment and might conclude, "oh- this doesn't work". i'll let you figure out the best way to convey this if you think it's useful.

[edshee]

Out of interest, did you install Istio or Ambassador? I think Istio is the thing that requires 3 nodes (or a big node) as it's pretty resource hungry. I'm pretty sure the Seldon controller would run fine on a single node.

[vinayvinay]

Hi Ed, I installed Istio, which explains.

[edshee]

Ok great. I merged this earlier so that others wouldn't have the same issue you had. I'll reword it to highlight Istio's node requirement on a subsequent commit.

[vinayvinay]

Thanks for updating it to be accurate. I wonder if seldon maintainers have a preference: Istio or Ambassador?

[edshee]

It's a difficult area because we don't really want to be saying one stack is better than the other. Our general preference right now is for Istio as it allows you to do more complex rollout strategies with the service mesh. In future releases we're looking to abstract away the ingress/service mesh a little more so that people can use whatever tool the like (or no tool).

[RafalSkolasinski]

Sorry for not following the discussion here more carefully. I'd suggest to have in our docs just a general note saying that with the --num-nodes flag one can customise number of GCP nodes, give three nodes as an example, and highlight that actual resource requirement may vary depending on other infrastructure components installed in the cluster / number of models.

A general advice here could be that "if you see pods in Pending state and not scheduled due to resource limitations, add more nodes to your cluster"