[New hubs] Dandi

[GeorgianaElena]

Closes #3824

Also, updates the eksctl jsonnet file to not have dask nodes (I assumed there will be daskhubs on setup) and the terraform to use the requested buckets name (staging and prod were assumed by the template).

[github-actions]

Merging this PR will trigger the following deployment actions.

Support and Staging deployments

Cloud Provider	Cluster Name	Upgrade Support?	Reason for Support Redeploy	Upgrade Staging?	Reason for Staging Redeploy
aws	dandi	Yes	Following helm chart values files were modified: support.values.yaml	Yes	Following helm chart values files were modified: staging.values.yaml, enc-staging.secret.values.yaml, common.values.yaml
aws	earthscope	No		Yes	Following helm chart values files were modified: common.values.yaml
aws	bican	No		Yes	Following helm chart values files were modified: common.values.yaml, staging.values.yaml, enc-staging.secret.values.yaml

Production deployments

Cloud Provider	Cluster Name	Hub Name	Reason for Redeploy
aws	dandi	prod	Following helm chart values files were modified: prod.values.yaml, enc-prod.secret.values.yaml, common.values.yaml
aws	earthscope	prod	Following helm chart values files were modified: common.values.yaml
aws	bican	prod	Following helm chart values files were modified: common.values.yaml, prod.values.yaml, enc-prod.secret.values.yaml

[GeorgianaElena]

I don't know what's wrong, but I can't get the autoscaler to trigger when starting a new server. It's just stuck in:

I0329 11:55:18.773905       1 scheduling_queue.go:1070] "About to try and schedule pod" pod="staging/jupyter-georgianaelena"
I0329 11:55:18.773926       1 schedule_one.go:81] "Attempting to schedule pod" pod="staging/jupyter-georgianaelena"
I0329 11:55:18.774113       1 preemption.go:216] "Preemption will not help schedule pod on any node" pod="staging/jupyter-georgianaelena"
I0329 11:55:18.774149       1 schedule_one.go:860] "Unable to schedule pod; no fit; waiting" pod="staging/jupyter-georgianaelena" err="0/1 nodes are available: 1 node(s) didn't match Pod's node affinity/selector. preemption: 0/1 nodes are available: 1 Preemption is not helpful for scheduling.."
I0329 11:55:18.774231       1 schedule_one.go:936] "Updating pod condition" pod="staging/jupyter-georgianaelena" conditionType=PodScheduled conditionStatus=False conditionReason="Unschedulable"

I even recreated the cluster, thinking I missed something during the first phases :|

[yuvipanda]

@GeorgianaElena i can take a look at this today.

[yuvipanda]

On AWS, we unfortunately have to manually install and manage the cluster autoscaler. I looked to see if it was crashing, with k -n support get pod, and noticed it wasn't there. We need to explicitly enable it in https://github.com/2i2c-org/infrastructure/blob/main/config/clusters/dandi/support.values.yaml, like in

infrastructure/config/clusters/bican/support.values.yaml

Line 30 in 40fca24

cluster-autoscaler:

.

I have opened #3881 to automate this. Otherwise, I think if you enable the cluster autoscaler in the support chart and redeploy it, that should work!

And sorry for not catching it when I was reviewing #3866

[GeorgianaElena]

On AWS, we unfortunately have to manually install and manage the cluster autoscaler.

Thank you @yuvipanda. I think this bit was recently removed from docs probably under the assumption that was already automated. Will work on fixing #3881. Thank you 🙏🏼

[GeorgianaElena]

I believe this PR is now ready to be reviewed/deployed 🚀

[yuvipanda]

Two changes but otherwise looks good to me!

[yuvipanda]

Good to go!

[github-actions]

🎉🎉🎉🎉

Monitor the deployment of the hubs here 👉 https://github.com/2i2c-org/infrastructure/actions/runs/8527749373

[yuvipanda]

Looks like I missed spotting the equivalent of #3895 here too!