Skip to content

Feature/eks container path fixes - #24

Open
KamalCharles wants to merge 4 commits into
developfrom
feature/eks-container-path-fixes
Open

Feature/eks container path fixes#24
KamalCharles wants to merge 4 commits into
developfrom
feature/eks-container-path-fixes

Conversation

@KamalCharles

Copy link
Copy Markdown
Collaborator

Title:
Sync EKS config with live cluster state and fix container paths

Body:

Summary

  • Fix hardcoded data paths in app.R, drought, and ibtracs sim scripts to use absolute /srv/shiny-server paths expected in the EKS container; revert the ShinyProxy image tag from v1-logo-fix to latest now that the fix is baked into the default image
  • Add sp-restart-cronjob.yaml and sp-hpa.yaml to capture the scheduled restart CronJob (plus its ServiceAccount/Role/RoleBinding) and the HorizontalPodAutoscaler
  • stale duplicate Deployment definition left over from an old branch merge that no longer matches the live shinyproxy-new deployment
  • Rename sp-deployment-new.yaml, sp-service-new.yaml, and sp-service-new-risk.yaml to drop the -new suffix for clarity (file names only — underlying Kubernetes resource names are unchanged)

Test plan

  • Confirm EKS-Config manifests match kubectl get output for shinyproxy-new deployment, hpa, and cronjob in the shiny namespace
  • Verify app still loads correctly after image rebuild with updated app.R/sim script paths

Update drought and TC ibtracs simulation scripts plus app.R mapping/
python source paths to use absolute /srv/shiny-server paths expected
in the EKS container, and revert the ShinyProxy deployment image tag
from v1-logo-fix back to latest now that the fix is baked into the
default image.
Capture the shinyproxy-restart CronJob (with its ServiceAccount, Role,
and RoleBinding) and the shinyproxy-hpa HorizontalPodAutoscaler that
were applied directly to the cluster, so autoscaling and the scheduled
restart are tracked in the repo. Also remove
sp-deployment-serviceaccount.yaml, a stale duplicate Deployment
definition left over from an old branch merge that no longer matches
the live shinyproxy-new deployment.
Rename sp-deployment-new.yaml, sp-service-new.yaml, and
sp-service-new-risk.yaml to sp-deployment.yaml, sp-service.yaml, and
sp-service-risk.yaml for clarity. File names only — the underlying
Kubernetes resource names (e.g. shinyproxy-new) are unchanged, so
there's no impact on the live cluster.
The riskexplorer-service targetPort still pointed at 3838 (pre-ShinyProxy
migration) instead of 8080, causing Cloudflare 520s. Session pods also had
no CPU/memory requests/limits, letting nodes get overpacked and OOM.
Also patches the ShinyProxy tab title and points sp-deployment.yaml at the
oasislmf-resource-limits image actually running in the cluster.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant