Skip to content

Accessing Production

The production instance of codeforphilly.org is deployed via the cfp-live-cluster via the release configured at cfp-live-cluster/code-for-philly

Deployment process

  1. Merge one or more PRs into develop under the codeforphilly.org repository
    • A release PR will automatically be generated with a suggested version number in the title, a template for release notes in the description, and a copyable draft changelog in the first comment
    • The first comment with the draft changelog will be automatically updated as additional commits/PRs are added to develop
    • The title or description will not be automatically changed after the PR is initially created, so they are safe to manually edit
  2. Copy bullet points from the generated changelog comment and use them to fill out the release notes template in the description, deleting any unused section. The suggested version number in the title will default to incrementing the patch number, and may be manually edited to switch to incrementing the minor number
  3. Merge the release PR when ready
    • A GitHub release+tag will automatically get created
    • The new release/tag will kick off automatically building and tagging a new Docker container image
  4. After the new image is built, edit code-for-philly/release-values.yaml in the cfp-live-cluster repository to update the image tag to the new version either as a direct commit to the main branch or a PR (see this example commit)
  5. Ask the Code for Philly ops team to merge your version bump PR if needed
    • A deployment PR will automatically be generated after that, previewing the pending changes to the cluster
  6. Ask the Code for Philly ops team to merge the pending deployment
    • The cluster will automatically be updated and the new image deployed

Accessing instance

The production database runs within the Kubernetes cluster, and is snapshotted every hour by restic to a Linode cloud storage bucket.

The database, as well as application logs and shell, can be accessed via a KUBECONFIG file with access to the needed resources. Ask Code for Philly ops to provide one if needed.

Using admin service account

  1. Activate the provided KUBECONFIG in your current terminal session:

    export KUBECONFIG=~/.kube/code-for-philly-admin.yaml

  2. Get the name of the currently running pods and store them in shell variables:

    POD_NAME=$(kubectl -n code-for-philly get pod --field-selector=status.phase=Running -l app.kubernetes.io/name=code-for-philly -o jsonpath='{.items[0].metadata.name}')

Open interactive backend shell

kubectl -n code-for-philly exec -it $POD_NAME -- bash

Open interactive database shell

kubectl -n code-for-philly exec -it $POD_NAME -- hab pkg exec codeforphilly/site-composite mysql

Run an emergence-console command

kubectl -n code-for-philly exec -it $POD_NAME -- hab pkg exec codeforphilly/site-composite emergence-console-run migrations:execute

Run an SQL query

echo 'SELECT * FROM people' | kubectl -n code-for-philly exec -i $POD_NAME -- hab pkg exec codeforphilly/site-composite mysql

Dump the entire database

kubectl -n code-for-philly exec -it $POD_NAME -- hab pkg exec codeforphilly/site-composite mysqldump 'codeforphilly' > /tmp/codeforphilly.sql

Forward MySQL port

kubectl -n code-for-philly port-forward pods/$POD_NAME 3306:3306

Database logins

Default database credentials can be found in helm-chart/values.yaml (typically admin / admin)