Initial commit for Postgres Operator UI (#613)

* Initial commit for our basic Postgres Operator UI: * Create and modify Postgres manifests * Watch Operator Logs in the UI * Observe cluster creation progress * S3 Backup browser for clone and restore Many thanks to Manuel Gomez and Jan Mussler for the initial UI work a long time ago!
2026-10-06 04:52:11 +02:00 · 2019-07-12 17:02:55 +02:00
parent 4fc5822b24
commit 4eabc8f5db
60 changed files with 12752 additions and 13 deletions
@@ -0,0 +1,117 @@
+#!/usr/bin/env bash
+set -e
+
+# NOTE: You still need to start the frontend bits of this application separately
+# as starting it here as a child process would leave leftover processes on
+# script termination; it appears there is some complication that does not allow
+# the shell to clean up nodejs grandchild processes correctly upon script exit.
+
+
+# Static bits:
+export APP_URL="${API_URL-http://localhost:8081}"
+export OPERATOR_API_URL="${OPERATOR_API_URL-http://localhost:8080}"
+export TARGET_NAMESPACE="${TARGET_NAMESPACE-*}"
+
+default_operator_ui_config='{
+  "docs_link":"https://postgres-operator.readthedocs.io/en/latest/",
+  "dns_format_string": "{1}-{0}.{2}",
+  "databases_visible": true,
+  "nat_gateways_visible": false,
+  "resources_visible": true,
+  "users_visible": true,
+  "postgresql_versions": [
+    "11",
+    "10",
+    "9.6"
+  ],
+  "static_network_whitelist": {
+    "localhost": ["172.0.0.1/32"]
+  }
+}'
+export OPERATOR_UI_CONFIG="${OPERATOR_UI_CONFIG-${default_operator_ui_config}}"
+
+# S3 backup bucket:
+export SPILO_S3_BACKUP_BUCKET="postgres-backup"
+
+# defines teams
+teams='["acid"]'
+export TEAMS="${TEAMS-${teams}}"
+
+# Kubernetes API URL (e.g. minikube):
+kubernetes_api_url="https://192.168.99.100:8443"
+
+# Enable job control:
+set -m
+
+# Clean up child processes on exit:
+trap 'kill $(jobs -p)' EXIT
+
+
+# PostgreSQL Operator UI application name as deployed:
+operator_ui_application='postgres-operator-ui'
+
+
+# Hijack the PostgreSQL Operator UI pod as a proxy for its AWS instance profile
+# on the pod's localhost:1234 which allows the WAL-E code to list backups there:
+kubectl exec \
+  "$(
+    kubectl get pods \
+      --server="${kubernetes_api_url}" \
+      --selector="application=${operator_ui_application}" \
+      --output='name' \
+      | head --lines=1 \
+      | sed 's@^[^/]*/@@'
+  )" \
+  -- \
+    sh -c '
+      apk add --no-cache socat;
+      pkill socat;
+      socat -v TCP-LISTEN:1234,reuseaddr,fork,su=nobody TCP:169.254.169.254:80
+    ' \
+  &
+
+
+# Forward localhost:1234 to localhost:1234 on the PostgreSQL Operator UI pod to
+# get to the AWS instance metadata endpoint:
+echo "Port forwarding to the PostgreSQL Operator UI's instance metadata service"
+kubectl port-forward \
+  --server="${kubernetes_api_url}" \
+  "$(
+    kubectl get pods \
+      --server="${kubernetes_api_url}" \
+      --selector="application=${operator_ui_application}" \
+      --output='name' \
+      | head --lines=1 \
+      | sed 's@^[^/]*/@@'
+  )" \
+  1234 \
+  &
+
+
+# Forward localhost:8080 to localhost:8080 on the PostgreSQL Operator pod, which
+# allows access to the Operator REST API
+# when using helm chart use --selector='app.kubernetes.io/name=postgres-operator'
+echo 'Port forwarding to the PostgreSQL Operator REST API'
+kubectl port-forward \
+  --server="${kubernetes_api_url}" \
+  "$(
+    kubectl get pods \
+      --server="${kubernetes_api_url}" \
+      --selector='name=postgres-operator' \
+      --output='name' \
+      | head --lines=1 \
+      | sed 's@^[^/]*/@@'
+  )" \
+  8080 \
+  &
+
+
+# Start a local proxy on localhost:8001 of the target Kubernetes cluster's API:
+kubectl proxy &
+
+
+# Start application:
+python3 \
+  -m operator_ui \
+  --clusters='localhost:8001' \
+  $@