From 31e6e49015d84782e1c5687175dc11e36bfae4f8 Mon Sep 17 00:00:00 2001
From: Aidan Hilt <11202897+AidanHilt@users.noreply.github.com>
Date: Wed, 20 Mar 2024 15:41:29 -0400
Subject: [PATCH] Creating a new type of DB dump that grabs stuff for
 va-testing (#2508)

* Creating a new type of DB dump that grabs stuff for va-testing

* Missed changes to dbdump script

* Changing job name
---
 gen3/bin/dbbackup.sh                          | 13 ++-
 .../jobs/psql-db-dump-va-testing-job.yaml     | 80 +++++++++++++++++++
 2 files changed, 92 insertions(+), 1 deletion(-)
 create mode 100644 kube/services/jobs/psql-db-dump-va-testing-job.yaml

diff --git a/gen3/bin/dbbackup.sh b/gen3/bin/dbbackup.sh
index 29f267221..eb9611a90 100644
--- a/gen3/bin/dbbackup.sh
+++ b/gen3/bin/dbbackup.sh
@@ -173,6 +173,10 @@ db_restore() {
     gen3 job run psql-db-prep-restore
 }
 
+va_testing_db_dump() {
+  gen3 job run psql-db-dump-va-testing
+}
+
 
 # main function to determine whether dump or restore
 main() {
@@ -191,8 +195,15 @@ main() {
             create_s3_bucket
             db_restore
             ;;
+        va-dump)
+            gen3_log_info "Running a va-testing DB dump..."
+            create_policy
+            create_service_account_and_role
+            create_s3_bucket
+            va_testing_db_dump
+            ;;
         *)
-            echo "Invalid command. Usage: gen3 dbbackup [dump|restore]"
+            echo "Invalid command. Usage: gen3 dbbackup [dump|restore|va-dump]"
             return 1
             ;;
     esac
diff --git a/kube/services/jobs/psql-db-dump-va-testing-job.yaml b/kube/services/jobs/psql-db-dump-va-testing-job.yaml
new file mode 100644
index 000000000..8a8037e16
--- /dev/null
+++ b/kube/services/jobs/psql-db-dump-va-testing-job.yaml
@@ -0,0 +1,80 @@
+---
+# NOTE: This job was created specifically to dump all the databases in va-testing, in preparation for a move to second cluster
+# If you aren't doing that, this probably is not the job you're looking for
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: psql-db-dump-va-testing
+spec:
+  template:
+    metadata:
+      labels:
+        app: gen3job
+    spec:
+      affinity:
+        nodeAffinity:
+          preferredDuringSchedulingIgnoredDuringExecution:
+            - weight: 100
+              preference:
+                matchExpressions:
+                  - key: karpenter.sh/capacity-type
+                    operator: In
+                    values:
+                      - on-demand
+            - weight: 99
+              preference:
+                matchExpressions:
+                  - key: eks.amazonaws.com/capacityType
+                    operator: In
+                    values:
+                      - ONDEMAND
+      serviceAccountName: dbbackup-sa
+      containers:
+        - name: pgdump
+          image: quay.io/cdis/awshelper:master
+          imagePullPolicy: Always
+          env:
+            - name: gen3Env
+              valueFrom:
+                configMapKeyRef:
+                  name: global
+                  key: environment
+            - name: JENKINS_HOME
+              value: "devterm"
+            - name: GEN3_HOME
+              value: /home/ubuntu/cloud-automation
+          command: ["/bin/bash"]
+          args:
+            - "-c"
+            - |
+              source "${GEN3_HOME}/gen3/lib/utils.sh"
+              gen3_load "gen3/gen3setup"
+              account_id=$(aws sts get-caller-identity --query "Account" --output text)
+              default_bucket_name="gen3-db-backups-${account_id}"
+              default_databases=("fence" "indexd" "sheepdog" "peregrine" "arborist" "argo" "atlas" "metadata" "ohdsi" "omop-data" "wts")
+              s3_dir="va-testing-$(date +"%Y-%m-%d-%H-%M-%S")"
+              databases=("${default_databases[@]}")
+              bucket_name=$default_bucket_name
+
+              for database in "${databases[@]}"; do
+                gen3_log_info "Starting database backup for ${database}"
+                gen3 db backup "${database}" > "${database}.sql"
+
+                if [ $? -eq 0 ] && [ -f "${database}.sql" ]; then
+                  gen3_log_info "Uploading backup file ${database}.sql to s3://${bucket_name}/${s3_dir}/${database}.sql"
+                  aws s3 cp "${database}.sql" "s3://${bucket_name}/${s3_dir}/${database}.sql"
+
+                  if [ $? -eq 0 ]; then
+                    gen3_log_info "Successfully uploaded ${database}.sql to S3"
+                  else
+                    gen3_log_err "Failed to upload ${database}.sql to S3"
+                  fi
+                  gen3_log_info "Deleting temporary backup file ${database}.sql"
+                  rm -f "${database}.sql"
+                else
+                  gen3_log_err "Backup operation failed for ${database}"
+                  rm -f "${database}.sql"
+                fi
+              done
+              sleep 600
+      restartPolicy: Never