From 99397ce7198818e478d859402cde031e60925576 Mon Sep 17 00:00:00 2001
From: hyunlee <Hyun.H.Lee@jpl.nasa.gov>
Date: Wed, 3 Apr 2024 11:09:02 -0700
Subject: [PATCH] ES snapshot policy is now daily instead of hourly

---
 cluster_provisioning/modules/common/mozart.tf | 31 +++++++++++++++++--
 1 file changed, 28 insertions(+), 3 deletions(-)

diff --git a/cluster_provisioning/modules/common/mozart.tf b/cluster_provisioning/modules/common/mozart.tf
index 2cba0fd61..6027a7c2e 100644
--- a/cluster_provisioning/modules/common/mozart.tf
+++ b/cluster_provisioning/modules/common/mozart.tf
@@ -470,6 +470,20 @@ resource "aws_instance" "mozart" {
     ]
   }
 
+  # Copy down latest opera-sds-int and opera-sds-ops repos for convenience
+  provisioner "remote-exec" {
+    inline = [<<-EOT
+      while [ ! -f /var/lib/cloud/instance/boot-finished ]; do echo 'Waiting for cloud-init...'; sleep 5; done
+      set -ex
+      cd ~/mozart/ops
+      wget https://github.com/nasa/opera-sds-int/archive/refs/heads/main.zip -O opera-sds-int.zip
+      wget https://github.com/nasa/opera-sds-ops/archive/refs/heads/main.zip -O opera-sds-ops.zip
+      unzip opera-sds-int.zip
+      unzip opera-sds-ops.zip
+    EOT
+    ]
+  }
+
   provisioner "remote-exec" {
     inline = [<<-EOT
       while [ ! -f /var/lib/cloud/instance/boot-finished ]; do echo 'Waiting for cloud-init...'; sleep 5; done
@@ -538,9 +552,13 @@ resource "aws_instance" "mozart" {
       fi
       cp -pr ~/mozart/ops/opera-pcm ~/verdi/ops/opera-pcm
       echo buckets are ---- ${local.code_bucket} ${local.dataset_bucket} ${local.isl_bucket}
+
+      sed -i "s/RELEASE_VERSION: '{{ RELEASE_VERSION }}'/RELEASE_VERSION: '${var.pcm_branch}'/g" ~/mozart/ops/opera-pcm/conf/settings.yaml
+
       if [ "${var.pge_sim_mode}" = false ]; then
         sed -i 's/PGE_SIMULATION_MODE: !!bool true/PGE_SIMULATION_MODE: !!bool false/g' ~/mozart/ops/opera-pcm/conf/settings.yaml
       fi
+      sed -i "s/DATASET_BUCKET: '{{ DATASET_BUCKET }}'/DATASET_BUCKET: '${local.dataset_bucket}'/g" ~/mozart/ops/opera-pcm/conf/settings.yaml
 
 	  sleep 180
       if [ "${var.use_artifactory}" = true ]; then
@@ -570,7 +588,13 @@ resource "aws_instance" "mozart" {
       cd ~/mozart/ops/opera-pcm
       echo # download dependencies for CLI execution of daac_data_subscriber.py
       pip install '.[subscriber]'
+      pip install '.[audit]'
+      pip install '.[cmr_audit]'
       pip install --progress-bar off -e .
+
+      # For daac_data_subscriber utility tool
+      mkdir ~/Downloads/
+      aws s3 cp  s3://opera-ancillaries/mgrs_tiles/dswx_s1/MGRS_tile_collection_v0.3.sqlite ~/Downloads/
     EOT
     ]
   }
@@ -613,6 +637,7 @@ resource "aws_instance" "mozart" {
   }
 
   // Snapshot repositories and lifecycles for GRQ mozart and metrics ES, also set shard max
+  // Snapshot schedule is in UTC, 5 AM UTC is 9/10 PM PST, depending on daylight savingss
   provisioner "remote-exec" {
     inline = [<<-EOT
      while [ ! -f /var/lib/cloud/instance/boot-finished ]; do echo 'Waiting for cloud-init...'; sleep 5; done
@@ -620,17 +645,17 @@ resource "aws_instance" "mozart" {
       source ~/.bash_profile
       echo // grq
       ~/mozart/bin/snapshot_es_data.py --es-url ${local.grq_es_url} create-repository --repository snapshot-repository --bucket ${var.es_snapshot_bucket} --bucket-path ${var.project}-${var.venue}-${var.counter}/grq --role-arn ${var.es_bucket_role_arn}
-      ~/mozart/bin/snapshot_es_data.py --es-url ${local.grq_es_url} create-lifecycle --repository snapshot-repository --policy-id hourly-snapshot --snapshot grq-backup --index-pattern grq_*,*_catalog
+      ~/mozart/bin/snapshot_es_data.py --es-url ${local.grq_es_url} create-lifecycle --repository snapshot-repository --policy-id daily-snapshot --snapshot grq-backup --index-pattern grq_*,*_catalog --schedule="0 0 5 * * ?"
       curl -XPUT ${local.grq_es_url}/_cluster/settings -H 'Content-type: application/json' --data-binary $'{"transient":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}, "persistent":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}}'
 
       echo // mozart
       ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.mozart.private_ip}:9200 create-repository --repository snapshot-repository --bucket ${var.es_snapshot_bucket} --bucket-path ${var.project}-${var.venue}-${var.counter}/mozart --role-arn ${var.es_bucket_role_arn}
-      ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.mozart.private_ip}:9200 create-lifecycle --repository snapshot-repository --policy-id hourly-snapshot --snapshot mozart-backup --index-pattern *_status-*,user_rules-*,job_specs,hysds_ios-*,containers
+      ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.mozart.private_ip}:9200 create-lifecycle --repository snapshot-repository --policy-id daily-snapshot --snapshot mozart-backup --index-pattern *_status-*,user_rules-*,job_specs,hysds_ios-*,containers --schedule="0 0 5 * * ?"
       curl -XPUT http://${aws_instance.mozart.private_ip}:9200/_cluster/settings -H 'Content-type: application/json' --data-binary $'{"transient":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}, "persistent":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}}'
 
       echo // metrics
       ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.metrics.private_ip}:9200 create-repository --repository snapshot-repository --bucket ${var.es_snapshot_bucket} --bucket-path ${var.project}-${var.venue}-${var.counter}/metrics --role-arn ${var.es_bucket_role_arn}
-      ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.metrics.private_ip}:9200 create-lifecycle --repository snapshot-repository --policy-id hourly-snapshot --snapshot metrics-backup --index-pattern logstash-*,sdswatch-*,mozart-logs-*,factotum-logs-*,grq-logs-*
+      ~/mozart/bin/snapshot_es_data.py --es-url http://${aws_instance.metrics.private_ip}:9200 create-lifecycle --repository snapshot-repository --policy-id daily-snapshot --snapshot metrics-backup --index-pattern logstash-*,sdswatch-*,mozart-logs-*,factotum-logs-*,grq-logs-* --schedule="0 0 5 * * ?"
       curl -XPUT http://${aws_instance.metrics.private_ip}:9200/_cluster/settings -H 'Content-type: application/json' --data-binary $'{"transient":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}, "persistent":{"cluster.max_shards_per_node": 6000, "search.max_open_scroll_context": 6000}}'
 
     EOT