From b05be36067ee128c35052590c9694e5e854dc6c3 Mon Sep 17 00:00:00 2001
From: rlskoeser <rebecca.s.koeser@princeton.edu>
Date: Mon, 4 Nov 2024 14:44:27 -0500
Subject: [PATCH] Update PPA configurations and playbook

- enable crontab
- configure cron job to reindex gale pages every first Saturday
- revise gale local ocr path
- remove unneeded transitional extra dependency
---
 inventory/group_vars/prosody/vars.yml            | 11 +++++++----
 inventory/group_vars/prosody_production/vars.yml | 15 +++++++++++++++
 inventory/group_vars/prosody_staging/vars.yml    |  3 +++
 playbooks/prosody.yml                            | 13 ++++++++-----
 roles/configure_crontab/tasks/main.yml           |  2 ++
 5 files changed, 35 insertions(+), 9 deletions(-)

diff --git a/inventory/group_vars/prosody/vars.yml b/inventory/group_vars/prosody/vars.yml
index e186d75e..a6b8edc1 100644
--- a/inventory/group_vars/prosody/vars.yml
+++ b/inventory/group_vars/prosody/vars.yml
@@ -43,7 +43,7 @@ marc_data_path: "{{ data_path }}/marc/"
 # path to EEBO TCP data
 eebo_data_path: "{{ data_path }}/eebo_tcp/"
 # path to Gale OCR data on tigerdata
-gale_ocr_data_path: "{{ tigerdata_mount_dir }}/prosody/ppa-ocr/Gale"
+gale_ocr_data_path: "{{ tigerdata_mount_dir }}/prosody/ppa-ocr/Gale-by-vol"
 
 technical_contact: cdhdevteam@princeton.edu
 # flag that csp is enabled
@@ -80,8 +80,11 @@ tigerdata_enabled: true
 # media nfs path
 media_root: /mnt/nfs/cdh/prosody/media/
 
-# app-specific system dependencies
-app_dependencies:
-  - libmysqlclient-dev # for mysql installation during the migration
+# app-specific system dependencies - no longer needed
+#app_dependencies:
 # common datadog configuration set in all vars
 datadog_app_name: cdh_prosody
+
+# configure scripts to run as cron jobs
+crontab: []
+# currently production only
diff --git a/inventory/group_vars/prosody_production/vars.yml b/inventory/group_vars/prosody_production/vars.yml
index 7b974b2f..02ce6e50 100644
--- a/inventory/group_vars/prosody_production/vars.yml
+++ b/inventory/group_vars/prosody_production/vars.yml
@@ -13,3 +13,18 @@ deploy_contexts:
     - 'javascript unit tests'
     - 'Python unit tests'
     # - 'codecov/project'
+
+# run cron jobs on the first vm
+cronjob_host: "cdh-prosody1.princeton.edu"
+
+# configure scripts to run as cron jobs
+crontab:
+  # schedule monthly reindex of Gale pages to update encrypted image urls
+  - name: "{{ django_app }} reindex Gale pages"
+    # run at 2am on the first Saturday of every month
+    minute: 0
+    hour: 2
+    weekday: "SAT"
+    day: "1-7"
+    job: "bin/cron-wrapper {{ deploy }}/env/bin/python {{ deploy }}/manage.py index_pages -p 4 --gale --no-progress >> {{ logging_dir }}/cron_reindex_gale_pages.log  2>&1"
+    state: present
diff --git a/inventory/group_vars/prosody_staging/vars.yml b/inventory/group_vars/prosody_staging/vars.yml
index 69113300..c22cea38 100644
--- a/inventory/group_vars/prosody_staging/vars.yml
+++ b/inventory/group_vars/prosody_staging/vars.yml
@@ -19,3 +19,6 @@ zk_host: "lib-zk-staging4:2181,lib-zk-staging5:2181,lib-zk-staging6:2181/solr9"
 solr_url: "http://lib-solr9-staging.princeton.edu:8983/solr/"
 solr_server: "{{ groups['solr9_staging'][0] }}"
 solr_version: 9
+
+# run cron jobs on the first vm
+cronjob_host: "cdh-test-prosody1.princeton.edu"
diff --git a/playbooks/prosody.yml b/playbooks/prosody.yml
index 91597e52..7f388feb 100644
--- a/playbooks/prosody.yml
+++ b/playbooks/prosody.yml
@@ -31,10 +31,13 @@
     - solr_collection
     - prosody_setup
     - finalize_deploy
+    - configure_crontab  # used in production but not always in staging
     - close_deployment
   tasks:
-  - ansible.builtin.include_role:   # datadog for logging
-      name: pulibrary.princeton_ansible.datadog
-    when: runtime_env | default('staging') == "production"
-    tags:
-      - datadog
+    - ansible.builtin.include_role:   # timezone needed for cron jobs
+        name: pulibrary.princeton_ansible.timezone
+    - ansible.builtin.include_role:   # datadog for logging
+        name: pulibrary.princeton_ansible.datadog
+      when: runtime_env | default('staging') == "production"
+      tags:
+        - datadog
diff --git a/roles/configure_crontab/tasks/main.yml b/roles/configure_crontab/tasks/main.yml
index ca8f92c3..e4af0ed1 100644
--- a/roles/configure_crontab/tasks/main.yml
+++ b/roles/configure_crontab/tasks/main.yml
@@ -1,5 +1,7 @@
 - name: Configure cron jobs
   when: inventory_hostname == cronjob_host
+  tags:
+    - crontab
   block:
   - name: "Configure crontab — special times"
     cron: