From ecd8ab838339764ea606eac1e8edcd2ab4cc4b06 Mon Sep 17 00:00:00 2001
From: Nils Philippsen <nils@redhat.com>
Date: Mon, 30 Aug 2021 15:41:04 +0200
Subject: [PATCH] Merge syncing and combining logs into one cronjob

This should prevent race conditions of the form that logs are attempted
to be combined while syncing those of individual hosts hasn't finished.

Signed-off-by: Nils Philippsen <nils@redhat.com>
---
 .../web-data-analysis/files/combineHttp.cron  |  1 -
 .../files/sync-http-logs-and-merge.sh         |  7 +++++
 roles/web-data-analysis/tasks/main.yml        | 28 ++++++++++++++++---
 3 files changed, 31 insertions(+), 5 deletions(-)
 delete mode 100644 roles/web-data-analysis/files/combineHttp.cron
 create mode 100755 roles/web-data-analysis/files/sync-http-logs-and-merge.sh

diff --git a/roles/web-data-analysis/files/combineHttp.cron b/roles/web-data-analysis/files/combineHttp.cron
deleted file mode 100644
index 016a796727..0000000000
--- a/roles/web-data-analysis/files/combineHttp.cron
+++ /dev/null
@@ -1 +0,0 @@
-0 06  * * * root /usr/local/bin/combineHttpLogs.sh
diff --git a/roles/web-data-analysis/files/sync-http-logs-and-merge.sh b/roles/web-data-analysis/files/sync-http-logs-and-merge.sh
new file mode 100755
index 0000000000..888ae82cf2
--- /dev/null
+++ b/roles/web-data-analysis/files/sync-http-logs-and-merge.sh
@@ -0,0 +1,7 @@
+#!/bin/bash
+
+# Don't attempt to combine if syncing the individual logs failed.
+set -e
+
+/usr/local/bin/sync-http-logs.py
+/usr/local/bin/combineHttpLogs.sh
diff --git a/roles/web-data-analysis/tasks/main.yml b/roles/web-data-analysis/tasks/main.yml
index bd1379b321..b44814c313 100644
--- a/roles/web-data-analysis/tasks/main.yml
+++ b/roles/web-data-analysis/tasks/main.yml
@@ -178,13 +178,22 @@
 - name: install a script to sync httpd logs via cron only on log01
   copy:
     src: sync-http-logs.py
-    dest: /etc/cron.daily/sync-http-logs.py
+    dest: /usr/local/bin/sync-http-logs.py
     mode: 0755
   when: inventory_hostname.startswith('log01')
   tags:
     - web-data
     - cron
 
+- name: remove sync-http-logs.py from cron.daily directory
+  file:
+    path: /etc/cron.daily/sync-http-logs.py
+    state: absent
+  when: inventory_hostname.startswith('log01')
+  tags:
+    - web-data
+    - cron
+
 - name: install awstats package
   package: state=present name=awstats
   tags:
@@ -202,8 +211,19 @@
     - config
     - web-data
 
-- name: daily cron job to merge old logs
-  copy: src=combineHttp.cron dest=/etc/cron.d/ mode=0644
+- name: remove separate daily cron job to merge old logs
+  file:
+    path: /etc/cron.d/combineHttp.cron
+    state: absent
   tags:
-    - cron
     - web-data
+    - cron
+
+- name: install daily cron job to sync and merge log files
+  copy:
+    src: sync-http-logs-and-merge.sh
+    dest: /etc/cron.d
+    mode: 0755
+  tags:
+    - web-data
+    - cron