Update elasticsearch index playbooks (#892)

* Update elasticsearch index swapping playbooks

* Add ansible syntax check

* Document index swapping

* Update .travis.yml

* Update .travis.yml

* Update .travis.yml

* Update .travis.yml

* Update .travis.yml

.travis.yml

@@ -3,11 +3,15 @@ go:
 - 1.x
 gobuild_args: -x -ldflags "-X main.buildversion=$(date -u +.%Y%m%d.%H%M%S)"
 before_install:
-- sudo apt-get install gcc-mingw-w64 gcc-mingw-w64-i686 gcc-mingw-w64-x86-64 binutils-mingw-w64-i686 binutils-mingw-w64-x86-64
+- sudo apt-add-repository ppa:ansible/ansible -y
+- sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 93C4A3FD7BB9C367
+- sudo apt-get update
+- sudo apt-get install gcc-mingw-w64 gcc-mingw-w64-i686 gcc-mingw-w64-x86-64 binutils-mingw-w64-i686 binutils-mingw-w64-x86-64 ansible
 - go get -u github.com/tools/godep
 before_script:
 - go vet
 - go test -v ./...
+- cd deploy/ansible/ && for playbook in *.yml; do ansible-playbook -i hosts $playbook --syntax-check; done && cd ../..
 install:
   # Add Godeps dependencies to GOPATH and PATH
   - export GOPATH="${TRAVIS_BUILD_DIR}/vendor:$GOPATH"

deploy/README.md

@@ -109,28 +109,40 @@ $ ansible-playbook -i hosts restore_database.yml
 ```
 
 
-### Create Elasticsearch Index Playbook
+### Elasticsearch Index Playbooks
 
-This playbook creates the elasticsearch index for our database from
-[ansible/roles/elasticsearch/files/elasticsearch_settings.yml](ansible/roles/elasticsearch/files/elasticsearch_settings.yml)
+We are using index aliases for a zero downtime and index hotswapping. You can
+find more information [here](https://www.elastic.co/guide/en/elasticsearch/guide/current/index-aliases.html).
 
+I'll assume you already have an index named `nyaapantsu_old` that is aliased to
+`nyaapantsu` and you want to create a new index to become the new `nyaapantsu`.
+
+First you'll need to modify the variables in the [group_vars/all](group_vars/all)
+file.
+
+The new index will be called `nyaapantsu_new`.
+
+```yaml
+nyaapantsu_elasticsearch_alias: nyaapantsu
+nyaapantsu_elasticsearch_index: nyaapantsu_new
+nyaapantsu_elasticsearch_old_index: nyaapantsu_old
 ```
+
+Now you need to run three playbooks.
+
+```bash
+# Creates the new index
 $ ansible-playbook -i hosts create_elasticsearch_index.yml
-```
-
-
-### Populate Elasticsearch Index Playbook
-
-This playbook uses a python script to populate the elasticsearch index from all
-the data inside the database.
-
-> WARNING: Make sure the python script is in sync with the mapping defined in
-> the elasticsearch index configuration.
-
-```
+# Populate the new index and disable the reindexing cron job. This avoid
+# losing new entries.
 $ ansible-playbook -i hosts populate_elasticsearch_index.yml
+# Remove the alias nyaapantsu from nyaapantsu_old and adds it to nyaapantsu_new
+$ ansible-playbook -i hosts swap_elasticsearch_index.yml
 ```
 
+Nyaa can now access the new index `nyaapantsu_new` by using the alias
+`nyaapantsu`.
+
 ## Playbook Testing
 
 You can easily test these playbooks by using vagrant. Once you have vagrant
@@ -138,7 +150,7 @@ installed:
 
 ```
 # Download centos/7 image
-$ vagrant  init centos/7
+$ vagrant init centos/7
 
 # Create and boot the vm
 $ vagrant up

deploy/ansible/create_elasticsearch_index.yml

@@ -16,8 +16,3 @@
         url: "http://localhost:9200/{{ nyaapantsu_elasticsearch_index }}"
         method: PUT
         body: "{{ config.stdout }}"
-
-    - name: Create alias
-      uri:
-        url: "http://localhost:9200/{{ nyaapantsu_elasticsearch_index }}/_alias/{{ nyaapantsu_elasticsearch_alias }}"
-        method: PUT

deploy/ansible/group_vars/all

@@ -9,6 +9,7 @@ nyaapantsu_elasticsearch_alias: nyaapantsu
 # nyaapantsu_elasticsearch_alias: sukebei
 nyaapantsu_elasticsearch_index: nyaapantsu_v1
 # nyaapantsu_elasticsearch_index: nyaapantsu_sukebei_v1
+nyaapantsu_elasticsearch_old_index: CHANGETHIS
 nyaapantsu_torrent_tablename: torrents
 # nyaapantsu_torrent_tablename: torrents_sukebei
 nyaapantsu_max_open_files: 200000

deploy/ansible/populate_elasticsearch_index.yml

@@ -6,6 +6,13 @@
     - elasticsearch
 
   tasks:
+    # Needed otherwise we'll lose information when people upload when we're
+    # populating to the new index.
+    # DONT FORGET TO RE-ENABLE THE CRON JOB AFTER
+    - name: Disable reindex cron job
+      command: mv "/etc/cron.d/reindex_{{ nyaapantsu_torrent_tablename }}" /tmp/
+      become: true
+
     - name: Index the database into elasticsearch
       command: python "{{ nyaapantsu_directory }}/index_nyaapantsu.py"
       environment:

deploy/ansible/swap_elasticsearch_index.yml

@@ -0,0 +1,28 @@
+# We're not putting elasticsearch role because we don't want to re-enable the
+# reindexing cron job before we've actually made the swap
+- name: Populate elasticsearch index from database
+  hosts: dbs
+  roles:
+    - common
+    - postgresql
+
+  tasks:
+    - name: Swap elasticsearch index
+      uri:
+        url: "http://localhost:9200/_aliases"
+        method: POST
+        body_format: json
+        body:
+          actions:
+            - remove:
+                index: "{{ nyaapantsu_elasticsearch_old_index }}"
+                alias: "{{ nyaapantsu_elasticsearch_alias }}"
+            - add:
+                index: "{{ nyaapantsu_elasticsearch_index }}"
+                alias: "{{ nyaapantsu_elasticsearch_alias }}"
+
+    - name: Re-enable reindex cron job
+      command: mv "/tmp/reindex_{{ nyaapantsu_torrent_tablename }}" /etc/cron.d/
+      become: true
+      ignore_errors: yes # Can ignore error here if the file had previously been
+                         # moved.