Deployment with Docker Compose

Dockerfile

@@ -0,0 +1,28 @@
+FROM debian:bullseye
+
+RUN apt update
+RUN apt install -y apache2 git subversion curl cron \
+                   libconfig-ini-perl libdbd-mysql-perl libdbi-perl libxml-parser-perl \
+                   php7.4 php7.4-mysql php7.4-curl php7.4-intl libapache2-mod-php7.4 php7.4-mbstring php7.4-xml
+
+WORKDIR /var/www/html
+RUN rm -rf /var/www/html/* && git clone https://github.com/OpenCorpora/opencorpora.git .
+RUN curl -sS https://getcomposer.org/installer | php
+RUN php composer.phar self-update --1
+RUN php composer.phar install
+
+WORKDIR /var/www
+RUN mkdir -p smarty_dir/cache && mkdir -p smarty_dir/configs && mkdir -p smarty_dir/templates_c
+RUN chown -R www-data:www-data smarty_dir
+
+WORKDIR /var/www/html
+RUN sed -E 's/^(host\s+=\s+)127\.0\.0\.1/\1opencorpora_db/' config.default.ini | \
+    sed -E 's/^(template_dir\s+=\s+)\/var\/www\/templates\//\1\/var\/www\/html\/templates\//'  > config.ini
+RUN sed -E 's/^(\s+\"host\"\s*:\s*\")127\.0\.0\.1/\1opencorpora_db/' config.default.json | \
+    sed -E 's/^(\s+\"template_dir\"\s*:\s+\")\/var\/www\/templates\//\1\/var\/www\/html\/templates\//' > config.json
+RUN mkdir logs && chown -R www-data:www-data logs
+
+RUN ln -s /var/www/html /corpus
+
+CMD ["apachectl", "-D", "FOREGROUND"]
+

README.md

@@ -2,3 +2,50 @@ opencorpora
 ===========
 
 A web-based engine for creating and annotating textual corpora
+
+
+How-to run with Docker Compose
+==============================
+
+- Modify MySql root password in `docker-compose.yml`
+
+- Execure following commands in the first terminal:
+```
+docker-compose up --build
+```
+
+- Execute following commands in the second terminal:
+
+```
+docker-compose exec opencorpora_db /bin/bash
+
+# Inside container:
+
+# Create database in MySql and grant access
+mysql -u root -p
+
+# Inside MySql shell:
+CREATE DATABASE corpora DEFAULT CHARSET utf8;
+CREATE USER corpora@localhost IDENTIFIED BY 'password';
+GRANT DELETE, INSERT, LOCK TABLES, SELECT, UPDATE, ALTER, CREATE ON corpora.* TO corpora;
+exit;
+
+exit
+```
+
+```
+docker-compose exec opencorpora_front /bin/bash
+
+# Inside contianer:
+
+# Add password for MySql user 'corpora' into both config.ini and config.json
+
+# Fill the database
+zcat database-dev.sql.gz | mysql -Dcorpora -uroot -p
+
+# Add cron tasks
+crontab scripts/conf/crontab.list
+
+exit
+```
+

docker-compose.yml

@@ -0,0 +1,24 @@
+version: "3.9"
+services:
+
+  opencorpora_front:
+    build: .
+    container_name: opencorpora_front
+    restart: always
+    ports:
+      - 80:80
+
+  opencorpora_db:
+    image: mysql:latest
+    container_name: opencorpora_db
+    restart: always
+    cap_add:
+      - SYS_NICE  # CAP_SYS_NICE
+    environment:
+      MYSQL_ROOT_PASSWORD: 'test_pass' # TODO: Change this
+    volumes:
+      - ocsqldata:/var/lib/mysql
+
+volumes:
+  ocsqldata:
+

scripts/conf/crontab.list

@@ -1,33 +1,33 @@
 # m h  dom mon dow   command
 # less than once a day
-0 6 1 * * php /corpus/scripts/check_dog_achievement.php >> /home/madfriend/error.log 2&>1
+0 6 1 * * php /corpus/scripts/check_dog_achievement.php >> /corpus/logs/madfriend_error.log 2&>1
 30 3 * * 1 /corpus/export/database/backup.sh 2>>/corpus/logs/db-backup.stderr.log
 
 # once a day
 # date to logs
-0 0 * * * date +\%F >>/home/grand/error.log
+0 0 * * * date +\%F >>/corpus/logs/grand_error.log
 0 0 * * * date +\%F >>/corpus/logs/consistency.stderr.log
 0 0 * * * date +\%F >>/corpus/logs/db-backup.stderr.log
 
 # other
-30 2 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/run_validators.sh" 2>>/home/grand/error.log
-#0 3 * * * nice /corpus/scripts/tokenizer/collect_data.pl /corpus/config.ini --dry-run >/dev/null 2>/home/grand/tokenizer.log
-0 4 * * * /corpus/scripts/backup.sh 2>>/home/grand/error.log
-30 4 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/export_and_stats.sh" 2>>/home/grand/error.log
-30 6 * * * nice /corpus/scripts/cronrunner.pl "/corpus/export/pools/export_pools.sh" 2>>/home/grand/error.log
+30 2 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/run_validators.sh" 2>>/corpus/logs/grand_error.log
+#0 3 * * * nice /corpus/scripts/tokenizer/collect_data.pl /corpus/config.ini --dry-run >/dev/null 2>/corpus/logs/grand_tokenizer.log
+0 4 * * * /corpus/scripts/backup.sh 2>>/corpus/logs/grand_error.log
+30 4 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/export_and_stats.sh" 2>>/corpus/logs/grand_error.log
+30 6 * * * nice /corpus/scripts/cronrunner.pl "/corpus/export/pools/export_pools.sh" 2>>/corpus/logs/grand_error.log
 0 7 * * * /corpus/scripts/invalidate_auth_tokens.pl /corpus/config.ini
-5 7 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/populate_pools.pl /corpus/config.ini" 2>>/home/grand/error.log
-20 7 * * * php /corpus/scripts/ma_pools/autopublish.php 2>>/home/grand/error.log
+5 7 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/populate_pools.pl /corpus/config.ini" 2>>/corpus/logs/grand_error.log
+20 7 * * * php /corpus/scripts/ma_pools/autopublish.php 2>>/corpus/logs/grand_error.log
 45 7 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/train_tokenizer.sh" >/dev/null 2>/dev/null
 
 # several times a day
-2,22,42 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/update_user_rating.py /corpus/config.json" 2>>/home/grand/error.log
-20 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/unpublish_pools.py /corpus/config.json" 2>>/home/grand/error.log
-15 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/tokenizer/check_sentence_split.pl /corpus/config.ini" 2>>/home/grand/error.log
-0,30 0-3,7-23 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/stats/update_annot_stats.pl /corpus/config.ini" 2>>/home/grand/error.log
-* * * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/dict_check.pl /corpus/config.ini" 2>>/home/grand/error.log
+2,22,42 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/update_user_rating.py /corpus/config.json" 2>>/corpus/logs/grand_error.log
+20 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/unpublish_pools.py /corpus/config.json" 2>>/corpus/logs/grand_error.log
+15 * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/tokenizer/check_sentence_split.pl /corpus/config.ini" 2>>/corpus/logs/grand_error.log
+0,30 0-3,7-23 * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/stats/update_annot_stats.pl /corpus/config.ini" 2>>/corpus/logs/grand_error.log
+* * * * * nice /corpus/scripts/cronrunner.pl "/corpus/scripts/dict_check.pl /corpus/config.ini" 2>>/corpus/logs/grand_error.log
 * * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/consistency/form2lemma.pl /corpus/config.ini" 2>>/corpus/logs/consistency.stderr.log
 * * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/consistency/form2tf.pl /corpus/config.ini" 2>>/corpus/logs/consistency.stderr.log
 * * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/consistency/dict_update_forms.pl /corpus/config.ini" 2>>/corpus/logs/consistency.stderr.log
-* * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/merge_pools_data.py /corpus/config.json" 2>>/home/grand/error.log
+* * * * * /corpus/scripts/cronrunner.pl "/corpus/scripts/ma_pools/merge_pools_data.py /corpus/config.json" 2>>/corpus/logs/grand_error.log
 # empty line must be here