borg-backup: implement saving a stream, use for database backups

Add facility to borg-backup role to run a command and save the output
of it to a separate archive file during the backup process.

This is mostly useful for database backups.  Compressed on-disk logs
are terrible for differential backups because revisions have
essentially no common data.  By saving the uncompressed stream
directly from mysqldump, we allow borg the chance to de-duplicate,
saving considerable space on the backup servers.

This is implemented for our ansible-managed servers currently doing
dumps.  We also add it to the testinfra.

This also separates the archive names for the filesystem and stream
backup with unique prefixes so they can be pruned separately.
Otherwise we end up keeping only one of the stream or filesystem
backups which isn't the intention.  However, due to issues with
--append-only mode we are not issuing prune commands at this time.

Note the updated dump commands are updated slightly, particularly with
"--skip-extended-insert" which was suggested by mordred and
significantly improves incremental diff-ability by being slightly more
verbose but keeping much more of the output stable across dumps.

Change-Id: I500062c1c52c74a567621df9aaa716de804ffae7

inventory/service/host_vars/etherpad01.opendev.org.yaml

@@ -8,3 +8,5 @@ etherpad_redirect_vhost: etherpad.openstack.org
 borg_backup_excludes_extra:
   # live db; we store daily dumps
   - /var/etherpad/*
+  # local db backups, we store stream
+  - /var/backups/etherpad-mariadb

inventory/service/host_vars/gitea01.opendev.org.yaml

@@ -7,3 +7,5 @@ borg_backup_excludes_extra:
   - /var/gitea/data/
   # db is backed up in dumps, don't capture live files
   - /var/gitea/db
+  # backed up by streaming backup
+  - /var/backups/gitea-mariadb

inventory/service/host_vars/review01.openstack.org.yaml

@@ -76,3 +76,5 @@ borg_backup_excludes_extra:
   - /home/gerrit2/review_site/cache/*
   - /home/gerrit2/review_site/tmp/*
   - /home/gerrit2/review_site/index/*
+  # dump directly via stream
+  - /home/gerrit2/mysql_backups/*

playbooks/roles/borg-backup/README.rst

@@ -15,6 +15,12 @@ correctly on the backup server.  This role sets a tuple ``borg_user``
 with the username and public key; the ``borg-backup-server`` role uses this
 variable for each host in the ``borg-backup`` group to initalise users.
 
+Hosts can place into ``/etc/borg-streams`` which should be a script
+that outputs to stdout data to be fed into a backup archive on each
+run.  This will be saved to an archive with the name of the file.
+This is useful for raw database dumps which allow ``borg`` to
+deduplicate as much as possible.
+
 **Role Variables**
 
 .. zuul:rolevar:: borg_username

playbooks/roles/borg-backup/templates/borg-backup.j2

@@ -9,6 +9,7 @@ if [ -z "$1" ]; then
 fi
 
 BORG="/opt/borg/bin/borg"
+BORG_CREATE="${BORG} create --verbose --filter AME --list --stats --show-rc --compression lz4 --exclude-caches "
 
 # Setting this, so the repo does not need to be given on the commandline:
 export BORG_REPO="ssh://{{ borg_username}}@${1}/opt/backups/{{ borg_username }}/backup"
@@ -24,31 +25,35 @@ export BORG_UNKNOWN_UNENCRYPTED_REPO_ACCESS_IS_OK=1
 
 # Backup the most important directories into an archive named after
 # the machine this script is currently running on:
-${BORG} create                      \
-    --verbose                       \
-    --filter AME                    \
-    --list                          \
-    --stats                         \
-    --show-rc                       \
-    --compression lz4               \
-    --exclude-caches                \
+${BORG_CREATE} \
 {% for item in borg_backup_excludes + borg_backup_excludes_extra -%}
     --exclude '{{ item }}'          \
 {% endfor -%}
-                                    \
-    ::'{hostname}-{now}'            \
+    ::'{hostname}-filesystem-{now}' \
 {% for item in borg_backup_dirs + borg_backup_dirs_extra -%}
     {{ item }} {{ '\\' if not loop.last }}
 {% endfor -%}
 
 backup_exit=$?
 
-if [ ${backup_exit} -eq 0 ]; then
-    info "Running prune"
-    ${BORG} prune --verbose --list --prefix '{hostname}-' \
-            --show-rc --keep-daily 7 --keep-weekly 4 --keep-monthly 12
-    backup_exit=$?
+for f in $(shopt -s nullglob; echo /etc/borg-streams/*)
+do
+    stream_name=$(basename $f)
+    info "Backing up stream archive $stream_name"
+    bash $f | ${BORG_CREATE} --stdin-name ${stream_name} \
+                      ::"{hostname}-${stream_name}-{now}" -
+
+    if [[ ${PIPESTATUS[0]} -ne 0 ]]; then
+        info "Streaming script ${f} failed!"
+        stream_exit=${PIPESTATUS[0]}
+    elif [[ ${PIPESTATUS[1]} -ne 1 ]]; then
+        info "Borg failed!"
+        stream_exit=${PIPESTATUS[1]}
+    else
+        stream_exit=0
     fi
+    (( backup_exit = backup_exit || stream_exit ))
+done
 
 if [ ${backup_exit} -eq 0 ]; then
     info "Backup finished successfully"

playbooks/roles/etherpad/tasks/main.yaml

@@ -123,7 +123,7 @@
     owner: root
     group: root
 
-- name: Set up cron job to backup the database
+- name: Set up cron job for local database backup
   cron:
     name: etherpad-db-backup
     state: present
@@ -142,3 +142,17 @@
     logrotate_rotate: 2
     logrotate_file_name: /var/backups/etherpad-mariadb/etherpad-mariadb.sql.gz
     logrotate_compress: false
+
+- name: Setup db backup streaming job
+  block:
+    - name: Create backup streaming config dir
+      file:
+        path: /etc/borg-streams
+        state: directory
+
+    - name: Create db streaming file
+      copy:
+        content: >-
+          /usr/local/bin/docker-compose -f /etc/etherpad-docker/docker-compose.yaml exec -T mariadb
+          bash -c '/usr/bin/mysqldump --skip-extended-insert --databases etherpad-lite --single-transaction -uroot -p"$MYSQL_ROOT_PASSWORD"'
+        dest: /etc/borg-streams/mysql

playbooks/roles/gerrit/tasks/main.yaml

@@ -338,3 +338,16 @@
     job: 'find /home/gerrit2/review_site/logs/*.gz -mtime +30 -exec rm -f {} \;'
     minute: 1
     hour: 6
+
+- name: Setup db backup streaming job
+  block:
+    - name: Create backup streaming config dir
+      file:
+        path: /etc/borg-streams
+        state: directory
+
+    - name: Create db streaming file
+      copy:
+        content: >-
+          /usr/bin/mysqldump --defaults-file=/root/.gerrit_db.cnf --skip-extended-insert --ignore-table mysql.event --all-databases --single-transaction
+        dest: /etc/borg-streams/mysql

playbooks/roles/gitea/tasks/main.yaml

@@ -186,3 +186,16 @@
   vars:
     logrotate_file_name: /var/backups/gitea-mariadb/gitea-mariadb.sql.gz
     logrotate_compress: false
+- name: Setup db backup streaming job
+  block:
+    - name: Create backup streaming config dir
+      file:
+        path: /etc/borg-streams
+        state: directory
+
+    - name: Create db streaming file
+      copy:
+        content: >-
+            /usr/local/bin/docker-compose -f /etc/gitea-docker/docker-compose.yaml exec -T mariadb
+            bash -c '/usr/bin/mysqldump --skip-extended-insert --databases gitea --single-transaction -uroot -p"$MYSQL_ROOT_PASSWORD"'T_PASSWORD"'
+        dest: /etc/borg-streams/mysql

playbooks/test-borg-backup.yaml

@@ -0,0 +1,14 @@
+- hosts: "borg-backup"
+  tasks:
+    - name: Setup db backup streaming job
+      block:
+        - name: Create backup streaming config dir
+          file:
+            path: /etc/borg-streams
+            state: directory
+
+        - name: Create sample streaming file
+          copy:
+            content: >-
+              dd if=/dev/urandom bs=1M count=5
+            dest: /etc/borg-streams/random

zuul.d/system-config-run.yaml

@@ -347,7 +347,10 @@
     vars:
       run_playbooks:
         - playbooks/service-borg-backup.yaml
+      run_test_playbook: playbooks/test-borg-backup.yaml
     files:
+      - playbooks/service-borg-backup.yaml
+      - playbooks/test-borg-bcakup.yaml
       - playbooks/install-ansible.yaml
       - playbooks/roles/install-borg
       - playbooks/roles/borg-backup