On-Premises Private

CARTO can be installed on a local server, or set of servers, either directly or using our commercial installer.

Backup and Recovery with CARTO Builder On-Premises

It is possible (and strongly recommended) to perform periodical backups of CARTO Builder On-Premises. This will allow you to recover both data, and configurations, in case of disaster.

Note
Two scripts have been provided for these tasks. Unless your sudo configuration allows you to load commands from onpremises paths loaded, these scripts are not going to work properly under sudo. To ensure that they work correction, run them directly as root.

Backup

Warning
This tool will not check if there is enough disk space to host the entire backup. The server should have (in the configured backup path) at least the same disk space used for the data (the default /data) directory.

backup.rb - It offers two methods for creating a backup: rsync and tar. Backups take everything under the directory provided by --data-dir and stores it in --backup-dir directory. The script has the ability to check if this is a mount point backup; which are done to an external drive, e.g network drive.

Depending on the backup method chosen, the following scenarios may apply:

rsync

If rsync is chosen, the backup directory provided by the --backup-dir flag (e.g --backup-dir /backup/production) that will be updated every time that the script is run. :

1
2
3
4
5
6
7
8
9
10

/opt/carto/tools/bin/cartoctl exec -- backup.rb --help
Usage: backup [options]
    -c, --check-mount                Check if backup dir is a mount point
    -r, --redis-retries=val          Retries to check Redis backup completed
    -m, --method=val                 Backup method: rsync (default) or tar
    -d, --data-dir=val               Data dir to backup
    -b, --backup-dir=val             Output dir to write backup to
    -h                               Shows this message

/opt/carto/tools/bin/cartoctl exec -- backup.rb --data-dir /data --backup-dir /backup

The script requires both --data-dir and backup-dir to be provided in order to run.

tar

If tar is chosen, there should be a separate tar file every time the backup script is run.

By default, backups are not performed automatically. The user is responsible for running the backup task periodically in cron, or any other method.

The following usage example describes how a backup script is applied:

1
2
3
4
5

$ /opt/carto/tools/bin/cartoctl exec -- backup.rb --data-dir /data --backup-dir /backup --method tar

$ ll /backup/
total 4316
-rw-r--r-- 1 root root 4417656 Sep 30 15:17 carto_backup_20190930151723.tar.gz

The script requires both --data-dir and backup-dir to be provided in order to run.

Restore

restore-backup.rb stops all services, takes a backup as generated by backup.rb tool, pointed to by the --rsync-dir or --tar-file, depending on which method has been used to make the backup, restores it into the directory provided by the --data-dir flag, and starts all services. By default, it tries to keep whatever --data-dir had unless told otherwise. Two different inputs can be provided:

  • A directory (made using rsync method).
  • A tar.gz file

The following usage example describes how a backup restore script is applied:

1
2
3
4
5
6
7

/opt/carto/tools/bin/cartoctl exec -- restore-backup.rb --help
Usage: restore-backup [options]
    -d, --rsync-dir=val              Rsync'ed dir to restore
    -f, --tar-file=val               Tar file to restore
    -o, --output-dir=val             Output data dir to restore backup to
    -r, --remove-data-dir            Removes data dir. We keep it by default
    -h                               Shows this message

It is required to pass either option -d or option -f, and the output directory (-o, –output-dir) to restore to.

The following example comes from the restore of a Builder installation, from a directory backup created with backup script:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

$ /opt/carto/tools/bin/cartoctl exec -- restore-backup.rb -d /backup/data/ -o /data

This script is going to restore /backup/data/
into /data local data dir. If the provided data dir exists, it will try
to keep it unless you have asked explicitly to not do it.

Please make sure backup comes from carto-backup.rb script in an unaltered form
or things won't work as expected

Do you want to proceed? (y/N)y
2019-09-30 14:58:04 UTC - INFO - ./tools/common-templates/restore-backup.rb starts
2019-09-30 14:58:04 UTC - INFO - Stopping all services
2019-09-30 14:58:04 UTC - DEBUG - Service postgresql stop
2019-09-30 14:58:04 UTC - DEBUG - Service redis stop
2019-09-30 14:58:04 UTC - INFO - Saving /data into /data.save
2019-09-30 14:58:04 UTC - DEBUG - Running command [mv /data /data.save]
2019-09-30 14:58:04 UTC - INFO - Recovering backup from /backup/data/
2019-09-30 14:58:04 UTC - DEBUG - Running command [rsync -a /backup/data /]
2019-09-30 14:58:05 UTC - DEBUG - Running command [mv /data/log/redis /data/log/redis.save]
2019-09-30 14:58:05 UTC - DEBUG - Running command [mkdir -p /data/log/redis]
2019-09-30 14:58:05 UTC - DEBUG - Running command [touch /data/log/redis/redis.log]
2019-09-30 14:58:05 UTC - DEBUG - Running command [chown redis: /data/log/redis/redis.log]
2019-09-30 14:58:05 UTC - INFO - Starting all services
2019-09-30 14:58:05 UTC - DEBUG - Service redis start
2019-09-30 14:58:05 UTC - DEBUG - Service postgresql start
2019-09-30 14:58:12 UTC - INFO - Remove /data.save if things are ok
2019-09-30 14:58:12 UTC - INFO - ./tools/common-templates/restore-backup.rb ends

In order to restore a server using a backup, the cleanest approach would be to install the product using the installer for the version. Then use the restore script from the latest backup.