Fix backup archive creation. (#1718)

Use exit error codes.

Fail if target path is a file.

Create target directory if it doesn't exist.

Use target directory at /data.

Add setting for git_ssh path.

Clarify sudo usage in the comments.

Add a dry run check

Remove development system handling.

Fix archive extraction.

Lots of fixes.

Escape double quotes in arguments.

Use sudo again.

Adjust the script to be run as root user.

Fix postgres dump not working on development system (wrong user).

Drop last leftover of open3 in favor of system().

Only restore data we own.

Use psql ontohub user instead of postgres.

Detect and use basename of data_root.

Add possibility to specify the data path via the environment.

Adjust data path to new deployments.

Adjust backup path to new deployments.

Add reference to the more beautiful documentation.

Check maintenance mode before activating it.

Omit the parameter z of tar: It is inferred by the file type.

Use pg_dump with file parameter This renders the Subprocess class obsolete. Instead, we use system now.

Disable running the backup script as root.

Fix typo.

Improve code style.

Add documentation to the backup script.

Fix arguments for pg dump.

Obey Hound.

Remove pry.

Adjust constants for pruning.

fix wrong sudo password issue

Print more information on the whole process.

Don't rescue from error when moving data dirs.

Use verbose flag for tar command as well.

print errors on stderr

Use open3 to send system calls.

Backup data dirs on restore, set correct permissions Backup of data dirs: When restoring a backup, we need to remove the currently active data directories in order to restore the ones from the backup archive. This is done by moving them to a temp directory, then extracting the archive and deleting the temp directory. This way, if something goes wrong during extraction, one can still recover the pre-restore repositories. Since they are git repositories and we don't allow force-pushes, they may still be of use. Permissions: To reset the permissions on the extracted repositories, super user privileges are needed. This is accomplished by first trying to get those privileges for the extraction. If the user refuses to enter the password (wrong one three times or Ctrl-C), the extraction will be done as the normal user. This way, the repository archive is still extracted, but all the files belong to the restorign user.

chmod -R (allow deletion) before rm -r

Simplify pg_{dump,restore} calls.

Don't sudo -u postgres call on pg_{dump,restore}.

Set different backup directory on develop system.

Add pruning of backups.

put backups into ontohub home

add helper method for maintenance file path

rename module to Backup

add backup script