backups etc should be set up again
Opened this issue · 1 comments
I think the OLD backup scripts are still running regularly, but we should instead be backing up the NEW site. Let's get this fixed.
I'm putting this into the PM tracker at this point because we need to document how the backups work - and in general, document as much as we can about the way the server works too!
Old scripts:
From Aaron:
45 1 * * * /var/www/noosphere/bin/pm_backup.pl /dev/null 2>&1
is in the pm user's crontab.
This is relevant to #33.
Roughly speaking, all that typically needs to happen is a mysqldump on the upstream site (planetmath.org) and a mysql restore on the downstream site (metameso.org/beta). There are also drush commands that will back up the entire codebase, but as things stabilize, we would typically expect planetmath.org to be downstream to code changes, rather than upstream.
Also, the drush commands have been at least a little tempermental at times; see https://drupal.org/node/1953524.
Roughly speaking, the old script(s) should work with minor modifications, we just need to find out where they are...
Long term, it would be nice to look into things like MySQL replication, and Git/SVN integration (MathHubInfo/Legacy-planetary#68) - or any other features that would keep from having to regenerate everything everything just to make a coherent backup copy.
In particular, rather than putting time into re-formulating the snapshotting system, I think we should just work to get Git logging and and git2svn exporting set up, then the "snapshot" will be the SVN repository.
Another issue
Simply git pulling code from the Github tends to overwrite files that have special per-server configurations in them (e.g. paths in pyrdfa or planetmath_view_pdf). That's not good -- the special configurations should really live someplace where they can be restored after a git pull.