These are the basic strategies used to backup Riak's data (to protect yourself from cases like total cluster failure or data deletion by tired developer).
The common issues to note:
- The backed up node should be stopped.
- Backups are inconsistent (no coordination between backuped nodes)
- Data directories depends on OS (the list in official docs)
Note: If you have installed riak on OS X using homebrew, the data directory is something like
Note: If you want to setup the local cluster on OS X and you have installed riak with homebrew, follow this tutorial.
The general info available here: https://github.com/drewkerrigan/riak-differential-backup
- Copy only the keys, that were modified from the previous backup
- Riak Migration Tool-based (may be an issue: see Caveats)
- More complex, compared to other options
Filesystem data dirs copy
This approach is the one in the official documentation : http://docs.basho.com/riak/latest/ops/running/backups/
- Official recommendation
- Different paths on different OSes
- backup the full cluster data
Using riak-admin cli tool
This strategy was recently deprecated in favor of filesystem backup, described former. See docs for details
riak-admin backup <node name> <erlang cookie> <file name with path> all