dan at basho.com
Fri Jul 29 16:18:25 EDT 2011
When tarring up the data directories on nodes it's important to remember that the partition to node mapping as defined in the ring file will determine where a cluster looks for data.
Here is a method for migrating data to a new cluster:
Using "riak-admin reip" to restore Bitcask tarballs to a new cluster
1. Tarball the Bitcask and ring data directories for each node
* Package installed versions of Riak usually place this data in "/var/lib/riak/bitcask" and "/var/lib/riak/ring"
1. Deploy a set of Riak nodes; same number of nodes as the backed up cluster
* Don't start or cluster the nodes
2. Restore the Bitcask and ring data to the new set of nodes
3. On each node run "riak-admin reip" for all nodes (explained in the example below)
4. Start the Riak nodes
* there is no need to cluster the nodes; the restored ring file will form the cluster
We have a cluster with 4 nodes (1, 2, 3, 4) which we want to backup and restore to 4 new nodes (A, B, C, D)
Backup the Bitcask and ring data directories for nodes 1, 2, 3, 4
Install Riak on nodes A, B, C, D
On node A:
Restore the Bitcask and ring data directories from node 1
Reip the ring file:
riak-admin reip node1 nodeA
riak-admin reip node2 nodeB
riak-admin reip node3 nodeC
riak-admin reip node4 nodeD
Repeat on nodes B, C, D using the back ups from 2, 3, 4 respectively
* The reip commands will be the same on all nodes
Start nodes A, B, C, D
Nodes A, B, C, D will form a cluster that is a clone of nodes 1, 2, 3, 4
Backups must be restored to the same number of nodes from which they were taken
Sent from my iPhone
On Jul 29, 2011, at 3:22 PM, francisco treacy <francisco.treacy at gmail.com> wrote:
> Hi all,
> I'm setting up a brand-new staging server, with Riak 0.14.2.
> As I want to get all production data, I thought I could scp the data
> directory into the new one (besides the ring dir).
> So that's what I did, and then started the staging db for the first
> time. It doesn't really work though. Many documents are present, but
> for others I get 404s and possibly 500s as I see tons of crashes in
> the logs (certainly many related to Luwak).
> I believed this kind of operation was legal... Is there something I'm
> missing, or something that could help?
> Ideally, I want to transfer fresh production data to staging via a
> cron job on a daily basis.
> riak-users mailing list
> riak-users at lists.basho.com
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users