Database recovery on same cluster but with one less node.

>> but what happens if one node is damaged and you want to restore last backup on the same cluster with one less node?
Where is a problem or question? Restore backup should repair failed node - its a goal of backup. Don't you agree with me?

Hi Amador,

You and Daniel have the right idea here -- if you take a full database backup you need to restore it to the same number of nodes.  If a node has failed, the restore will therefore work fine as soon as you have replaced the node, and not before.

In general, because we optimize our data layout so heavily according to the design of your cluster, it's much more efficient to stamp the whole cluster back into place all at once than to try to shuffle data around to compensate for a node that is down.

We're aware that people would like to restore to different-sized clusters.  Take a look a the upcoming Vertica 7.0 release as soon as you have access to it; it has some new backup/restore functionality.  More new functionality will likely continue to appear in future releases.

Adam

Hi Daniel , When I referred to a damaged node I did not mean at the database level I rather meant that for instance the node broke down because of a hardware failure and is not able to start , Should I hold a good node out of the cluster to take over a damaged node and be able to restore a previous backup because someone dropped a table too ? , I know it is an unlikely chain of events but I really think that you should be allowed to recover a previous backup regardless of the nodes left after a disaster.

Thanks Adam , I am looking forward to put my hands on Vertica 7.0.