When quorum is lost does that mean data is potentially lost aswell?

As i understand database is valid only if quorum is maintained. So if a software bug brings down the entire cluster does that mean only way to get back up and running is backup? What about planned shutdown of entire cluster? When i shutdown nodes one by one and quorum was lost what happens to data? Can i start a single node from cluster after full shutdown and access the data it has?

1 Like

Sorry for the delay in response @rene98c !

So if a software bug brings down the entire cluster does that mean only way to get back up and running is backup?

It depends. A bug is undefined behavior, so if you are asking what would happen if there is a bug within CockroachDB itself, then it’s hard to say. Of course we do our best to ensure that we don’t create bugs that bring down the whole cluster, and we try to write code in a way to limit the impact of such bugs. But yes, in the worst-case, there could be a hypothetical bug that requires you to restore from a backup.

What about planned shutdown of entire cluster? When i shutdown nodes one by one and quorum was lost what happens to data?

In this case, your data is assured to be safe. Check out the docs on using quit to stop a node. As long as nodes shut down gracefully, then the data will be persisted on disk, and you can bring up your cluster later. At some point, if quorum cannot be reached, then the cluster will stop responding to writes. You can see the docs on replication for more details.

Can i start a single node from cluster after full shutdown and access the data it has?

Yes this should work, but I may not fully understand the question. Could you provide an example (i.e. some sort of script) of what you are trying to do?