All the docs say data loss is inevitable in a failure scenario for a single server setup however they don’t say what that failure scenario actually is. If consul itself is the only thing that crashes, is it still going to lose data or would it require a catastrophic failure that wipes out storage, or something in between?
If the Consul process dies, probably no data loss. But if you run it out of memory and get something corrupt, absolutely. That can also happen with a cluster.
But the docs are stressing, if you lose a storage device on a single node cluster you’re going out of service until you restore the device and restore your snapshot of data.
Do you have a use case that somehow needs a single-instance Consul deployment?
Thanks for the clarification.
The use case is really only configuration & state, but I’d like to be able to scale down to one server(everything, not just Consul), at which point if something catastrophic happens like a disk failing or running out of memory, Consul crashing is probably the least of my worries. Consul feels like the best solution for what I’m doing but I’m considering just using a file when there’s only one server running.