Deploying with Litestream

[boiserunner]

We’re interested in giving Litestream a spin in a non-critical application. I read in the Fly blog post that Litestream is intended for single-node applications.

Does this mean it is not suitable for situations where we’re using a n>1 scale count that share a persistent storage volume? In our ideal scenario, this would mean my application would be running on multiple application servers, each with a read-write connection… to a single sqlite database.

I suppose as I write this I can think of at least one problem. We could end up with three Docker images, with three Litestream instances each running our application as a sub process. That seems not very good, unless I somehow constrain all writes to one application node.

Apologies if Fly is not yet the appropriate forum for this question.

Chris

[ignoramous]

I read in the Fly blog post that Litestream is intended for single-node applications.

litestream has two modes:

• single-node database with periodic backups to any supported blob store.
• single-node writer, with multiple read-replicas, which is in beta.

application would be running on multiple application servers, each with a read-write connection… to a single sqlite database.

This is tricky right now (here’s a reference dockerfile and relevant documentation), but I fully expect it to improve in the coming weeks! See this discussion.

For single-node setup on Fly, see this entry on litestream.io.

Apologies if Fly is not yet the appropriate forum for this question.

Fly is absolutely the approp forum for it, especially since Fly now more or less owns litestream.