mirror of
https://github.com/neondatabase/neon.git
synced 2026-05-23 16:10:37 +00:00
066324d6ec
Move most of the code to compute.rs, so that all the major startup steps are visible in one place. You can now get a pretty good picture of what happens in the latency-critical path at compute startup by reading ComputeNode::start_compute(). This also clarifies the error handling in start_compute. Previously, the start_postgres function sometimes returned an Err, and sometimes Ok but with the compute status already set to Failed. Now the start_compute function always returns Err on failure, and it's the caller's responsibility to change the compute status to Failed. Separately from that, it returns a handle to the Postgres process via a `&mut` reference if it had already started Postgres (i.e. on success, or if the failure happens after launching the Postgres process). --------- Co-authored-by: Alexey Kondratov <kondratov.aleksey@gmail.com>