Server spams logs hard with "Network is unreachable"

The libSQL server logs in production are often spammed hard with log entries such as:

[info] 2023-11-08T20:52:12.368098Z WARN load_region{provider=ImdsRegionProvider { client: LazyClient { client: OnceCell { value: None }, builder: Builder { max_attempts: None, endpoint: None, mode_override: None, token_ttl: None, connect_timeout: None, read_timeout: None, config: Some(ProviderConfig { env: Env(Real), fs: Fs(Real), sleep: Some(TokioSleep), region: None }) } }, env: Env(Real) }}: aws_config::imds::region: failed to load region from IMDS err=failed to load IMDS session token: dispatch failure: io error: error trying to connect: tcp connect error: Network is unreachable (os error 101): tcp connect error: Network is unreachable (os error 101): Network is unreachable (os error 101) (FailedToLoadToken(FailedToLoadToken { source: DispatchFailure(DispatchFailure { source: ConnectorError { kind: Io, source: hyper::Error(Connect, ConnectError("tcp connect error", Os { code: 101, kind: NetworkUnreachable, message: "Network is unreachable" })), connection: Unknown } }) }))

Start by tracing how the libSQL server configures the AWS IMDS region provider and where the reported production warning is emitted. Confirm the conditions that cause repeated “Network is unreachable” entries; done means the server no longer spams these logs while preserving useful diagnostics for genuine failures.