facebook/flow · critical

Unknown exception reading from the server: {}

Error message

Unknown exception reading from the server: {}

What it means

The flow CLI client reads MonitorToClientMessage frames from the running flow server over a socket. Known disconnect-style io::Error kinds are handled gracefully (spinner line cleared, Err(()) returned so the retry loop can reconnect); this panic fires for any other read error, i.e. an unexpected OS-level or decode failure while waiting for a server response.

Source

Thrown at rust_port/crates/flow_cli/src/command_utils.rs:3058

                            | std::io::ErrorKind::UnexpectedEof
                    ) =>
                {
                    if !quiet && flow_utils_tty::spinner_used() {
                        let stderr = std::io::stderr();
                        let mut stderr = stderr.lock();
                        flow_utils_tty::print_clear_line(&mut stderr)
                            .expect("failed to clear spinner line");
                    }
                    return Err(());
                }
                Err(e) => {
                    if !quiet && flow_utils_tty::spinner_used() {
                        let stderr = std::io::stderr();
                        let mut stderr = stderr.lock();
                        flow_utils_tty::print_clear_line(&mut stderr)
                            .expect("failed to clear spinner line");
                    }
                    panic!("Unknown exception reading from the server: {}", e)
                }
            };
            match response {
                MonitorToClientMessage::PleaseHold(server_status, watcher_status) => {
                    let status_string = {
                        if server_status::is_free(&server_status) {
                            // Let's ignore messages from the server that it is free.
                            // It's a confusing message for the user
                            if watcher_status.1 == file_watcher_status::StatusKind::Ready {
                                None
                            } else {
                                Some(file_watcher_status::string_of_status(&watcher_status))
                            }
                        } else {
                            Some(server_status::string_of_status(
                                use_emoji,
                                false,
                                &server_status,

View on GitHub (pinned to f88ac94bcf)

Solutions

1. Stop the stale server (flow stop, or kill the daemon) and rerun the command so a fresh server spawns
2. Check server logs and dmesg for OOM kills or crashes; reduce memory pressure (fewer workers, smaller project, more RAM)
3. Ensure the CLI and the daemon resolve to the same flow installation/version (one flow on PATH)
4. If it persists, capture the underlying io::Error text from the panic and report it

Example fix

# before
flow check   # panics: Unknown exception reading from the server: <e>

# after
flow stop
flow check

When it happens

Trigger: Reading a MonitorToClientMessage from the server socket returns an io::Error whose kind is not in the handled set: server crashes mid-write leaving truncated data, socket reset in an unusual way, garbage bytes causing a decode failure, or transport errors during machine suspend/resume.

Common situations: Flow server OOM-killed while processing a large check; CLI and server from different flow installations sharing a socket dir (protocol mismatch); laptop suspend/resume dropping the connection; stale daemon left running after a flow upgrade.

Related errors

• Error sending command to server: {}
• failed to dup server->monitor channel
• failed to dup monitor->server channel
• failed to write {}: {}
• failed to read executable at {} for flow build id: {}

AI-assisted analysis of facebook/flow@f88ac94bcf (2026-08-20). Data as JSON: /api/errors/6ecaae651904b0dd. Report an issue: GitHub.