Every live market is subscribed on two connections chosen at random, and its book takes each change from whichever connection delivers it first: the book hash says whether a change is a duplicate, the next one, or one to hold until the change before it arrives. A change one connection misses costs nothing, each Delta carries the earliest receive time, and a connection that drops is covered by the other. When a change fits on no connection in time, the book follows one connection unchecked and asks the other for a fresh book. Connections now decode and route events to 16 book threads, independent of them; a move leaves its old connection only once the new one delivers, so it costs no change; a connection closed with 1013 twice within ten minutes is split. reconcile reads the collector's log and fails only on a mismatch no cause line explains. SIGTERM and SIGHUP stop the collector cleanly. Breaking: lanes are gone (connections and book threads are separate), and reconcile takes the log as a second argument.
2.0 KiB
0001. Concurrency model: threads, shared-nothing
Context
The collector moves a firehose of websocket messages from many sockets to a
single output sink. This is the project's central bet (see
../project.md): it must be correct (no dropped messages, no
races) and time-accurate (receive time stamped at the socket edge) before it is
fast. The runtime is the one foundational, costly-to-change choice. The template
already uses std threads + crossbeam channels and pulls in no tokio.
Decision
V1 uses OS threads, not an async runtime — a sync websocket client
(tungstenite), with one read loop per socket. State is shared-nothing: no
mutexes. Each pipeline stage either owns its state on a single thread or is
sharded so each shard owns its slice, and stages communicate over channels.
Async (tokio) is deferred as a possible optimization, re-evaluated at the start of extension 1 (M4), when socket-management complexity (reconnect, timeouts, redundancy) makes async ergonomics worth their cost.
Why
Thread-per-socket is comfortable at this scale (symbols shard across tens of sockets, not thousands) and is the simplest, most readable option — correctness first, speed later. Shared-nothing + channels removes lock contention and data races by construction, and keeps the runtime pluggable: only the socket layer is runtime-specific, so a later async swap is localized to that boundary rather than a pipeline rewrite. Deferring async is therefore safe through the base challenge.
The M4 checkpoint was held in sprint 04 (2026-11-11): threads stay. The new logic of redundancy sits in the book threads, not in socket handling, and about 200 threads (175 connections, 16 book threads) are comfortable.
Trade-off: thread-per-socket does not scale to very large socket counts, and the extension-era socket-management logic (select / timeout / reconnect) is where threads and async diverge most — hence the explicit M4 re-evaluation point rather than "never."