Zachery Aaron Shores-Chmielewski
599678e7c9
Introduce a gossip-converged pooled datastore protocol layered on SWIM piggybacking, backed by a reusable gossip-channel abstraction, plus a Docker-free multi-process cluster test runner. - crates/distribution/src/gossip_channel.rs: add the GossipChannel trait (piggyback on SWIM messages) and a budget-limited DisseminationBuffer<T> that replaces the four duplicated Lambda*ceil(log2(n)) dissemination copies - crates/datastore/src/pool: add PoolDisseminator (CRDT state for membership/capacity/content-location/ACL with join/leave/announce) and PoolCoordinator (placement-aware CRUD actor delegating to the co-located DatastoreNode) - crates/shared-types/src/pool.rs: add shared pool protocol types (PoolId plus member/capacity/content-location/ACL entries and PoolConfig) consumed by both distribution and datastore - crates/dashboard/src/pool_html.rs: add a live pool dashboard page (membership, capacity, content locations) and add pool_tests integration coverage - xtask/src/sim_cluster.rs: add the sim-cluster runner that spawns N swactor nodes over iroh through a local relay server, reusing the docker cluster scenarios without Docker - crates/distribution/src/iroh_driver.rs: add relay-URL resolution (cache, then SWIM gossip, then home relay) with a 2s connect timeout to back the relay-based connections Signed-off-by: Zachery Aaron Shores-Chmielewski <zacheryasc@gmail.com> |
||
|---|---|---|
| .. | ||
| examples | ||
| src | ||
| tests | ||
| .gitignore | ||
| AGENTS.md | ||
| Cargo.toml | ||
| README.md | ||
runtime-dashboard
Visual dashboard for the swactor runtime. Provides a live HTTP dashboard, a terminal UI (TUI), trace recording/replay, and an HTTP API for programmatic runtime investigation.
Features
| Feature | Default | Description |
|---|---|---|
distribution |
yes | /distribution page with SWIM membership, Kademlia routing, and location cache |
tui |
no | Terminal UI with overview, worker detail, and distribution views |
HTTP Dashboard
Start the dashboard demo and open it in a browser:
cargo run -p runtime-dashboard --example dashboard_demo
Pages:
http://localhost:9090— live overview (workers, actors, message rates)http://localhost:9090/actors— actor tablehttp://localhost:9090/distribution— SWIM membership, Kademlia routing, cache entries
The demo creates a 4-worker runtime with ping-pong and counter actors, plus a 9-node distribution cluster (1 main node + 8 peers) with simulated SWIM membership and actor registrations in the directory/cache.
TUI
A standalone binary that connects to any running dashboard over SSE:
cargo run -p runtime-dashboard --features tui --bin swactor-tui
# or point at a specific endpoint
cargo run -p runtime-dashboard --features tui --bin swactor-tui -- http://localhost:9090
Views (cycle with Tab):
- Overview — htop-style worker bars, summary line, sortable actor table
- Worker Detail — focused view of a single worker's actors and phase breakdown
- Distribution — cluster summary, scrollable members table, cache entries, routing bucket histogram
Key bindings: q quit, Tab cycle views, s sort column, r reverse sort,
arrow keys/j/k scroll, Enter drill into worker, Esc back to overview.
Agent HTTP API (Investigate)
All diagnostic commands are available as HTTP endpoints when the dashboard server is running. See AGENTS.md for full protocol documentation.
curl 'http://localhost:9090/api/investigate?cmd=overview'
curl 'http://localhost:9090/api/investigate?cmd=hot&n=5'
curl 'http://localhost:9090/api/investigate?cmd=workers'
curl 'http://localhost:9090/api/investigate?cmd=worker&id=2'
curl 'http://localhost:9090/api/investigate?cmd=actors&sort=mailbox&limit=10'
curl 'http://localhost:9090/api/investigate?cmd=diff&seconds=2'
The same commands are also available via a stdin/stdout REPL for direct
programmatic use (see investigate::run_investigate).
Demos
All examples are run from the workspace root.
HTTP dashboard — live workload with distribution cluster, Ctrl+C to stop:
cargo run -p runtime-dashboard --example dashboard_demo
# http://localhost:9090 — runtime overview
# http://localhost:9090/distribution — cluster view
Benchmarks — four automated scenarios (~20 s total):
cargo run -p runtime-dashboard --example bench_dashboard
# open http://localhost:9090
Record & replay — records ~10 s of activity, then serves a replay:
cargo run -p runtime-dashboard --example record_and_replay_demo
# live dashboard at http://localhost:9090 during recording
# replay dashboard at http://localhost:9091 after recording finishes
# Ctrl+C to stop