feat: websocket for health

[tribhuwan-kumar]

feat: #1218 (comment)

maybe we have to update all the instance of fetchHealth & debouncedFetchHealth in all react components,
anyway i still kept these functions in health hook even tho they aren't necessary now

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
screenpipe	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Feb 3, 2025 3:47pm

[louis030195]

might be risky port, maybe 14346? or can it be same port than the tauri server? not sure if it conflict with normal http server

[louis030195]

btw why not using axum like we do already?

https://github.com/tokio-rs/axum/blob/main/examples/websockets/src/main.rs

[louis030195]

screenpipe/screenpipe-server/src/server.rs

Line 1880 in 9311d63

async fn handle_socket(socket: WebSocket, query: Query<EventsQuery>) {

[louis030195]

hmm actually i have second thought

what about we do this health event on our existing event websocket?

[tribhuwan-kumar]

yes, that'd be more easy!

[tribhuwan-kumar]

what about we do this health event on our existing event websocket?

screenpipe/screenpipe-server/src/server.rs

Line 1880 in 9311d63

async fn handle_socket(socket: WebSocket, query: Query<EventsQuery>) {

i tried to implement it but the problem is when screenpipe backend goes down the websocket server goes down too, cz the ws server is from backend,

screenpipe/screenpipe-server/src/server.rs

Line 1980 in 9311d63

.route("/ws/events", get(ws_events_handler))

after that its not possible to emit the health error which is something like this

    let error_health = HealthCheckResponse {
    status: "error".to_string(),
    last_frame_timestamp: None,
    last_audio_timestamp: None,
    last_ui_timestamp: None,
    frame_status: "error".to_string(),
    audio_status: "error".to_string(),
    ui_status: "error".to_string(),
    message: format!("health check failed: {}", e),
    status_code: 500,
    verbose_instructions: None,
};

one possible solution it'd be emit the error health before doing the shutdown, but it can create problem with react states!

[louis030195]

i guess the hook would just show error when ws is disconnected

[tribhuwan-kumar]

@louis030195
now the health websocket is on /ws/health endpoint of app's server

[louis030195]

i meant as custom event

https://github.com/mediar-ai/screenpipe/blob/main/screenpipe-events/src/events_manager.rs

[tribhuwan-kumar]

i meant as custom event

https://github.com/mediar-ai/screenpipe/blob/main/screenpipe-events/src/events_manager.rs

yes, I tried it but I'm facing problems to getting values of error health

#1256 (comment)

[tribhuwan-kumar]

i guess the hook would just show error when ws is disconnected

@louis030195
then what's the benefit of using webSocket if we have to keep connecting & disconnecting :(

[tribhuwan-kumar]

@louis030195

any second thought?
also in windows it takes ~10s just to recognize is backend down or not this pr can fix this.