Feature Request: API to expose underlying memory profiling stats

[ReppCodes]

Hi!

I'm working on a project that already uses foundations and exposes a telemetry server running in a background tokio task, with memory profiling enabled. I have a use case where I'd like to query my current memory usage from within this process to make some runtime determinations. I'm able to do this by importing tikv_jemalloc_ctl directly and using its API, but it feels in-scope for foundations to expose this without needing to reach for the lower-level library.

Would it be possible to pass-through tikv_jemalloc_ctl::stats in some way, to query the actual stats available there from within the process using foundations?

Thanks for your work on this, love the library!

[inikulin]

Is https://docs.rs/foundations/latest/foundations/telemetry/struct.MemoryProfiler.html#method.heap_stats what you're looking for?

[ReppCodes]

Possibly, but I think probably not for 2 reasons.

1. This prints out large string that I then need to parse to get resident/allocated/whatever. Contrast that with tikv_jemallocater_ctl just handing me a size_t that I can work with easily, from let resident = stats::resident::read().unwrap();
2. I'm not sure how I would use that API if my profiler is stuck into a telemetry server serving endpoints on a separate tokio task. If I do the below to generate a future for the telemetry server, and then tokio::task::spawn(async move {}) it to a to task to await it there, how can I query the profiler with the heap_stats method you've linked?

        let metrics_server_fut = telemetry::init(TelemetryConfig {
            service_info: &ServiceInfo {
                name: "sql-query-cache",
                name_in_metrics: "sql_query_cache".to_string(),
                version: SQC_VERSION,
                author: "Hyperdrive team",
                description: "Hyperdrive edge service",
            },
            settings: &TelemetrySettings {
                metrics: MetricsSettings {
                    service_name_format: telemetry::settings::ServiceNameFormat::MetricPrefix,
                    report_optional: false,
                },
                memory_profiler: MemoryProfilerSettings {
                    enabled: profiler_enabled,
                    ..Default::default()
                },
                server: telemetry_server_settings,
            },
            custom_server_routes: vec![],
        })
        .expect("metrics server");