Merge tag 'v3.12.0' into merge_v3.11.0

CHANGELOG.md

@@ -1,5 +1,17 @@
 ## [Unreleased]
 
+## [3.12.0] - 2024-03-28
+
+## Added
+- MacOS cpu usage sampling. (#203)
+- Metric unit annotations are added and exposed as metadata.
+- Logs version number on startup. (#213)
+
+## Fixed
+
+- Incorrect summary stats (percentiles) were reported in version 3.10.2, 3.10.3,
+  and 3.11.0. (#216)
+
 ## [3.11.0] - 2024-03-25
 
 ## Changed
@@ -185,8 +197,9 @@
 - Rewritten implementation of Rezolus using libbpf-rs and perf-event2 to provide
   a more modern approach to BPF and Perf Event instrumentation. 
 
-[unreleased]: https://github.com/iopsystems/rezolus/compare/v3.11.0...HEAD
-[3.10.3]: https://github.com/iopsystems/rezolus/compare/v3.10.3...v3.11.0
+[unreleased]: https://github.com/iopsystems/rezolus/compare/v3.12.0...HEAD
+[3.12.0]: https://github.com/iopsystems/rezolus/compare/v3.11.0...v3.12.0
+[3.11.0]: https://github.com/iopsystems/rezolus/compare/v3.10.3...v3.11.0
 [3.10.3]: https://github.com/iopsystems/rezolus/compare/v3.10.2...v3.10.3
 [3.10.2]: https://github.com/iopsystems/rezolus/compare/v3.10.1...v3.10.2
 [3.10.1]: https://github.com/iopsystems/rezolus/compare/v3.10.0...v3.10.1

Cargo.toml

@@ -7,7 +7,7 @@ edition = "2021"
 description = "High resolution systems performance telemetry agent"
 
 [workspace.package]
-version = "3.11.0"
+version = "3.12.0"
 license = "MIT OR Apache-2.0"
 
 [dependencies]
@@ -22,6 +22,7 @@ linkme = "0.3.15"
 metriken =  "0.5.1"
 metriken-exposition = { version = "0.2.0", features = ["serde", "msgpack"] }
 memmap2 = "0.5.10"
+num_cpus = "1.16.0"
 once_cell = "1.18.0"
 ouroboros = "0.17.2"
 ringlog = "0.6.0"

src/main.rs

@@ -10,6 +10,8 @@ use std::time::SystemTime;
 use std::time::{Duration, Instant};
 use tokio::sync::RwLock;
 
+const VERSION: &str = env!("CARGO_PKG_VERSION");
+
 type HistogramSnapshots = HashMap<String, HistogramSnapshot>;
 
 static SNAPSHOTS: Lazy<Arc<RwLock<Snapshots>>> =
@@ -50,8 +52,9 @@ impl Snapshots {
 
         for (name, previous) in &self.previous {
             if let Some(histogram) = current.get(name).cloned() {
-                let _ = histogram.wrapping_sub(previous);
-                delta.insert(name.to_string(), histogram);
+                if let Ok(d) = histogram.wrapping_sub(previous) {
+                    delta.insert(name.to_string(), d);
+                }
             }
         }
 
@@ -165,7 +168,7 @@ fn main() {
         }
     });
 
-    info!("rezolus");
+    info!("rezolus {VERSION}");
 
     // spawn http exposition thread
     rt.spawn(exposition::http(config.clone()));

src/samplers/block_io/stats.rs

@@ -1,96 +1,148 @@
+use crate::common::HISTOGRAM_GROUPING_POWER;
 use crate::*;
 use metriken::*;
 
-bpfhistogram!(
-    BLOCKIO_LATENCY,
-    "blockio/latency",
-    "distribution of block IO latencies"
-);
-bpfhistogram!(
-    BLOCKIO_SIZE,
-    "blockio/size",
-    "distribution of block IO sizes"
-);
+#[metric(
+    name = "blockio/latency",
+    description = "Distribution of blockio operation latency in nanoseconds",
+    metadata = { unit = "nanoseconds" }
+)]
+pub static BLOCKIO_LATENCY: RwLockHistogram = RwLockHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
+
+#[metric(
+    name = "blockio/size",
+    description = "Distribution of blockio operation sizes in bytes",
+    metadata = { unit = "bytes" }
+)]
+pub static BLOCKIO_SIZE: RwLockHistogram = RwLockHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/operations",
     description = "The number of completed read operations for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "read" }
+    metadata = { op = "read", unit = "operations" }
 )]
 pub static BLOCKIO_READ_OPS: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_READ_OPS_HISTOGRAM, "blockio/read/operations");
+#[metric(
+    name = "blockio/read/operations",
+    description = "Distribution of blockio read operation rates from sample to sample",
+    metadata = { unit = "operations/second" }
+)]
+pub static BLOCKIO_READ_OPS_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/operations",
     description = "The number of completed write operations for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "write" }
+    metadata = { op = "write", unit = "operations" }
 )]
 pub static BLOCKIO_WRITE_OPS: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_WRITE_OPS_HISTOGRAM, "blockio/write/operations");
+#[metric(
+    name = "blockio/write/operations",
+    description = "Distribution of blockio write operation rates from sample to sample",
+    metadata = { unit = "operations/second" }
+)]
+pub static BLOCKIO_WRITE_OPS_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/operations",
     description = "The number of completed discard operations for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "discard" }
+    metadata = { op = "discard", unit = "operations" }
 )]
 pub static BLOCKIO_DISCARD_OPS: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_DISCARD_OPS_HISTOGRAM, "blockio/discard/operations");
+#[metric(
+    name = "blockio/discard/operations",
+    description = "Distribution of blockio discard operation rates from sample to sample",
+    metadata = { unit = "operations/second" }
+)]
+pub static BLOCKIO_DISCARD_OPS_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/operations",
     description = "The number of completed flush operations for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "flush" }
+    metadata = { op = "flush", unit = "operations" }
 )]
 pub static BLOCKIO_FLUSH_OPS: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_FLUSH_OPS_HISTOGRAM, "blockio/flush/operations");
+#[metric(
+    name = "blockio/flush/operations",
+    description = "Distribution of blockio flush operation rates from sample to sample",
+    metadata = { unit = "operations/second" }
+)]
+pub static BLOCKIO_FLUSH_OPS_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/bytes",
     description = "The number of bytes read for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "read" }
+    metadata = { op = "read", unit = "bytes" }
 )]
 pub static BLOCKIO_READ_BYTES: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_READ_BYTES_HISTOGRAM, "blockio/read/bytes");
+#[metric(
+    name = "blockio/read/bytes",
+    description = "Distribution of blockio read throughput from sample to sample",
+    metadata = { unit = "bytes/second" }
+)]
+pub static BLOCKIO_READ_BYTES_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/bytes",
     description = "The number of bytes written for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "write" }
+    metadata = { op = "write", unit = "bytes" }
 )]
 pub static BLOCKIO_WRITE_BYTES: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_WRITE_BYTES_HISTOGRAM, "blockio/write/bytes");
+#[metric(
+    name = "blockio/write/bytes",
+    description = "Distribution of blockio write throughput from sample to sample",
+    metadata = { unit = "bytes/second" }
+)]
+pub static BLOCKIO_WRITE_BYTES_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/bytes",
     description = "The number of bytes discarded for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "discard" }
+    metadata = { op = "discard", unit = "bytes" }
 )]
 pub static BLOCKIO_DISCARD_BYTES: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_DISCARD_BYTES_HISTOGRAM, "blockio/discard/bytes");
+#[metric(
+    name = "blockio/discard/bytes",
+    description = "Distribution of blockio discard throughput from sample to sample",
+    metadata = { unit = "bytes/second" }
+)]
+pub static BLOCKIO_DISCARD_BYTES_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 #[metric(
     name = "blockio/bytes",
     description = "The number of bytes flushed for block devices",
     formatter = blockio_metric_formatter,
-    metadata = { op = "flush" }
+    metadata = { op = "flush", unit = "bytes" }
 )]
 pub static BLOCKIO_FLUSH_BYTES: LazyCounter = LazyCounter::new(Counter::default);
 
-histogram!(BLOCKIO_FLUSH_BYTES_HISTOGRAM, "blockio/flush/bytes");
+#[metric(
+    name = "blockio/flush/bytes",
+    description = "Distribution of blockio flush throughput from sample to sample",
+    metadata = { unit = "bytes/second" }
+)]
+pub static BLOCKIO_FLUSH_BYTES_HISTOGRAM: AtomicHistogram =
+    AtomicHistogram::new(HISTOGRAM_GROUPING_POWER, 64);
 
 /// A function to format the blockio metrics that allows for export of ops and
 /// byte counters by operation type.

src/samplers/cpu/linux/mod.rs

@@ -1,2 +1,4 @@
 mod perf;
 mod usage;
+
+pub mod stats;

src/samplers/cpu/linux/perf/mod.rs

@@ -1,5 +1,4 @@
 use crate::common::Nop;
-use crate::samplers::cpu::stats::*;
 use crate::samplers::cpu::*;
 use metriken::{DynBoxedMetric, MetricBuilder};
 use perf_event::events::x86::{Msr, MsrId};