Add build info metric to pageserver, safekeeper and proxy (#2596)

* Test that we emit build info metric for pageserver, safekeeper and proxy with some non-zero length revision label

* Emit libmetrics_build_info on startup of pageserver, safekeeper and
proxy with label "revision" which tells the git revision.

libs/metrics/src/lib.rs

@@ -77,6 +77,16 @@ pub const DISK_WRITE_SECONDS_BUCKETS: &[f64] = &[
     0.000_050, 0.000_100, 0.000_500, 0.001, 0.003, 0.005, 0.01, 0.05, 0.1, 0.3, 0.5,
 ];
 
+pub fn set_build_info_metric(revision: &str) {
+    let metric = register_int_gauge_vec!(
+        "libmetrics_build_info",
+        "Build/version information",
+        &["revision"]
+    )
+    .expect("Failed to register build info metric");
+    metric.with_label_values(&[revision]).set(1);
+}
+
 // Records I/O stats in a "cross-platform" way.
 // Compiles both on macOS and Linux, but current macOS implementation always returns 0 as values for I/O stats.
 // An alternative is to read procfs (`/proc/[pid]/io`) which does not work under macOS at all, hence abandoned.

pageserver/src/bin/pageserver.rs

@@ -10,6 +10,8 @@ use clap::{App, Arg};
 use daemonize::Daemonize;
 
 use fail::FailScenario;
+use metrics::set_build_info_metric;
+
 use pageserver::{
     config::{defaults::*, PageServerConf},
     http, page_cache, page_service, profiling, task_mgr,
@@ -359,6 +361,8 @@ fn start_pageserver(conf: &'static PageServerConf, daemonize: bool) -> Result<()
         },
     );
 
+    set_build_info_metric(GIT_VERSION);
+
     // All started up! Now just sit and wait for shutdown signal.
     signals.handle(|signal| match signal {
         Signal::Quit => {

proxy/src/main.rs

@@ -23,6 +23,7 @@ use anyhow::{bail, Context};
 use clap::{self, Arg};
 use config::ProxyConfig;
 use futures::FutureExt;
+use metrics::set_build_info_metric;
 use std::{borrow::Cow, future::Future, net::SocketAddr};
 use tokio::{net::TcpListener, task::JoinError};
 use tracing::info;
@@ -166,6 +167,7 @@ async fn main() -> anyhow::Result<()> {
     ]
     .map(flatten_err);
 
+    set_build_info_metric(GIT_VERSION);
     // This will block until all tasks have completed.
     // Furthermore, the first one to fail will cancel the rest.
     let _: Vec<()> = futures::future::try_join_all(tasks).await?;

safekeeper/src/bin/safekeeper.rs

@@ -17,6 +17,7 @@ use toml_edit::Document;
 use tracing::*;
 use url::{ParseError, Url};
 
+use metrics::set_build_info_metric;
 use safekeeper::broker;
 use safekeeper::control_file;
 use safekeeper::defaults::{
@@ -363,6 +364,7 @@ fn start_safekeeper(mut conf: SafeKeeperConf, given_id: Option<NodeId>, init: bo
             })?,
     );
 
+    set_build_info_metric(GIT_VERSION);
     // TODO: put more thoughts into handling of failed threads
     // We probably should restart them.
 

test_runner/fixtures/neon_fixtures.py

@@ -1961,6 +1961,11 @@ class NeonProxy(PgProtocol):
     def _wait_until_ready(self):
         requests.get(f"http://{self.host}:{self.http_port}/v1/status")
 
+    def get_metrics(self) -> str:
+        request_result = requests.get(f"http://{self.host}:{self.http_port}/metrics")
+        request_result.raise_for_status()
+        return request_result.text
+
     def __enter__(self):
         return self
 

test_runner/regress/test_build_info_metric.py

@@ -0,0 +1,19 @@
+from fixtures.metrics import parse_metrics
+from fixtures.neon_fixtures import NeonEnvBuilder, NeonProxy
+
+
+def test_build_info_metric(neon_env_builder: NeonEnvBuilder, link_proxy: NeonProxy):
+    neon_env_builder.num_safekeepers = 1
+    env = neon_env_builder.init_start()
+
+    parsed_metrics = {}
+
+    parsed_metrics["pageserver"] = parse_metrics(env.pageserver.http_client().get_metrics())
+    parsed_metrics["safekeeper"] = parse_metrics(env.safekeepers[0].http_client().get_metrics_str())
+    parsed_metrics["proxy"] = parse_metrics(link_proxy.get_metrics())
+
+    for component, metrics in parsed_metrics.items():
+        sample = metrics.query_one("libmetrics_build_info")
+
+        assert "revision" in sample.labels
+        assert len(sample.labels["revision"]) > 0