remove prometheus metrics

Cargo.toml

@@ -39,7 +39,6 @@ thiserror = "2.0.3"
 comrak = { version = "0.48.0", default-features = false }
 syntect = { version = "5.0.0", default-features = false, features = ["parsing", "html", "dump-load", "regex-onig"] }
 toml = "0.9.2"
-prometheus = { version = "0.14.0", default-features = false }
 opentelemetry = "0.31.0"
 opentelemetry-otlp = { version = "0.31.0", features = ["grpc-tonic",  "metrics"] }
 opentelemetry-resource-detectors = "0.10.0"

docker-compose.yml

@@ -21,7 +21,6 @@
 #     - release-rebuild-enqueuer
 #
 # optional profile: `metrics`:
-# * `prometheus` -> configured prometheus instance
 # * `opentelemetry` -> a debug opentelemetry receiver
 #
 # optional profile: `full`: all of the above.
@@ -44,12 +43,6 @@
 #   * docker-native for DB, S3, rustwide workspace, crates.io index
 #   * mounts for prefix
 #
-# * prometheus scrape config is set to collect from the web server, the
-#   registry watcher, and the build servers. Scraping is not dynamic, so
-#   the local prometheus server will try to fetch from all service
-#   instances (web, watcher, builder), and just error in case the specific
-#   server isn't accessible.
-
 x-docker-cache: &docker-cache
     # shared configuration to cache docker layers across CI runs.
     # can just always be in in the `docker-compose.yml`, and will
@@ -119,9 +112,6 @@ x-builder: &builder
         <<: *builder-environment
     env_file:
         - .docker.env
-    healthcheck:
-        <<: *healthcheck-interval
-        test: curl --silent --fail localhost:3000/about/metrics
 
 x-registry-watcher: &registry-watcher
     build:
@@ -144,9 +134,6 @@ x-registry-watcher: &registry-watcher
         DOCSRS_MAX_QUEUED_REBUILDS: ${DOCSRS_MAX_QUEUED_REBUILDS:-10}
     env_file:
         - .docker.env
-    healthcheck:
-        <<: *healthcheck-interval
-        test: curl --silent --fail localhost:3000/about/metrics
 
 services:
     web:
@@ -166,7 +153,7 @@ services:
             - .docker.env
         healthcheck:
             <<: *healthcheck-interval
-            test: curl --silent --fail localhost:3000/about/metrics
+            test: curl --silent --fail localhost:3000/
         profiles:
             - web
             - full
@@ -275,28 +262,6 @@ services:
             <<: *healthcheck-interval
             test: mc ready local
 
-    prometheus:
-        build:
-            context: ./dockerfiles
-            dockerfile: ./Dockerfile-prometheus
-            <<: *docker-cache
-        ports:
-            - "127.0.0.1:9090:9090"
-        # we intentionally don't define depends_on here.
-        # While the scrapers are configured to fetch from eventually running
-        # web or build-servers, adding these as dependency would mean we can't
-        # test metrics just with a webserver.
-        # Prometheus will just scrape from the working endpoints, and skip/error
-        # on the broken ones.
-        healthcheck:
-            <<: *healthcheck-interval
-            test: promtool check healthy
-        profiles:
-            # we rarely need to test with actual prometheus, so always running
-            # it is a waste.
-            - metrics
-            - full
-
     opentelemetry:
         build:
             context: ./dockerfiles
@@ -309,8 +274,8 @@ services:
             test: curl --silent --fail http://localhost:13133/health
 
         profiles:
-            # we rarely need to test with actual prometheus, so always running
-            # it is a waste.
+            # we rarely need to test with actual opentelemetry, so always
+            # running it is a waste.
             - metrics
             - full
 

justfiles/services.just

@@ -20,7 +20,7 @@ compose-up-builder:
 compose-up-watcher: 
   just compose-up watcher
 
-# Launch prometheus server in the background 
+# Launch metrics collector in the background 
 [group('compose')]
 compose-up-metrics: 
   just compose-up metrics

src/bin/cratesfyi.rs

@@ -4,7 +4,7 @@ use clap::{Parser, Subcommand, ValueEnum};
 use docs_rs::{
     Config, Context, Index, PackageKind, RustwideBuilder,
     db::{self, CrateId, Overrides, add_path_into_database, types::version::Version},
-    queue_rebuilds_faulty_rustdoc, start_background_metrics_webserver, start_web_server,
+    queue_rebuilds_faulty_rustdoc, start_web_server,
     utils::{
         ConfigName, daemon::start_background_service_metric_collector, get_config,
         get_crate_pattern_and_priority, list_crate_priorities, queue_builder,
@@ -137,8 +137,6 @@ enum CommandLine {
     },
 
     StartRegistryWatcher {
-        #[arg(name = "SOCKET_ADDR", default_value = "0.0.0.0:3000")]
-        metric_server_socket_addr: SocketAddr,
         /// Enable or disable the repository stats updater
         #[arg(
             long = "repository-stats-updater",
@@ -152,10 +150,7 @@ enum CommandLine {
         queue_rebuilds: Toggle,
     },
 
-    StartBuildServer {
-        #[arg(name = "SOCKET_ADDR", default_value = "0.0.0.0:3000")]
-        metric_server_socket_addr: SocketAddr,
-    },
+    StartBuildServer,
 
     /// Starts the daemon
     Daemon {
@@ -186,7 +181,6 @@ impl CommandLine {
         match self {
             Self::Build { subcommand } => subcommand.handle_args(ctx)?,
             Self::StartRegistryWatcher {
-                metric_server_socket_addr,
                 repository_stats_updater,
                 cdn_invalidator,
                 queue_rebuilds,
@@ -205,18 +199,12 @@ impl CommandLine {
                 // metrics from the registry watcher, which should only run once, and all the time.
                 start_background_service_metric_collector(&ctx)?;
 
-                start_background_metrics_webserver(Some(metric_server_socket_addr), &ctx)?;
-
                 ctx.runtime.block_on(docs_rs::utils::watch_registry(
                     &ctx.async_build_queue,
                     &ctx.config,
                 ))?;
             }
-            Self::StartBuildServer {
-                metric_server_socket_addr,
-            } => {
-                start_background_metrics_webserver(Some(metric_server_socket_addr), &ctx)?;
-
+            Self::StartBuildServer => {
                 queue_builder(&ctx, RustwideBuilder::init(&ctx)?)?;
             }
             Self::StartWebServer { socket_addr } => {

src/build_queue.rs

@@ -1,5 +1,5 @@
 use crate::{
-    BuildPackageSummary, Config, Context, Index, InstanceMetrics, RustwideBuilder,
+    BuildPackageSummary, Config, Context, Index, RustwideBuilder,
     cdn::{self, CdnMetrics},
     db::{
         CrateId, Pool, delete_crate, delete_version,
@@ -70,7 +70,6 @@ pub struct AsyncBuildQueue {
     config: Arc<Config>,
     storage: Arc<AsyncStorage>,
     pub(crate) db: Pool,
-    metrics: Arc<InstanceMetrics>,
     queue_metrics: BuildQueueMetrics,
     builder_metrics: Arc<BuilderMetrics>,
     cdn_metrics: Arc<CdnMetrics>,
@@ -80,7 +79,6 @@ pub struct AsyncBuildQueue {
 impl AsyncBuildQueue {
     pub fn new(
         db: Pool,
-        metrics: Arc<InstanceMetrics>,
         config: Arc<Config>,
         storage: Arc<AsyncStorage>,
         cdn_metrics: Arc<CdnMetrics>,
@@ -90,7 +88,6 @@ impl AsyncBuildQueue {
             max_attempts: config.build_attempts.into(),
             config,
             db,
-            metrics,
             storage,
             queue_metrics: BuildQueueMetrics::new(otel_meter_provider),
             builder_metrics: Arc::new(BuilderMetrics::new(otel_meter_provider)),
@@ -427,7 +424,6 @@ impl AsyncBuildQueue {
                             "{}-{} added into build queue",
                             release.name, release.version
                         );
-                        self.metrics.queued_builds.inc();
                         self.queue_metrics.queued_builds.add(1, &[]);
                         crates_added += 1;
                     }
@@ -637,12 +633,10 @@ impl BuildQueue {
             let instant = Instant::now();
             let res = f(&to_process);
             let elapsed = instant.elapsed().as_secs_f64();
-            self.inner.metrics.build_time.observe(elapsed);
             self.inner.builder_metrics.build_time.record(elapsed, &[]);
             res
         };
 
-        self.inner.metrics.total_builds.inc();
         self.inner.builder_metrics.total_builds.add(1, &[]);
 
         self.runtime.block_on(
@@ -665,7 +659,6 @@ impl BuildQueue {
             )?;
 
             if attempt >= self.inner.max_attempts {
-                self.inner.metrics.failed_builds.inc();
                 self.inner.builder_metrics.failed_builds.add(1, &[]);
             }
             Ok(())
@@ -1384,12 +1377,6 @@ mod tests {
         })?;
         assert!(!called, "there were still items in the queue");
 
-        // Ensure metrics were recorded correctly
-        let metrics = env.instance_metrics();
-        assert_eq!(metrics.total_builds.get(), 9);
-        assert_eq!(metrics.failed_builds.get(), 1);
-        assert_eq!(metrics.build_time.get_sample_count(), 9);
-
         let collected_metrics = env.collected_metrics();
 
         assert_eq!(