Add configurable namespace to telemetry

.changelog/unreleased/improvements/ibc-relayer/4380-telemetry-namespace.md

@@ -0,0 +1,3 @@
+- Add a new optional configuration `prefix` for telemetry, allowing users
+  to specify a prefixed namespace for Prometheus metrics.
+  ([\#4380](https://github.com/informalsystems/hermes/issues/4380))

Cargo.toml

@@ -85,8 +85,9 @@ num-bigint               = "0.4"
 num-rational             = "0.4.1"
 once_cell                = "1.20.2"
 oneline-eyre             = "0.1"
-opentelemetry            = "0.19.0"
-opentelemetry-prometheus = "0.12.0"
+opentelemetry            = "0.20.0"
+opentelemetry_sdk        = { version = "0.20.0", features = ["metrics"] }
+opentelemetry-prometheus = "0.13.0"
 primitive-types          = { version = "0.12.1", default-features = false }
 prometheus               = "0.13.4"
 prost                    = "0.13"

config.toml

@@ -128,6 +128,18 @@ host = '127.0.0.1'
 # by the telemetry service. Default: 3001
 port = 3001
 
+# Specify the prefix used for Prometheus metrics. This configuration is optional, if
+# it is missing the result is equivalent to setting it to empty string: `prefix = ""`
+# For example setting `prefix = "hermes"` will result in:
+#   * `hermes_acknowledgement_events_total`
+#   * `hermes_wallet_balance`
+#   * etc...
+# The default is no prefix, which results in:
+#   * `acknowledgement_events_total`
+#   * `wallet_balance`
+#   * etc...
+# prefix = ""
+
 [telemetry.buckets]
 # Specify the range of the 10 histogram buckets in ms for the `tx_latency_submitted` metric.
 # Default: { start = 500, end = 10000, buckets = 10 }

crates/relayer-cli/src/commands/start.rs

@@ -178,6 +178,7 @@ fn spawn_telemetry_server(config: &Config) {
         config.telemetry.buckets.latency_submitted.buckets,
         config.telemetry.buckets.latency_confirmed.range.clone(),
         config.telemetry.buckets.latency_confirmed.buckets,
+        config.telemetry.prefix.as_str(),
     );
     let telemetry = config.telemetry.clone();
 

crates/relayer/src/config.rs

@@ -498,6 +498,8 @@ pub struct TelemetryConfig {
     pub port: u16,
     #[serde(default = "HistogramBuckets::default")]
     pub buckets: HistogramBuckets,
+    #[serde(default)]
+    pub prefix: String,
 }
 
 #[derive(Clone, Debug, Deserialize, Serialize)]
@@ -562,6 +564,7 @@ impl Default for TelemetryConfig {
             host: "127.0.0.1".to_string(),
             port: 3001,
             buckets: HistogramBuckets::default(),
+            prefix: Default::default(),
         }
     }
 }

crates/telemetry/Cargo.toml

@@ -19,7 +19,8 @@ axum                     = { workspace = true }
 dashmap                  = { workspace = true }
 moka                     = { workspace = true, features = ["sync"] }
 once_cell                = { workspace = true }
-opentelemetry            = { workspace = true, features = ["metrics"] }
+opentelemetry            = { workspace = true }
+opentelemetry_sdk        = { workspace = true, features = ["metrics"] }
 opentelemetry-prometheus = { workspace = true }
 prometheus               = { workspace = true }
 serde                    = { workspace = true }

crates/telemetry/src/lib.rs

@@ -20,12 +20,14 @@ pub fn new_state(
     tx_latency_submitted_buckets: u64,
     tx_latency_confirmed_range: Range<u64>,
     tx_latency_confirmed_buckets: u64,
+    namespace: &str,
 ) -> Arc<TelemetryState> {
     Arc::new(TelemetryState::new(
         tx_latency_submitted_range,
         tx_latency_submitted_buckets,
         tx_latency_confirmed_range,
         tx_latency_confirmed_buckets,
+        namespace,
     ))
 }
 
@@ -36,12 +38,14 @@ pub fn init(
     tx_latency_submitted_buckets: u64,
     tx_latency_confirmed_range: Range<u64>,
     tx_latency_confirmed_buckets: u64,
+    namespace: &str,
 ) -> &'static Arc<TelemetryState> {
     let new_state = new_state(
         tx_latency_submitted_range,
         tx_latency_submitted_buckets,
         tx_latency_confirmed_range,
         tx_latency_confirmed_buckets,
+        namespace,
     );
     match GLOBAL_STATE.set(new_state) {
         Ok(_) => debug!("initialised telemetry global state"),
@@ -68,6 +72,7 @@ pub fn global() -> &'static Arc<TelemetryState> {
                     end: 20000,
                 },
                 10,
+                "",
             )
         }
     }