new tracing: package bumps, changelogs, docs

Author: mgmeier

cardano-node/cardano-node.cabal

@@ -211,8 +211,8 @@ library
                       , sop-extras
                       , text >= 2.0
                       , time
-                      , trace-dispatcher ^>= 2.10.0
-                      , trace-forward ^>= 2.3.0
+                      , trace-dispatcher ^>= 2.11.0
+                      , trace-forward ^>= 2.4.0
                       , trace-resources ^>= 0.2.4
                       , tracer-transformers
                       , transformers

cardano-tracer/CHANGELOG.md

@@ -1,5 +1,14 @@
 # ChangeLog
 
+## 0.3.6 (November 2025)
+* Implement Prometheus HTTP service discovery (SD) under the URL `/targets`
+* Add optional config field `"prometheusLabels": { "<labelname>": "<labelvalue>", ... }` for custom labels to be attached with Prometheus SD
+* Use `TracerTrace.forMachine` directly instead of going via derived `TracerTrace.toJSON`; remove unused `TracerTrace` JSON instances
+* Use proper 'camelCase' for machine-readable `TracerTrace`
+* Proper tracing (vs. dumping to stdout) for `showProblemIfAny` and for forwarding connection interruptions
+* Remove redundant `runInLoop` in favour of `trace-dispatcher`'s implementation
+* Split up journal handler implementation into internal modules `Systemd` and `NoSystemd` (maintenance)
+
 ## 0.3.5 (October, 2025)
 * Updated to `ekg-forward-1.0`, `ouroboros-network-0.22.3`, `ouroboros-network-api-0.16` and `ouroboros-network-0.22.3`.
 * Updated metric names

cardano-tracer/cardano-tracer.cabal

@@ -1,7 +1,7 @@
 cabal-version: 3.0
 
 name:                   cardano-tracer
-version:                0.3.5
+version:                0.3.6
 synopsis:               A service for logging and monitoring over Cardano nodes
 description:            A service for logging and monitoring over Cardano nodes.
 category:               Cardano,
@@ -200,8 +200,8 @@ library
                       , string-qq
                       , text
                       , time
-                      , trace-dispatcher ^>= 2.10.0
-                      , trace-forward ^>= 2.3.0
+                      , trace-dispatcher ^>= 2.11.0
+                      , trace-forward ^>= 2.4.0
                       , trace-resources ^>= 0.2.4
                       , wai ^>= 3.2
                       , warp ^>= 3.4

cardano-tracer/configuration/metrics_help.json

@@ -35,6 +35,7 @@
   "forgedSlotLast": "Slot number of the last forged block",
   "forging_enabled": "A node without forger credentials or started as non-producing has forging disabled.",
   "forks": "counter for forks",
+  "GSM.state": "The state of the Genesis State Machine. 0 = PreSyncing, 1 = Syncing, 2 = CaughtUp.",
   "haskell_compiler_major": "Cardano compiler version information",
   "haskell_compiler_minor": "Cardano compiler version information",
   "haskell_compiler_patch": "Cardano compiler version information",

cardano-tracer/docs/cardano-tracer.md

@@ -19,6 +19,7 @@
   - [Logging](#logging)
   - [Logs Rotation](#logs-rotation)
   - [Prometheus](#prometheus)
+    - [Prometheus HTTP service discovery](#prometheus-http-sd)
   - [EKG Monitoring](#ekg-monitoring)
   - [Verbosity](#verbosity)
   - [RTView](#rtview)
@@ -462,7 +463,28 @@ or
 }
 ```
 
+### Prometheus HTTP service discovery
 
+The `/targets` path can be used for Prometheus HTTP service discovery. This lets
+Prometheus dynamically discover all connected nodes, and scrape their metrics.
+Below is a minimal example of a corresponding job definition that goes into the
+`prometheus.yml` configuration:
+
+```yaml
+  - job_name: "cardano-tracer"
+    http_sd_configs:
+      - url: 'http://127.0.0.1:3200/targets'    # <-- Your cardano-tracer's real hostname:prometheus port
+```
+
+Each target will have a label `node_name` which corresponds to the `TraceOptionNodeName` setting in the respective node config.
+
+In `cardano-tracer`'s config, you can optionally provide additional labels to be attached to *all* targets
+(default is no additional labels):
+```json
+  "prometheusLabels": {
+      "<labelname>": "<labelvalue>", ...
+    }
+```
 
 ## EKG Monitoring
 

trace-dispatcher/CHANGELOG.md

@@ -1,8 +1,14 @@
 # Revision history for trace-dispatcher
 
-## NEXT -- Nov 2025
-
-* Replaced both `tofDisconnQueueSize` and `tofConnQueueSize` with `tofQueueSize` (See #6361 for details). 
+## 2.11.0 -- Nov 2025
+
+* `class LogFormatting`: remove redundant `forHumanFromMachine` and `forHumanOrMachine` (the system already does that inherently)
+* Introduce type `Cardano.Logging.Types.TraceMessage.TraceMessage` with explicit codecs for JSON and CBOR
+* Rework `PreFormatted` type and formatters to use `TraceMessage`; slightly optimize `humanFormatter'`
+* Add CBOR formatting via `FormattedMessage.FormattedCBOR` constructor and a `cborFormatter'` function
+* Replaced both `disconnectedQueueSize` and `connectedQueueSize` with `queueSize` in `TraceOptionForwarder` while keeping config parsing backwards compatible
+* Add retry delay reset in `runInLoop` when the action runs sufficiently long
+* Safely stop `standardTracer`'s stdout thread when there are no more producers
 
 ## 2.10.0 -- July, 2025
 * Forwarding protocol supports connections over TCP socket, in addition to Unix domain sockets.

trace-dispatcher/doc/trace-dispatcher.md

@@ -186,11 +186,11 @@ For the effective integration of trace messages into the tracing system, two ess
 
 The `LogFormatting` typeclass governs the presentation of trace messages, encompassing the mapping of traces to metrics and messages. It includes the following methods:
 
-- The `forMachine` method caters to a machine-readable representation, adaptable based on the detail level. Implementation is mandatory for the trace author.
+- The `forMachine` method caters to a machine-readable representation, adaptable based on the detail level. Implementation is mandatory for the trace author. The system will render this,
+along with trace metadata, as JSON of type `Cardano.Logging.Types.TraceMessage.TraceMessage`.
 
 - The `forHuman` method renders the message in a human-readable form. Its default implementation is an
-empty text, which will be replaced by forMachine in the rendering, if forHuman is empty. The forMachine
-will by default rendered with a DNormal detsil level, if no other information is provided.
+empty text. Whenever the system encounters the empty text, it will replace it with the machine-readable JSON, rendering it as a value in `{"data": <value>}`, preventing potential loss of log information
 
 - The `asMetrics` method portrays the message as 0 to n metrics. The default implementation assumes no metrics. Each metric can optionally specify a hierarchical identifier as a `[Text]`.
 
@@ -700,30 +700,30 @@ logs and forwarded tracing output.
 
 As mentioned earlier, trace backends serve as the final destinations for all traces once they have undergone trace interpretation, resulting in metrics and messages. The system defines three trace backends:
 
-1. __Standard Tracer:__ This is the fundamental standard output tracer. Notably, it can accept both regular and confidential traces. It's important to construct only one standard tracer in any application, as attempting to create a new one will result in an exception.
+1. __Standard Tracer:__ This is the fundamental standard output tracer. Notably, it can accept both regular and confidential traces.
 
     ```haskell
     standardTracer :: forall m. (MonadIO m)
       => m (Trace m FormattedMessage)
     ```
 
-2. __Trace-Forward Tracer:__ This is a network-only sink dedicated to forwarding messages using typed protocols over TCP or local sockets. It exclusively handles public traces and should be instantiated only once per application.
+2. __Trace-Forward Tracer:__ This is a network-only sink dedicated to forwarding messages using typed protocols over TCP or local sockets. It exclusively handles public traces.
 
     ```haskell
     forwardTracer :: forall m. (MonadIO m)
       => ForwardSink TraceObject
       -> Trace m FormattedMessage
     ```
 
-3. __EKG Tracer:__ This tracer submits metrics to a local EKG store, which then further forwards the messages.
+3. __EKG Tracer:__ This tracer submits metrics to a local EKG store (which then can be exposed directly via the `PrometheusSimple` backend and/or forwarded).
 
     ```haskell
     ekgTracer :: MonadIO m
       => Metrics.Store
       -> m (Trace m FormattedMessage)
     ```
 
-It's imperative to note that constructing more than one instance of each tracer in an application may lead to exceptions and should be avoided.
+It's imperative to note that constructing more than one instance of each tracer in an application should absolutely be avoided, as it may result in unexpected behaviour.
 
 ## Data Points Overview and Deprecation Notice
 

trace-dispatcher/trace-dispatcher.cabal

@@ -1,7 +1,7 @@
 cabal-version: 3.0
 
 name:                   trace-dispatcher
-version:                2.10.0
+version:                2.11.0
 synopsis:               Tracers for Cardano
 description:            Package for development of simple and efficient tracers
                         based on the arrow based contra-tracer package

trace-forward/CHANGELOG.md

@@ -1,8 +1,10 @@
 # ChangeLog
 
-## NEXT - Nov 2025
+## 2.4.0 - Nov 2025
 
-* Replaced both `disconnectedQueueSize` and `connectedQueueSize` with `queueSize` (See #6361 for details).
+* Refactor `writeToSink` and `readFromSink` to simplify STM usage
+* Drop incongruous logic of switching queue capacity from `ForwardSink`, along with related fields and code
+* Provide `InitForwardingConfig` config record type for `initForwarding` and `initForwardingDelayed`
 
 ## 2.3.1 - Oct 2025
 

trace-forward/trace-forward.cabal

@@ -1,7 +1,7 @@
 cabal-version: 3.0
 
 name:                   trace-forward
-version:                2.3.1
+version:                2.4.0
 synopsis:               The forwarding protocols library for cardano node.
 description:            The library providing typed protocols for forwarding different
                         information from the cardano node to an external application.