Merge pull request #89 from oramasearch/feat/fix-locks

Move to tokio RwLock. Use tokio mpsc. Add metrics

.gitignore

@@ -17,3 +17,5 @@ flamegraph.svg
 .envrc
 
 .env
+
+log.log

Cargo.toml

@@ -78,7 +78,7 @@ quick-js = "0.4.1"
 
 # Tracing
 tracing = "0.1"
-tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+tracing-subscriber = { version = "0.3", features = ["env-filter", "json"] }
 duration-string = "0.5.2"
 deno_core = "0.330.0"
 serde_v8 = "0.239.0"
@@ -120,9 +120,9 @@ path = "src/bin/oramacore.rs"
 name = "embeddings"
 path = "src/bin/embeddings/embeddings.rs"
 
-[[bench]]
-name = "phrase_match"
-harness = false
+# [[bench]]
+# name = "phrase_match"
+# harness = false
 
 [features]
 default = ["benchmarking"]

README.md

@@ -104,6 +104,20 @@ hurl --very-verbose --test --variables-file api-test.hurl.property embedding-api
 
 NB: you need to have the server running before running the tests.
 
+## Opentelemetry
+
+OramaCore integrates Grafana and Prometheus for monitoring and collenting metrics.
+Make sure your OramaCore configuration enables the prometheus exporter (`http.with_prometheus: true`).
+Change the `otel/prometheus.yml` file to match your IP configuration and run the following command:
+```
+docker run --rm --name prometheus -d -p 9090:9090 -v $(pwd)/otel/prometheus.yml:/etc/prometheus/prometheus.yml prom/prometheus
+docker run --rm --name grafana -d -p 3000:3000 grafana/grafana
+```
+
+We have a Grafana dashboard available at `otel/OramaCore Dashboard.json`. You can import it into your Grafana instance.
+
+NB: the default username and password for Grafana are `admin` and `admin`.
+
 ## License
 
 [AGPLv3](/LICENSE.md)

config.yaml

@@ -4,6 +4,10 @@ http:
     allow_cors: true
     with_prometheus: true
 
+log:
+    # Comment the following line to disable the logging
+    file_path: "./log.log"
+
 writer_side:
     output: in-memory
     config:
@@ -17,13 +21,16 @@ writer_side:
         # The default embedding model used to calculate the embeddings
         # if not specified in the collection creation
         default_embedding_model: MultilingualE5Small
+        # The maximum number of request to javascript runtime that can be stored in the queue
+        # NB: the elements are in memory, so be careful with this value
+        javascript_queue_limit: 50
 
 reader_side:
     input: in-memory
     config:
         data_dir: ./.data/reader
         # The number of the write operation after the read side will commit the changes
-        insert_batch_commit_size: 300
+        insert_batch_commit_size: 3000
 
 ai_server:
     scheme: http
@@ -37,7 +44,7 @@ ai_server:
         default_model_group: small
         dynamically_load_models: false
         execution_providers:
-            - CUDAExecutionProvider
+            # - CUDAExecutionProvider
             - CPUExecutionProvider
         total_threads: 8
     LLMs: