🎨 Improve: Observability performance

services/jaeger/docker-compose.yml.j2

@@ -33,7 +33,7 @@ services:
           memory: 64M
           cpus: "0.1"
   otel-collector:
-    image: otel/opentelemetry-collector-contrib:0.123.0
+    image: otel/opentelemetry-collector-contrib:0.135.0
     configs:
       - source: opentelemetry-collector-config
         target: /etc/otel/config.yaml

services/jaeger/opentelemetry-collector-config.yaml

@@ -31,6 +31,11 @@ processors:
     error_mode: ignore
     traces:
       span:
+        - attributes["http.route"] == "/health"
+        - attributes["http.target"] == "/socket.io/"
+        - attributes["url.path"] == "/socket.io/"
+        - attributes["server.address"] == "monitoring.osparc-master.speag.com"
+        - attributes["net.peer.name"] == "/var/run/docker.sock"
         - attributes["http.route"] == "healthcheck_readiness_probe"
         - attributes["http.route"] == "healthcheck_liveness_probe"
         - attributes["http.target"] == "/metrics" and IsMatch(attributes["http.user_agent"], ".*Prometheus.*") == true

services/logging/docker-compose.yml.j2

@@ -138,7 +138,7 @@ services:
       start_period: 40s
 
   loki:
-    image: grafana/loki:3.5.0
+    image: grafana/loki:3.5.4
     configs:
       - source: loki_config
         target: /etc/loki/loki.yaml
@@ -160,8 +160,8 @@ services:
         delay: 5s
       resources:
         limits:
-          cpus: '1.0'
-          memory: 2G
+          cpus: '2.0'
+          memory: 4G
         reservations:
           cpus: '0.5'
           memory: 1G

services/logging/loki.yaml

@@ -1,8 +1,10 @@
 auth_enabled: false
 
 server:
+  log_level: info
   http_listen_port: 3100
-
+  grpc_server_max_recv_msg_size: 16777216  # 16MB (increase from 4MB default)
+  grpc_server_max_send_msg_size: 16777216  # 16MB
 common:
   path_prefix: /tmp/loki  # Required for internal directories
 
@@ -29,16 +31,30 @@ storage_config:
     active_index_directory: /tmp/loki/tsdb-index   # Local cache for index metadata
     cache_location: /tmp/loki/tsdb-cache          # Local cache for tsdb
   aws:
-    s3: s3://${S3_BUCKET_NAME_LOKI}
+    bucketnames: ${S3_BUCKET_NAME_LOKI}
     region: ${S3_REGION_LOKI}
     access_key_id: ${S3_ACCESS_KEY_LOKI}
     secret_access_key: ${S3_SECRET_KEY_LOKI}
-    s3forcepathstyle: ${S3_FORCE_PATH_STYLE_LOKI}     # Set to true if using MinIO or S3-compatible API; optional for AWS
-    endpoint: ${S3_ENDPOINT_LOKI} # Optional; use for non-default endpoints
+    endpoint: ${S3_ENDPOINT_LOKI}
+    s3forcepathstyle: true
 
 compactor:
   working_directory: /tmp/loki/compactor
   retention_enabled: false
 
 limits_config:
+  reject_old_samples: true
+  reject_old_samples_max_age: 4h
+  max_cache_freshness_per_query: 10m
+  split_queries_by_interval: 15m
+  # for big logs tune
+  per_stream_rate_limit: 512M
+  per_stream_rate_limit_burst: 1024M
+  cardinality_limit: 200000
+  ingestion_burst_size_mb: 1000
+  ingestion_rate_mb: 10000
+  max_entries_limit_per_query: 1000000
+  max_label_value_length: 20480
+  max_label_name_length: 10240
+  max_label_names_per_series: 300
   retention_period: ${LOKI_RETENTION_PERIOD}  # must be >= 24h and multiple of index period (24h)

services/monitoring/docker-compose.yml.j2

@@ -372,7 +372,7 @@ services:
           memory: 32M
           cpus: "0.1"
   tempo:
-    image: grafana/tempo:2.7.2
+    image: grafana/tempo:2.8.2
     command: "-target=scalable-single-binary -config.file=/etc/tempo.yaml"
     configs:
       - source: tempo_config
@@ -393,8 +393,8 @@ services:
         - traefik.http.routers.tempo.middlewares=ops_whitelist_ips@swarm, ops_gzip@swarm, tempo_replace_regex
       resources:
         limits:
-          memory: 2000M
-          cpus: "2.0"
+          memory: 4G
+          cpus: "5.0"
 
 configs:
   alertmanager_config:

services/monitoring/tempo_config.yaml.j2

@@ -12,9 +12,9 @@ distributor:
   log_discarded_spans:
     enabled: true
     include_all_attributes: false
-#ingester:
-#  max_block_duration: 5m               # cut the headblock when this much time passes. this should probably be left alone normally
-
+ingester:
+  max_block_duration: 5m               # cut the headblock when this much time passes. this should probably be left alone normally
+  trace_idle_period: 301s
 compactor:
   compaction:
     block_retention: 96h             # overall Tempo trace retention.
@@ -59,7 +59,7 @@ storage:
 querier:
   frontend_worker:
     frontend_address: 0.0.0.0:9095
-
+stream_over_http_enabled: true
 overrides:
   defaults:
     metrics_generator: