fix(BA-2179): Unify logging configuration in all services (#5614)

Co-authored-by: octodog <[email protected]>

Author: achimnol

changes/5614.fix.md

@@ -0,0 +1 @@
+Fix regression of agent's per-package log-level configurations

configs/agent/sample.toml

@@ -151,14 +151,24 @@
   ## sample-rate = 10
 
 [logging]
-  # Log level.
+  # The version used by logging.dictConfig().
+  version = 1
+  # The main log level to filter messages from all loggers.
   level = "INFO"
   # Array of log drivers to print.
   drivers = [ "console",]
+  # Disable the existing loggers when applying the config.
+  disable-existing-loggers = false
   ## file = "{ FileConfig }"
   ## logstash = "{ LogstashConfig }"
   ## graylog = "{ GraylogConfig }"
 
+  # The mapping of log handler configurations.
+  [logging.handlers]
+
+  # The mapping of per-namespace logger configurations.
+  [logging.loggers]
+
   [logging.console]
     # Opt to print colorized log.
     ## colored = true

configs/manager/sample.toml

@@ -10,6 +10,9 @@
   # Type of the database system to use.
   # Currently, only PostgreSQL is supported as the main database backend.
   type = "postgresql"
+  # Network address and port of the database server.
+  # Default is the standard PostgreSQL port on localhost.
+  addr = { host = "127.0.0.1", port = 5432 }
   # Database name to use.
   # This database must exist and be accessible by the configured user.
   # Length must be between 2 and 64 characters due to database naming constraints.
@@ -43,16 +46,6 @@
   # If connections cannot be acquired within this time, an exception is raised.
   lock-conn-timeout = 0
 
-  [db.addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
-
 [etcd]
   # Namespace prefix for etcd keys used by Backend.AI.
   # Allows multiple Backend.AI clusters to share the same etcd cluster.
@@ -94,6 +87,20 @@
   # If not specified, defaults to the GID of the server.py file.
   # Important for proper file permissions when creating files/sockets.
   ## group = "501"
+  # Network address and port where the manager service will listen.
+  # Default is all interfaces (0.0.0.0) on port 8080.
+  # For private deployments, consider using 127.0.0.1 instead.
+  service-addr = { host = "0.0.0.0", port = 8080 }
+  # Address and port to announce to other components.
+  # This is used for service discovery and should be accessible by other
+  # components.
+  announce-addr = { host = "127.0.0.1", port = 5432 }
+  # Address and port to announce for internal API requests.
+  # This is used for service discovery and should be accessible by other
+  # components.
+  announce-internal-addr = { host = "host.docker.internal", port = 18080 }
+  # Set the internal hostname/port to accept internal API requests.
+  internal-addr = { host = "0.0.0.0", port = 18080 }
   # Path to the keypair file used for RPC authentication.
   # This file contains key pairs used for secure communication between manager
   # components.
@@ -205,46 +212,11 @@
   # If specified, metrics will be available at this port.
   # Leave as None to disable public metrics exposure.
   ## public-metrics-port = 8080  # min=1 max=65535
-
-  [manager.service-addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
-
-  [manager.announce-addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
-
-  [manager.announce-internal-addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
-
-  [manager.internal-addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
+  # Whether to use the Sokovan orchestrator for session scheduling.
+  # When enabled, uses the new Sokovan orchestrator for improved scheduling
+  # performance.
+  # When disabled, falls back to the legacy scheduling system.
+  use-sokovan = true
 
   # Configuration for PostgreSQL advisory locks.
   # This is used when distributed_lock is set to pg_advisory.
@@ -270,14 +242,24 @@
   ssl-verify = true
 
 [logging]
-  # Log level.
+  # The version used by logging.dictConfig().
+  version = 1
+  # The main log level to filter messages from all loggers.
   level = "INFO"
   # Array of log drivers to print.
   drivers = [ "console",]
+  # Disable the existing loggers when applying the config.
+  disable-existing-loggers = false
   ## file = "{ FileConfig }"
   ## logstash = "{ LogstashConfig }"
   ## graylog = "{ GraylogConfig }"
 
+  # The mapping of log handler configurations.
+  [logging.handlers]
+
+  # The mapping of per-namespace logger configurations.
+  [logging.loggers]
+
   [logging.console]
     # Opt to print colorized log.
     ## colored = true
@@ -393,6 +375,11 @@
   # Controls how long operations wait before timing out.
   # If None, uses the default timeout configured in the Redis client.
   ## request_timeout =   
+  # Whether to use TLS for Redis connections.
+  use_tls = false
+  # Whether to skip TLS certificate verification.
+  # Set to True for self-signed certificates or development environments.
+  tls_skip_verify = false
   # Optional override configurations for specific Redis contexts.
   # Allows different Redis settings for different services within Backend.AI.
   # Each key represents a context name, and the value is a complete Redis
@@ -507,21 +494,13 @@
       ## TERMINATING = "..."
 
 [metric]
+  # Address for the metric collection service.
+  addr = { host = "127.0.0.1", port = 9090 }
   # Time window for metric collection.
   # Controls how often metrics are collected and reported.
   # Format is a duration string like "1h" for 1 hour.
   timewindow = "1m"
 
-  [metric.addr]
-    # Host address of the service.
-    # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-    # all interfaces.
-    host = "127.0.0.1"
-    # Port number of the service.
-    # Must be between 1 and 65535.
-    # Ports below 1024 require root/admin privileges.
-    port = 8080
-
 [volumes]
   # Default volume host for new virtual folders.
   # Format is "proxy_name:volume_name".

configs/storage-proxy/sample.toml

@@ -6,8 +6,13 @@
 #
 # Generated automatically from the StorageProxyUnifiedConfig schema.
 
-# Configuration for external registries.
-# Defines how to connect and interact with external model registries.
+# List of object storage configurations.
+# Each configuration defines how to connect to and use an object storage
+# service.
+storages = []
+# List of artifact registry configurations.
+# Each configuration defines how to connect to and use an artifact registry
+# service.
 registries = []
 
 [storage-proxy]
@@ -83,14 +88,24 @@ registries = []
   ## sample-rate = 10
 
 [logging]
-  # Log level.
+  # The version used by logging.dictConfig().
+  version = 1
+  # The main log level to filter messages from all loggers.
   level = "INFO"
   # Array of log drivers to print.
   drivers = [ "console",]
+  # Disable the existing loggers when applying the config.
+  disable-existing-loggers = false
   ## file = "{ FileConfig }"
   ## logstash = "{ LogstashConfig }"
   ## graylog = "{ GraylogConfig }"
 
+  # The mapping of log handler configurations.
+  [logging.handlers]
+
+  # The mapping of per-namespace logger configurations.
+  [logging.loggers]
+
   [logging.console]
     # Opt to print colorized log.
     ## colored = true
@@ -102,6 +117,9 @@ registries = []
 
 [api]
   [api.client]
+    # Network address and port where the client API service will listen.
+    # This is the address accessible by clients for file operations.
+    service-addr = { host = "127.0.0.1", port = 6021 }
     # Whether to enable SSL/TLS for client API connections.
     # Required for secure communication with clients.
     ssl-enabled = true
@@ -112,17 +130,19 @@ registries = []
     # Required if ssl_enabled is True.
     ## ssl-privkey = "/etc/ssl/private/storage-proxy.key"
 
-    [api.client.service-addr]
-      # Host address of the service.
-      # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-      # all interfaces.
-      host = "127.0.0.1"
-      # Port number of the service.
-      # Must be between 1 and 65535.
-      # Ports below 1024 require root/admin privileges.
-      port = 8080
-
   [api.manager]
+    # Network address and port where the manager API service will listen.
+    # This is the address accessible by managers for control operations.
+    service-addr = { host = "127.0.0.1", port = 6022 }
+    # Address and port to announce to managers for service discovery.
+    # Should be accessible by other manager components.
+    announce-addr = { host = "127.0.0.1", port = 6022 }
+    # Address and port to announce for internal manager API requests.
+    # Used for service discovery within container environments.
+    announce-internal-addr = { host = "host.docker.internal", port = 6023 }
+    # Internal address where manager API listens for internal requests.
+    # Used for internal communication between services.
+    internal-addr = { host = "127.0.0.1", port = 16023 }
     # Whether to enable SSL/TLS for manager API connections.
     # Required for secure communication with managers.
     ssl-enabled = true
@@ -136,46 +156,6 @@ registries = []
     # Must be shared between manager and storage-proxy instances.
     secret = "manager-secret-key"
 
-    [api.manager.service-addr]
-      # Host address of the service.
-      # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-      # all interfaces.
-      host = "127.0.0.1"
-      # Port number of the service.
-      # Must be between 1 and 65535.
-      # Ports below 1024 require root/admin privileges.
-      port = 8080
-
-    [api.manager.announce-addr]
-      # Host address of the service.
-      # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-      # all interfaces.
-      host = "127.0.0.1"
-      # Port number of the service.
-      # Must be between 1 and 65535.
-      # Ports below 1024 require root/admin privileges.
-      port = 8080
-
-    [api.manager.announce-internal-addr]
-      # Host address of the service.
-      # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-      # all interfaces.
-      host = "127.0.0.1"
-      # Port number of the service.
-      # Must be between 1 and 65535.
-      # Ports below 1024 require root/admin privileges.
-      port = 8080
-
-    [api.manager.internal-addr]
-      # Host address of the service.
-      # Can be a hostname, IP address, or special addresses like 0.0.0.0 to bind to
-      # all interfaces.
-      host = "127.0.0.1"
-      # Port number of the service.
-      # Must be between 1 and 65535.
-      # Ports below 1024 require root/admin privileges.
-      port = 8080
-
 # Volume configuration.
 # Defines available storage backends and their settings.
 [volume]

configs/webserver/sample.toml

@@ -210,6 +210,11 @@
     # Controls how long operations wait before timing out.
     # If None, uses the default timeout configured in the Redis client.
     ## request_timeout =   
+    # Whether to use TLS for Redis connections.
+    use_tls = false
+    # Whether to skip TLS certificate verification.
+    # Set to True for self-signed certificates or development environments.
+    tls_skip_verify = false
     # Optional override configurations for specific Redis contexts.
     # Allows different Redis settings for different services within Backend.AI.
     # Each key represents a context name, and the value is a complete Redis
@@ -254,15 +259,31 @@
   # OpenTelemetry endpoint.
   endpoint = "http://127.0.0.1:4317"
 
+[apollo_router]
+  # Whether to enable Apollo Router.
+  enabled = false
+  # Apollo Router endpoint.
+  endpoint = "http://127.0.0.1:4000"
+
 [logging]
-  # Log level.
+  # The version used by logging.dictConfig().
+  version = 1
+  # The main log level to filter messages from all loggers.
   level = "INFO"
   # Array of log drivers to print.
   drivers = [ "console",]
+  # Disable the existing loggers when applying the config.
+  disable-existing-loggers = false
   ## file = "{ FileConfig }"
   ## logstash = "{ LogstashConfig }"
   ## graylog = "{ GraylogConfig }"
 
+  # The mapping of log handler configurations.
+  [logging.handlers]
+
+  # The mapping of per-namespace logger configurations.
+  [logging.loggers]
+
   [logging.console]
     # Opt to print colorized log.
     ## colored = true
@@ -272,10 +293,6 @@
   # Override default log level for specific scope of package
   [logging.pkg_ns]
 
-[apollo-router]
-  # endpoint of apollo router
-  endpoint = "http://127.0.0.1:4000,http://api.example.com:4000"
-
 [debug]
   # Whether to enable debug mode.
   enabled = false