This topic describes the cluster-level and node-level parameters of Log Service (LogService) by scope.
Cluster-level parameters
Parameter name |
Description |
|---|---|
| min_oblogservice_version | The minimum oblogservice version allowed for the cluster. It should be increased only during an upgrade and must not be manually rolled back. |
| node_permanent_offline_time_sec | The node is marked as permanently offline after it remains offline continuously for longer than the configured duration. |
| stream_replica_task_history_retention_sec | The retention period (in seconds) of the history of log stream replica migration or replication tasks. |
| object_store_url | The access address of the object storage service. |
| log_local_target_retention_sec | The target retention period (in seconds) of local logs of each stream. This period is guaranteed as much as possible, but may be shortened under disk pressure. |
| retained_remote_log_blocks | The number of log blocks retained on the remote object storage. |
| log_local_disk_utilization_threshold | The local disk usage threshold (in percentage). When this threshold is exceeded, the disk pressure handling logic is triggered. |
| max_syslog_file_count | The maximum number of log files (oblogservice.log) retained after rolling. A value of 0 indicates no limit. |
| enable_rebalance | Whether to enable log stream load balancing. |
| load_balance_policy | The load balancing policy. load_based indicates balancing based on node load (such as disk usage), and replica_count_based indicates balancing based on the number of replicas. |
| rebalance_interval_sec | The interval (in seconds) at which load balancing is triggered. |
| metric_sample_save_interval_sec | The interval (in seconds) for sampling and saving monitoring metrics. |
| max_utilization_threshold | The upper limit (in percentage) of disk or resource utilization of a node in log stream load balancing. Migration is triggered when the limit is exceeded. |
| min_utilization_threshold | The lower limit (in percentage) of disk or resource utilization of a node in log stream load balancing. When the utilization falls below this limit, only UnderUtilized resource events are recorded for observation, and migration is not triggered. |
| log_disk_throttling_percentage | The threshold at which progressive throttling of the log disk starts, expressed as a percentage of the total log disk capacity. Throttling starts when the space occupied by unreclaimable logs exceeds this threshold. |
| log_disk_throttling_maximum_duration | The maximum latency duration (in milliseconds) for progressive throttling of the log disk. If this parameter is set to 0, progressive throttling is disabled. |
| rebuild_log_lag_threshold | The replica log lag threshold that triggers automatic Rebuild in Log Service. |
Node-level parameters
Parameter name |
Description |
|---|---|
| cluster_id | The unique identifier of the cluster. |
| region | The identifier of the region where the node is located. |
| az | The identifier of the availability zone (AZ) where the node is located. |
| local_ip | The IP address on which the node listens for external requests. |
| rpc_port | The port on which the node's gRPC service listens. It forms the listening address together with local_ip. |
| log_disk_size | The upper limit of the disk space occupied by log data. If this parameter is set to 0, the quota is determined by log_disk_percentage. |
| io_worker_count | The number of I/O worker threads. |
| http_port | The listening port of the HTTP status/management service. It forms the listening address together with local_ip. The port must be valid. Otherwise, an error is reported when the node starts, and the node fails to start. |
| status_thread_pool_size | The size of the thread pool of the HTTP status service. |
| stack_size | The stack size of worker threads. |
| net_thread_count | The number of network I/O threads. If this parameter is set to 0, the runtime automatically determines the number. |
| enable_heap_profile | Whether to enable heap memory profiling. |
| log_disk_percentage | The percentage of the total file system space allocated to logs. This parameter takes effect only when log_disk_size is set to 0. The value must be in the range [1, 99]. A value of 0 or greater than or equal to 100 causes disk quota parsing to fail. |
| memory_limit | The upper limit of the memory that the process can use. If this parameter is set to 0, the process uses 70% of the physical memory of the system by default. The total memory used by the process must not exceed the total physical memory of the system. |
| log_level | The log output level. |
| max_syslog_file_size | The size threshold for rolling over a single log file (oblogservice.log). When this threshold is exceeded, the system switches to a new file. |
| disk_max_iops | The estimated maximum disk IOPS, used for I/O rate limiting. The value must be greater than 0. |
| disk_max_bandwidth | The estimated maximum disk bandwidth, used for I/O rate limiting. The value must be greater than 0. |
| upload_block_concurrency | The concurrency of uploading log blocks to object storage. |
| log_cache_memory_percentage | The percentage of memory_limit that LogCache capacity can use (if memory_limit is set to 0, 70% of the total system memory is used). The actual capacity has a lower limit of 256 MB. If this parameter is set to 0, the cache is completely disabled. |
| log_cache_evict_scan_budget | The upper limit of the budget for scanning reusable space in the cache during a single log publish. |
| log_cache_enable_write | The LogCache write switch. It is enabled by default. |
| log_cache_enable_read | The LogCache read switch. It is enabled by default. |
| log_cache_metrics_dump_interval_sec | The interval (in seconds) at which each node outputs log_cache_metrics monitoring logs. A value of 0 disables this feature (the default, which causes no behavior change during an upgrade). If you set it to a value such as 10, time series data such as hit_rate, hit_bytes, and miss_bytes can be obtained during the verification period. |
| ln_id | The numeric ID of the current node (LN) in the cluster, which is written by the deployment system. |
| local_storage_dir | The path of the local data directory of the node. |