This topic describes the inspection metrics of obdiag V1.5.2 and later.
OBServer node-related metrics
Type |
Task |
Inspection purpose |
|---|---|---|
| cluster | core_file_find | Checks for OBServer nodes that have caused the generation of core dump files. |
| deadlocks | Checks for deadlocks. | |
| major | Checks for major compaction tasks that have been in execution for more than 36 hours. | |
| part_trans_action_max | Checks whether the number of transaction participants exceeds 200. | |
| tenant_log_size | Checks whether the size of logs in the log disk is three times or more of the memory size. | |
| data_path_settings | Checks whether the data disk and log disk are set to the same disk. | |
| mod_too_large | Checks whether the number of global indexes exceeds 20. | |
| mod_too_large | Checks for user-side modules that have occupied more than 10 GB of memory space. | |
| task_opt_stat_gather_fail | Checks for failed history collection tasks. | |
| ob_enable_plan_cache_bad_version | Checks for OceanBase Database versions for which ob_enable_plan_cache is enabled. |
|
| observer_not_active | Checks for OBServer nodes that are not started. | |
| sys_log_level | Checks the level of syslogs. | |
| optimizer_better_inlist_costing | Checks for OceanBase Database versions for which optimizer_better_inlist_costing is enabled. |
|
| resource_limit_max_session_num | Checks the maximum number of sessions. | |
| table_history_too_many | Checks for tables that have excessive historical records. | |
| tenant_number | Checks whether the number of tenants is excessively large. | |
| cpu | oversold | Checks for OBServer nodes that are not in the ACTIVE state. |
| disk | disk_full | Checks for full disk usage. A warning is sent if the disk usage exceeds 80% and an error is sent if it exceeds 90%. |
| disk_hole | Checks for disk holes. | |
| err_code | find_err_* | Checks for error codes of serious errors. |
| system | aio | Checks whether asynchronous I/O is correctly configured. |
| dependent_software | Checks whether the dependencies of basic software are correctly configured. For more information, see Host standardization check items. | |
| parameter* | Checks whether the kernel parameters are configured as required. For more information, see Host standardization check items. | |
| ulimit_parameter | Checks whether the ulimit parameters are configured as required. For more information, see Host standardization check items. |
|
| clock_source | Checks whether the time information matches the clock source. | |
| core_pattern | Checks whether the kernel.core_pattern parameter is normal. |
|
| dependent_software_swapon | Checks whether swapon is properly configured. | |
| instruction_set_avx2 | Checks whether the Advanced Vector Extensions 2 (AVX2) instruction set is normal. | |
| version | bad_version | Checks whether an unrecommended OceanBase Database version is used. |
| sysbench | sysbench_free_test_cpu_count | Checks whether the value of cpu_count is 80% to 100% of the total number of CPU cores of the server. |
| sysbench_run_test_tenant_memory_used | Checks whether the sum of used or held memory is less than 80% of the total memory for the tenant. | |
| sysbench_test_cpu_quota_concurrency | Checks whether the CPU quota concurrency is 2 or 4. |
|
| sysbench_test_tenant_log_disk_size | Checks whether the size of the log disk is less than 20 GB. | |
| sysbench_free_test_memory_limit | Checks whether the allocated memory is 80% to 90% of the memory for the server. | |
| sysbench_test_cluster_datafile_size | Checks whether the the percentage of disk space for data files is 60% if data files are stored on the same disk as logs, and 90% if data files are stored on an exclusive disk. | |
| sysbench_test_log_level | Checks whether the log level is INFO, WARN, or ERROR. |
|
| sysbench_test_tenant_primary_zone | Checks whether the primary zone is set to RANDOM. |
|
| sysbench_free_test_network_speed | Checks whether the network interface card (NIC) provides a bandwidth greater than 1 Gbit/s. | |
| sysbench_test_cluster_log_disk_size | Checks whether the percentage of disk space for log files is 30% if logs and data are stored on the same disk, and 90% if logs and data are stored on different disks. | |
| sysbench_test_sql_net_thread_count | Checks whether net_thread_count is set to 1/8 of the value of cpu_count, and whether sql_net_thread_count is set to 0 or the same value as net_thread_count. |
|
| sysbench_run_test_tenant_cpu_used | Checks whether the value is greater than 90% of the value of the max_cpu parameter. |
|
| sysbench_test_cluster_parameters | Checks whether the enable_sql_audit, enable_perf_event, and enable_record_trace_log parameters are set to false. |
|
| sysbench_test_tenant_cpu_parameters | Checks whether the max_iops parameter is set to a value greater than the value of max_cpu × 10000. |
|
| bugs | bug_182 | For more information, see issue #182. |
| table | information_schema_tables_two_data | Checks for tables that have two records in the information_schema.tables view. |
| column_storage | tenant_parameters | Checks proof of concept (POC) configurations for columnar storage. |
ODP-related metrics
Type |
Task |
Inspection purpose |
|---|---|---|
| version | bad_version | Checks whether an unrecommended OceanBase Database Proxy (ODP) version is used. |