OceanBase

·

Collecting OceanBase performance metrics through the sys tenant.

Already tested version:

• OceanBase Enterprise 3.2.4

Configuration¶

Precondition¶

• Create a monitoring account

Create a monitoring account using a sys tenant account and grant the following privileges:

CREATE USER 'datakit'@'localhost' IDENTIFIED BY '<UNIQUEPASSWORD>';

-- MySQL 8.0+ create the datakit user with the caching_sha2_password method
CREATE USER 'datakit'@'localhost' IDENTIFIED WITH caching_sha2_password by '<UNIQUEPASSWORD>';

-- Grant the required permissions 
GRANT SELECT ON *.* TO 'datakit'@'localhost';

Collector Configuration¶

[[inputs.oceanbase]]
  # host name
  host = "localhost"

  ## port
  port = 2883

  ## tenant name
  tenant = "sys"

  ## cluster name
  cluster = "obcluster"

  ## user name
  user = "datakit"

  ## password
  password = "<PASS>"

  ## database name
  database = "oceanbase"

  ## mode. mysql only.
  mode = "mysql"

  ## @param connect_timeout - number - optional - default: 10s
  # connect_timeout = "10s"

  interval = "10s"

  ## OceanBase slow query time threshold defined. If larger than this, the executed sql will be reported.
  slow_query_time = "0s"

  ## Set true to enable election
  election = true

  ## Run a custom SQL query and collect corresponding metrics.
  # [[inputs.oceanbase.custom_queries]]
    # sql = '''
    #   select
    #     CON_ID tenant_id,
    #     STAT_ID,
    #     replace(name, " ", "_") metric_name,
    #     VALUE
    #   from
    #     v$sysstat;
    # '''
    # metric = "oceanbase_custom"
    # tags = ["metric_name", "tenant_id"]
    # fields = ["VALUE"]

  [inputs.oceanbase.tags]
    # some_tag = "some_value"
    # more_tag = "some_other_value"

Long Running Queries¶

DataKit could reports the SQLs, those executed time exceeded the threshold time defined by user, to Guance, displays in the Logs side bar, the source name is oceanbase_log.

This function is disabled by default, user could enabling it by modify DataKit's OceanBase configuration like followings:

Change the string value after slow_query_time from 0s to the threshold time, minimal value is 1 millsecond. Generally, recommand it to 10s.

slow_query_time = "0s"

Metric¶

For all of the following data collections, the global election tags will added automatically, we can add extra tags in [inputs.oceanbase.tags] if needed:

 [inputs.oceanbase.tags]
  # some_tag = "some_value"
  # more_tag = "some_other_value"
  # ...

oceanbase_stat¶

OceanBase system statistics from gv$sysstat, reported by cluster, tenant, server IP, stat ID, and metric name.

oceanbase_cache_block¶

OceanBase block cache size from __all_virtual_kvcache_info, grouped by cluster, tenant, server, and cache name.

oceanbase_cache_plan¶

OceanBase plan cache access and hit counters from gv$plan_cache_stat, grouped by cluster, tenant, server IP, and server port.

oceanbase_event¶

OceanBase non-idle system wait event totals from gv$system_event, grouped by cluster, tenant, server IP, and event group.

oceanbase_session¶

OceanBase current session counts from __all_virtual_processlist, grouped by cluster, tenant, server IP, and server port.

oceanbase_clog¶

OceanBase clog synchronization delay from virtual clog statistics, grouped by cluster, tenant, server, and replica type.

Log¶

oceanbase_log¶

OceanBase slow-query log records collected from GV$SQL_AUDIT for the configured database service.