Internal time series databases
_internal_factry is a time series database used to store historian and collector logs and statistics. This database gets configured during the “First time setup” after installing the historian. The database is initialized with a retention policy named _internal_factry_rp of 12 weeks which means that data older than 12 weeks is not kept.
Logs
InfluxDB measurement: logs.
You can view these logs in the historian.
Only message is an InfluxDB field, plus error when the log entry carries one. Everything else below is a tag, which matters when querying: tags can be used in GROUP BY and are indexed, fields cannot.
Common tags
- application (2 options, reflects the type of the log)
- historian-server
- historian-collector
- level
- log_type, derived from the entity the entry relates to: system, event, forwarder, measurement, collector, worker_pool, timeseries_database or organization
Server logs
Logs from the historian server. They always carry the version and build tags. Which logs are stored are defined by the LOG_LEVEL environmental variableenvironmental variable .
Where possible server logs are linked to the relevant entity by a <entity type>_uuid tag; eg. collector_uuid, measurement_uuid, event_configuration_uuid,...
Collector logs
Logs sent by the collectors. Which logs get sent depends on the LogLevel setting on the collector.
Extra tags:
- collector_name
- collector_type
- collector_uuid
- measurement_uuid
- version
Statistics
InfluxDB measurement: statistics.
Statistics are written by both the historian server and the collectors, and the two write different field sets. Server points carry the tags application (value historian) and version. Collector points carry collector_uuid, collector_name, collector_type and version, and no application tag, so the two are told apart by the presence of the collector tags rather than by application.
Both producers write once per minute.
System
The two producers write different fields, so not every field below is present on every point.
Written by the historian server
- cpu_usage (%): average over the last minute
- memory_usage (%): the instantaneous used percentage at the last refresh, not an average
- system_uptime (seconds)
- uptime (seconds): Uptime of the Historian
- points_written: Points written per second, refreshed at most every 10 seconds
- cpu_count: Amount of cpu’s on machine
- cpu_load: Average cpu load
- memory_total: Total available memory
- memory_free: Available memory
- the cache_* fields described under Memory cache
The server also writes one extra point per mounted filesystem, tagged with mount_point, carrying total_disk_space, free_disk_space, read_count, write_count, read_bytes and write_bytes. Free disk space is only reported on these per-mount points.
Written by the collectors
- cpu_usage (%)
- memory_usage (%)
- free_disk_space (bytes): Free disk space of the drive the collector is installed on
- system_uptime (seconds)
Collector
- collector_name
- collector_type
- collector_uuid
- collector_uptime (seconds)
- version
- bad_points: amount per minute
- good_points: amount per minute
- points_sent: amount per minute
Buffer
- buffer_current_size: amount of items held in memory across all five collector buffers (logs, statistics, points, measurements and health), not points alone
- buffer_error_files: amount of invalid buffer files
- buffer_disk_flushes: amount of disk flushes since last statistics update