Skip to content

MongoDB FTDC and Grafana Integration

Jump to bottom
Ken Chen edited this page Nov 6, 2020 · 11 revisions

MongoDB FTDC Data and Grafana Integration

Keyhole can feed time series data into Grafana as a Simple JSON Datasource.

Instructions

  • Install Grafana
  • Install SimpleJSON plugin
  • Configure Simple JSON data store. Use http://<hostname>:5408/grafana as the HTTP URL.
  • Import Keyhole Analytic Dashboard template.
  • Use keyhole to read from diagnostic.data, for example
keyhole --web --diag /data/db/diagnostic.data
  • Modify from and to date/time in Grafana

Reinit Data Points

curl -XPOST http://localhost:5408/grafana/dir -d '{"dir": "/new/diagnostic.data"}'

The Docker Way

Keyhole Docker container is also available from Docker Hub. This greatly reduce the need to configure Grafana. See mongo-ftdc for details. The Dockerfile follows the instructions above to configure Granfana and to build a Docker container image.

Dashboard Snapshot

By default, Granfana uses port 3000, http://localhost:3000. After entering correct date/time range, you can see results similar to below. Note that if you use the docker-compose template, port 3030 is used.

MongoDB FTDC Analytics Snapshot

Time Series Data Points

Available time series data points are

  • mem_resident - resident memory used
  • mem_virtual - virtual memory used
  • mem_page_faults - page faults
  • conns_available - Connections available
  • conns_current - current connections opened
  • conns_created_per_minute - connections created per minute
  • ops_query - number of queries
  • ops_insert - number of insertions
  • ops_update - number of updates
  • ops_delete - number of deletions
  • ops_getmore - number of getmore command
  • ops_command - number of commands
  • q_active_read - number of active reads
  • q_active_write - number of active writes
  • q_queued_read - number of reads queued
  • q_queued_write - number of writes queued
  • latency_read - average read latency in milliseconds
  • latency_write - average write latency in milliseconds
  • latency_command - average command latency in milliseconds
  • scan_keys - number of index keys scanned
  • scan_objects - number of documents scanned
  • scan_sort - number of documents loaded into memory to sort
  • wt_cache_max - wiredTiger max cache bytes configured in GB
  • wt_cache_used - wiredTiger cache used in GB
  • wt_cache_dirty - wiredTiger cache modified in GB
  • wt_modified_evicted - modified pages evicted from wiredTiger cache
  • wt_unmodified_evicted - unmodified pages evicted from wiredTiger cache
  • wt_read_in_cache - pages read into wiredTiger cache
  • wt_written_from_cache - pages written written from wiredTiger cache
  • ticket_avail_read - available wiredTiger read tickets
  • ticket_avail_write - available wiredTiger write tickets
  • replication_lags - replication lags in seconds
  • cpu_idle - CPU idle %
  • cpu_iowait - CPU I/O wait %
  • cpu_nice - CPU nice process %
  • cpu_softirq - CPU soft interrupt %
  • cpu_steal - CPU steal %
  • cpu_system - CPU system %
  • cpu_user - CPU user %
  • disks_iops - disks IOPS
  • disks_utils - disks utilizations (%)
Clone this wiki locally