LATENCY GRAPH event

Available since 2.8.13.

Produces an ASCII-art style graph for the specified event.

LATENCY GRAPH lets you intuitively understand the latency trend of an event via state-of-the-art visualization. It can be used for quickly grasping the situation before resorting to means such parsing the raw data from LATENCY HISTORY or external tooling.

Valid values for event are: active-defrag-cycle aof-fsync-always aof-stat aof-rewrite-diff-write aof-rename aof-write aof-write-active-child aof-write-alone aof-write-pending-fsync command expire-cycle eviction-cycle eviction-del fast-command fork rdb-unlink-temp-file

  1. 127.0.0.1:6379> latency reset command
  2. (integer) 0
  3. 127.0.0.1:6379> debug sleep .1
  4. OK
  5. 127.0.0.1:6379> debug sleep .2
  6. OK
  7. 127.0.0.1:6379> debug sleep .3
  8. OK
  9. 127.0.0.1:6379> debug sleep .5
  10. OK
  11. 127.0.0.1:6379> debug sleep .4
  12. OK
  13. 127.0.0.1:6379> latency graph command
  14. command - high 500 ms, low 101 ms (all time high 500 ms)
  15. --------------------------------------------------------------------------------
  16.    #_
  17.   _||
  18.  _|||
  19. _||||
  21. 11186
  22. 542ss
  23. sss

The vertical labels under each graph column represent the amount of seconds, minutes, hours or days ago the event happened. For example "15s" means that the first graphed event happened 15 seconds ago.

The graph is normalized in the min-max scale so that the zero (the underscore in the lower row) is the minimum, and a # in the higher row is the maximum.

For more information refer to the Latency Monitoring Framework page.

*Return value

Bulk string reply