0

The chart below is a Redis Cluster Hit Ratio. For no reason one instance (6394) started missing keys dramatically since a week ago.

Redis Cluster Hit Ratio: 

#Stats
expired_keys:55437553
evicted_keys:2735570
keyspace_hits:8713704804
keyspace_misses:459444958

8 secs later 1,224 misses:

# Stats
expired_keys:55437845
evicted_keys:2735570
keyspace_hits:8713727951
keyspace_misses:459446182

No evictions. No max memory problems. The rest of the shards are running great. Monitor don't show anything wrong, just the boring GET, SET, DEL, etc.. Log level set on debug mode is all ok: GOSSIP ok, Accepted form XX, CLienmt closed connection, etc, etc..

It sounds silly but can someone please give me all the possible reasons for missed key? For example can a blocking command affect the "keyspace_misses" number? What could be happening?

Jirene
  • 151
  • 3
  • 9
  • Were you making requests for keys that didn't exist? – TheDude Aug 14 '18 at 21:26
  • Thanks @TheDude for answering. As far I've tested, they do. Most of the commands being executed are GET, SET, HGET, DEL, RPUSH, LRANGE, SADD and SETEX. My test was get several GET/HGET commands from Monitor and just check if those keys exists. Is there any other better way? I – Jirene Aug 15 '18 at 17:13

0 Answers0