Cbstats: what is chk_persistence_cmd?

Using cbstats command in our architecture we noticed that there is one point with high timings (>4sec):

chk_persistence_cmd (684 total)
64us - 128us  : (  0.44%)   3
128us - 256us : (  0.88%)   3
256us - 512us : (  1.32%)   3
512us - 1ms   : (  1.46%)   1
1ms - 2ms     : (  2.19%)   5 #
2ms - 4ms     : (  3.51%)   9 ##
4ms - 8ms     : (  4.68%)   8 ##
8ms - 16ms    : (  5.26%)   4 #
16ms - 32ms   : (  6.73%)  10 ###
32ms - 65ms   : (  8.92%)  15 ####
65ms - 131ms  : ( 10.96%)  14 ####
131ms - 262ms : ( 15.50%)  31 #########
262ms - 524ms : ( 22.66%)  49 ###############
524ms - 1s    : ( 34.36%)  80 ########################
1s - 2s       : ( 52.05%) 121 #####################################
2s - 4s       : ( 84.36%) 221 ####################################################################
4s - 8s       : ( 96.93%)  86 ##########################
8s - 16s      : (100.00%)  21 ######

What is this? Can someone explain it please?
We would like to decrease this time if it is correlated with Couchbase architecture performance.

Thank you
Matteo

It’s administrative / monitoring command used between ns_server (cluster manager) and memcached (KV-engine) to check when a particular sequence number is persisted.

You shouldn’t need to worry about it - it’s essentially a background task related to checking where persistence is up to, and shouldn’t have any impact on normal front-end operations.

Ok thank you @drigby.

We were checking cbstats output because we were experiencing an higher number of Client-side Timeout, so inspecting the cluster only that parameter had a value higher than 2 seconds. After about an hour client-side timeouts came again to zero.