Hi All.
My name is John Haan.
I've been testing Cache Pool using Jewel version on ubuntu 16.04 OS.
I implemented 2 types of cache tiers.
first one is cache pool + erasure pool and the other one is cache pool + replicated pool
I choose writeback mode of cache mode.
vdbench and rados bench are used for test tool.
1. cache pool + erasure pool
(1) write operation : objects are saved to cache pool.
(2) read operation : objects are loaded from cache pool.
2. cache pool + replicated pool
(1) write operation : objects are saved to replicated pool
(2) read operation : objects are loaded from cache pool.
As a result, objects are not saved to cache pool but replicated pool when we implement cache pool and replicated pool.
That was very weird things for me.
Could anybody explain this?
Below is my environment.
=======================================================
1. ceph osd dump | grep pool
pool 23 'erasure-pool' erasure size 13 min_size 10 crush_ruleset 1 object_hash rjenkins pg_num 512 pgp_num 512 last_change 630 lfor 577 flags hashpspool tiers 24 read_tier 24 write_tier 24 stripe_width 4160
pool 24 'cache-pool' replicated size 3 min_size 2 crush_ruleset 2 object_hash rjenkins pg_num 512 pgp_num 512 last_change 660 flags hashpspool,incomplete_clones tier_of 23 cache_mode writeback target_bytes 4000000000000 target_objects 1000000 hit_set bloom{false_positive_probability: 0.05, target_size: 0, seed: 0} 1200s x4 decay_rate 0 search_last_n 0 min_read_recency_for_promote 1 min_write_recency_for_promote 1 stripe_width 0
pool 32 'c-pool' replicated size 3 min_size 2 crush_ruleset 2 object_hash rjenkins pg_num 512 pgp_num 512 last_change 825 flags hashpspool,incomplete_clones tier_of 33 cache_mode writeback target_bytes 4000000000000 target_objects 1000000 hit_set bloom{false_positive_probability: 0.05, target_size: 0, seed: 0} 1200s x6 decay_rate 0 search_last_n 0 min_read_recency_for_promote 1 min_write_recency_for_promote 1 stripe_width 0
pool 33 'r-pool' replicated size 3 min_size 2 crush_ruleset 3 object_hash rjenkins pg_num 512 pgp_num 512 last_change 801 lfor 787 flags hashpspool tiers 32 read_tier 32 write_tier 32 stripe_width 0
2. crush rule list
rule erasure-pool {
ruleset 1
type erasure
min_size 3
max_size 13
step set_chooseleaf_tries 5
step set_choose_tries 100
step take sata
step choose indep 0 type osd
step emit
}
rule ssd_ruleset {
ruleset 2
type replicated
min_size 1
max_size 10
step take ssd
step chooseleaf firstn 0 type host
step emit
}
rule sata_ruleset {
ruleset 3
type replicated
min_size 1
max_size 10
step take sata
step chooseleaf firstn 0 type host
step emit
}
3. ceph df detail
GLOBAL:
SIZE AVAIL RAW USED %RAW USED OBJECTS
43518G 43327G 148G 0.34 35630
POOLS:
NAME ID CATEGORY QUOTA OBJECTS QUOTA BYTES USED %USED MAX AVAIL OBJECTS DIRTY READ WRITE RAW USED
rbd 0 - N/A N/A 0 0 14387G 1 1 8213k 2523k 0
erasure-pool 23 - N/A N/A 40 0 23118G 3 3 33 171k 52
cache-pool 24 - N/A N/A 29946M 0.20 4418G 17081 15030 8643k 7815k 89839M
c-pool 32 - N/A N/A 1021M 0 4418G 2876 206 27316 23965 3064M
r-pool 33 - N/A N/A 62127M 0.42 10018G 15669 15669 2901k 2501k 182G
4. cache pool option
hit_set_type: bloom
hit_set_count: 6
hit_set_period: 1200
min_read_recency_for_promote: 1
min_write_recency_for_promote: 1
cache_min_flush_age: 600
cache_min_evict_age: 1800
target_max_bytes: 4000000000000
target_max_objects: 1000000
cache_target_dirty_ratio: 0.6
cache_target_dirty_high_ratio: 0.7
cache_target_full_ratio: 0.8
===============================================================
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
