qa: Add overlapped roots test case

Added overlapped roots test case for CRUSH rules
rule zone-1-only
  step take zone us-east-1
  step chooseleaf firstn 3 type host
  step emit

rule zone-2-only
  step take zone us-east-2
  step chooseleaf firstn 3 type host
  step emit

rule zone-3-only
  step take zone us-east-3
  step chooseleaf firstn 3 type host
  step emit

rule default
   step take default
  step chooseleaf firstn 3 type host
  step emit

and osd tree
    -1         6.00000  root default
     -8         6.00000      region us-east
     -5         2.00000          zone us-east-1
    -13         1.00000              host host0
      0    ssd  1.00000                  osd.0
    -14         1.00000              host host1
      1    ssd  1.00000                  osd.1
     -6         2.00000          zone us-east-2
    -15         1.00000              host host2
      2    ssd  1.00000                  osd.2
    -16         1.00000              host host3
      3    ssd  1.00000                  osd.3
     -7         2.00000          zone us-east-3
    -17         1.00000              host host4
      4    ssd  1.00000                  osd.4
    -18         1.00000              host host5
      5    ssd  1.00000                  osd.5

    -1: pg_target = 300
    -5: pg_target = 100
    -6  pg_target = 100
    -7: pg_target = 100

Fixes: https://tracker.ceph.com/issues/73892

Signed-off-by: Eric Zhang <emzhang@ibm.com>
(cherry picked from commit 7f9e383ecb)

Conflicts:
	qa/workunits/mon/pg_autoscaler.sh
-Add new test cases for overlapped roots and separate different parts of script into
functions
This commit is contained in:
Eric Zhang 2026-05-27 15:55:59 -07:00 committed by Ericmzhang
parent 76e5bbc2b4
commit ddb0211214

View File

@ -46,151 +46,326 @@ function eval_actual_expected_val() {
fi
}
function test_autoscaler_basic() {
# pg_num_min
MON_TARGET_PG_PER_OSD=200
ceph config set global mon_target_pg_per_osd $MON_TARGET_PG_PER_OSD
ceph osd pool create meta0
ceph osd pool create bulk0 --bulk
ceph osd pool create bulk1 --bulk
ceph osd pool create bulk2 --bulk
ceph osd pool set meta0 pg_autoscale_mode on
ceph osd pool set bulk0 pg_autoscale_mode on
ceph osd pool set bulk1 pg_autoscale_mode on
ceph osd pool set bulk2 pg_autoscale_mode on
# set pool size
ceph osd pool set meta0 size 2
ceph osd pool set bulk0 size 2
ceph osd pool set bulk1 size 2
ceph osd pool set bulk2 size 2
# get num pools again since we created more pools
NUM_POOLS=$(ceph osd pool ls | wc -l)
# get bulk flag of each pool through the command ceph osd pool autoscale-status
BULK_FLAG_1=$(ceph osd pool autoscale-status | grep 'meta0' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_2=$(ceph osd pool autoscale-status | grep 'bulk0' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_3=$(ceph osd pool autoscale-status | grep 'bulk1' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_4=$(ceph osd pool autoscale-status | grep 'bulk2' | grep -o -m 1 'True\|False' || true)
# evaluate the accuracy of ceph osd pool autoscale-status specifically the `BULK` column
eval_actual_expected_val $BULK_FLAG_1 'False'
eval_actual_expected_val $BULK_FLAG_2 'True'
eval_actual_expected_val $BULK_FLAG_3 'True'
eval_actual_expected_val $BULK_FLAG_4 'True'
# This part of this code will now evaluate the accuracy of the autoscaler
# get pool size
POOL_SIZE_1=$(ceph osd pool get meta0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_2=$(ceph osd pool get bulk0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_3=$(ceph osd pool get bulk1 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_4=$(ceph osd pool get bulk2 size| grep -Eo '[0-9]{1,4}')
sleep 30
# Calculate target pg of each pools
TARGET_PG_1=$(ceph osd pool get meta0 pg_num| grep -Eo '[0-9]{1,4}')
PG_LEFT=$(($NUM_OSDS*$MON_TARGET_PG_PER_OSD - TARGET_PG_1 * $POOL_SIZE_1))
NUM_POOLS_LEFT=$NUM_POOLS-1
# Rest of the pool is bulk and even pools so pretty straight forward
# calculations.
TARGET_PG_2=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_2))))
TARGET_PG_3=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_3))))
TARGET_PG_4=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_4))))
# evaluate target_pg against pg num of each pools
wait_for 300 "ceph osd pool get meta0 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get bulk0 pg_num | grep $TARGET_PG_2"
wait_for 300 "ceph osd pool get bulk1 pg_num | grep $TARGET_PG_3"
wait_for 300 "ceph osd pool get bulk2 pg_num | grep $TARGET_PG_4"
# target ratio
ceph osd pool set meta0 target_size_ratio 5
ceph osd pool set bulk0 target_size_ratio 2
sleep 60
APGS=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
BPGS=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
test $APGS -gt 100
test $BPGS -gt 10
# small ratio change does not change pg_num
ceph osd pool set meta0 target_size_ratio 7
ceph osd pool set bulk0 target_size_ratio 2
sleep 60
APGS2=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
BPGS2=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
test $APGS -eq $APGS2
test $BPGS -eq $BPGS2
# target_size
ceph osd pool set meta0 target_size_bytes 1000000000000000
ceph osd pool set bulk0 target_size_bytes 1000000000000000
ceph osd pool set meta0 target_size_ratio 0
ceph osd pool set bulk0 target_size_ratio 0
wait_for 60 "ceph health detail | grep POOL_TARGET_SIZE_BYTES_OVERCOMMITTED"
ceph osd pool set meta0 target_size_bytes 1000
ceph osd pool set bulk0 target_size_bytes 1000
ceph osd pool set meta0 target_size_ratio 1
wait_for 60 "ceph health detail | grep POOL_HAS_TARGET_SIZE_BYTES_AND_RATIO"
# test autoscale warn
ceph osd pool create warn0 1 --autoscale-mode=warn
wait_for 120 "ceph health detail | grep POOL_TOO_FEW_PGS"
ceph osd pool create warn1 256 --autoscale-mode=warn
wait_for 120 "ceph health detail | grep POOL_TOO_MANY_PGS"
ceph osd pool rm meta0 meta0 --yes-i-really-really-mean-it
ceph osd pool rm bulk0 bulk0 --yes-i-really-really-mean-it
ceph osd pool rm bulk1 bulk1 --yes-i-really-really-mean-it
ceph osd pool rm bulk2 bulk2 --yes-i-really-really-mean-it
ceph osd pool rm warn0 warn0 --yes-i-really-really-mean-it
ceph osd pool rm warn1 warn1 --yes-i-really-really-mean-it
}
function test_pool_starvation() {
# test pool starvation where PGs not distributed evenly
MON_TARGET_PG_PER_OSD=200
ceph config set global mon_target_pg_per_osd $MON_TARGET_PG_PER_OSD
ceph osd pool create data0
ceph osd pool create data1
ceph osd pool create data2
ceph osd pool set data0 pg_autoscale_mode on
ceph osd pool set data1 pg_autoscale_mode on
ceph osd pool set data2 pg_autoscale_mode on
ceph osd pool set data0 target_size_ratio 0.3333
ceph osd pool set data1 target_size_ratio 0.3333
ceph osd pool set data2 target_size_ratio 0.3333
sleep 30
PG_LEFT=$(($NUM_OSDS*$MON_TARGET_PG_PER_OSD))
POOL_SIZE=$(ceph osd pool get data0 size| grep -Eo '[0-9]{1,4}')
NUM_POOLS=$(ceph osd pool ls | wc -l)
TARGET_PG=$(power2_floor $((($PG_LEFT)/($NUM_POOLS)/($POOL_SIZE))))
wait_for 300 "ceph osd pool get data0 pg_num | grep $TARGET_PG"
wait_for 300 "ceph osd pool get data1 pg_num | grep $TARGET_PG"
wait_for 300 "ceph osd pool get data2 pg_num | grep $TARGET_PG"
ceph osd pool rm data0 data0 --yes-i-really-really-mean-it
ceph osd pool rm data1 data1 --yes-i-really-really-mean-it
ceph osd pool rm data2 data2 --yes-i-really-really-mean-it
}
function test_exact_budget() {
# test edge case for exact budget
MON_TARGET_PG_PER_OSD=128
ceph config set global mon_target_pg_per_osd $MON_TARGET_PG_PER_OSD
ceph osd pool create data0
ceph osd pool create data1
ceph osd pool set data0 pg_autoscale_mode on
ceph osd pool set data1 pg_autoscale_mode on
ceph osd pool set data0 size 3
ceph osd pool set data1 size 3
ceph osd pool set data0 target_size_ratio 0.5
ceph osd pool set data1 target_size_ratio 0.5
sleep 60
POOL_SIZE_1=$(ceph osd pool get data0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_2=$(ceph osd pool get data1 size| grep -Eo '[0-9]{1,4}')
PG_LEFT=$(($NUM_OSDS * $MON_TARGET_PG_PER_OSD))
NUM_POOLS=$(ceph osd pool ls | wc -l)
TARGET_PG_1=$(power2_floor $((($PG_LEFT)/($NUM_POOLS)/($POOL_SIZE_1))))
TARGET_PG_2=$(power2_floor $((($PG_LEFT)/($NUM_POOLS)/($POOL_SIZE_2))))
wait_for 300 "ceph osd pool get data0 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data1 pg_num | grep $TARGET_PG_2"
ceph osd pool rm data0 data0 --yes-i-really-really-mean-it
ceph osd pool rm data1 data1 --yes-i-really-really-mean-it
}
function test_overlapping_roots() {
# Create custom CRUSH rules for zone-based placement
MON_TARGET_PG_PER_OSD=100
ceph config set global mon_target_pg_per_osd $MON_TARGET_PG_PER_OSD
ceph osd crush add-bucket us-east region
ceph osd crush move us-east root=default
ceph osd crush add-bucket us-east-1 zone
ceph osd crush add-bucket us-east-2 zone
ceph osd crush add-bucket us-east-3 zone
ceph osd crush move us-east-1 region=us-east
ceph osd crush move us-east-2 region=us-east
ceph osd crush move us-east-3 region=us-east
ceph osd crush add-bucket host0 host
ceph osd crush add-bucket host1 host
ceph osd crush add-bucket host2 host
ceph osd crush add-bucket host3 host
ceph osd crush add-bucket host4 host
ceph osd crush add-bucket host5 host
ceph osd crush move host0 zone=us-east-1
ceph osd crush move host1 zone=us-east-1
ceph osd crush move host2 zone=us-east-2
ceph osd crush move host3 zone=us-east-2
ceph osd crush move host4 zone=us-east-3
ceph osd crush move host5 zone=us-east-3
ceph osd crush set osd.0 1.0 host=host0
ceph osd crush set osd.1 1.0 host=host1
ceph osd crush set osd.2 1.0 host=host2
ceph osd crush set osd.3 1.0 host=host3
ceph osd crush set osd.4 1.0 host=host4
ceph osd crush set osd.5 1.0 host=host5
ceph osd crush move us-east root=default
# Add zone-based CRUSH rules
hostname=$(hostname -s)
ceph osd crush remove $hostname
ceph osd getcrushmap > crushmap
crushtool --decompile crushmap > crushmap.txt
sed 's/^# end crush map$//' crushmap.txt > crushmap_modified.txt
cat >> crushmap_modified.txt << EOF
rule zone-1-only {
id 1
type replicated
step take us-east-1
step chooseleaf firstn 3 type host
step emit
}
rule zone-2-only {
id 2
type replicated
step take us-east-2
step chooseleaf firstn 3 type host
step emit
}
rule zone-3-only {
id 3
type replicated
step take us-east-3
step chooseleaf firstn 3 type host
step emit
}
rule default {
id 4
type replicated
step take default
step chooseleaf firstn 3 type host
step emit
}
# end crush map
EOF
# compile the modified crushmap and set it
crushtool --compile crushmap_modified.txt -o crushmap.bin
ceph osd setcrushmap -i crushmap.bin
ceph osd pool create data0 --size=1
ceph osd pool create data1 --size=1
ceph osd pool create data2 --size=1
ceph osd pool create data3 --size=3
ceph osd pool set data0 pg_autoscale_mode on
ceph osd pool set data1 pg_autoscale_mode on
ceph osd pool set data2 pg_autoscale_mode on
ceph osd pool set data3 pg_autoscale_mode on
ceph osd pool set data0 crush_rule zone-1-only
ceph osd pool set data1 crush_rule zone-2-only
ceph osd pool set data2 crush_rule zone-3-only
ceph osd pool set data3 crush_rule default
ceph osd pool set data0 target_size_ratio 1.0
ceph osd pool set data1 target_size_ratio 1.0
ceph osd pool set data2 target_size_ratio 1.0
ceph osd pool set data3 target_size_ratio 1.0
# expect
# -1 6.00000 root default
# -8 6.00000 region us-east
# -5 2.00000 zone us-east-1
# -13 1.00000 host host0
# 0 ssd 1.00000 osd.0
# -14 1.00000 host host1
# 1 ssd 1.00000 osd.1
# -6 2.00000 zone us-east-2
# -15 1.00000 host host2
# 2 ssd 1.00000 osd.2
# -16 1.00000 host host3
# 3 ssd 1.00000 osd.3
# -7 2.00000 zone us-east-3
# -17 1.00000 host host4
# 4 ssd 1.00000 osd.4
# -18 1.00000 host host5
# 5 ssd 1.00000 osd.5
# -1: pg_target = 300
# -5: pg_target = 100
# -6 pg_target = 100
# -7: pg_target = 100
TARGET_PG_1=$(power2_floor $MON_TARGET_PG_PER_OSD)
TARGET_PG_2=$(power2_floor $(( ($NUM_OSDS - 3) * $MON_TARGET_PG_PER_OSD )))
wait_for 300 "ceph osd pool get data0 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data1 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data2 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data3 pg_num | grep $TARGET_PG_2"
ceph osd pool rm data0 data0 --yes-i-really-really-mean-it
ceph osd pool rm data1 data1 --yes-i-really-really-mean-it
ceph osd pool rm data2 data2 --yes-i-really-really-mean-it
ceph osd pool rm data3 data3 --yes-i-really-really-mean-it
}
# enable
ceph config set mgr mgr/pg_autoscaler/sleep_interval 60
ceph mgr module enable pg_autoscaler
# ceph config set global osd_pool_default_pg_autoscale_mode on
ceph osd pool set threshold 1.0
# pg_num_min
ceph osd pool create meta0 16
ceph osd pool create bulk0 16 --bulk
ceph osd pool create bulk1 16 --bulk
ceph osd pool create bulk2 16 --bulk
ceph osd pool set meta0 pg_autoscale_mode on
ceph osd pool set bulk0 pg_autoscale_mode on
ceph osd pool set bulk1 pg_autoscale_mode on
ceph osd pool set bulk2 pg_autoscale_mode on
# set pool size
ceph osd pool set meta0 size 2
ceph osd pool set bulk0 size 2
ceph osd pool set bulk1 size 2
ceph osd pool set bulk2 size 2
# get num pools again since we created more pools
NUM_POOLS=$(ceph osd pool ls | wc -l)
# get bulk flag of each pool through the command ceph osd pool autoscale-status
BULK_FLAG_1=$(ceph osd pool autoscale-status | grep 'meta0' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_2=$(ceph osd pool autoscale-status | grep 'bulk0' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_3=$(ceph osd pool autoscale-status | grep 'bulk1' | grep -o -m 1 'True\|False' || true)
BULK_FLAG_4=$(ceph osd pool autoscale-status | grep 'bulk2' | grep -o -m 1 'True\|False' || true)
# evaluate the accuracy of ceph osd pool autoscale-status specifically the `BULK` column
eval_actual_expected_val $BULK_FLAG_1 'False'
eval_actual_expected_val $BULK_FLAG_2 'True'
eval_actual_expected_val $BULK_FLAG_3 'True'
eval_actual_expected_val $BULK_FLAG_4 'True'
# This part of this code will now evaluate the accuracy of the autoscaler
# get pool size
POOL_SIZE_1=$(ceph osd pool get meta0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_2=$(ceph osd pool get bulk0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_3=$(ceph osd pool get bulk1 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_4=$(ceph osd pool get bulk2 size| grep -Eo '[0-9]{1,4}')
# Calculate target pg of each pools
# First Pool is a non-bulk so we do it first.
# Since the Capacity ratio = 0 we first meta pool remains the same pg_num
TARGET_PG_1=$(ceph osd pool get meta0 pg_num| grep -Eo '[0-9]{1,4}')
PG_LEFT=$(($NUM_OSDS*100 - $TARGET_PG_1 * $POOL_SIZE_1))
NUM_POOLS_LEFT=$NUM_POOLS-1
# Rest of the pool is bulk and even pools so pretty straight forward
# calculations.
TARGET_PG_2=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_2))))
TARGET_PG_3=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_3))))
TARGET_PG_4=$(power2_floor $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_4))))
# evaluate target_pg against pg num of each pools
wait_for 300 "ceph osd pool get meta0 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get bulk0 pg_num | grep $TARGET_PG_2"
wait_for 300 "ceph osd pool get bulk1 pg_num | grep $TARGET_PG_3"
wait_for 300 "ceph osd pool get bulk2 pg_num | grep $TARGET_PG_4"
# target ratio
ceph osd pool set meta0 target_size_ratio 5
ceph osd pool set bulk0 target_size_ratio 2
sleep 60
APGS=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
BPGS=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
test $APGS -gt 100
test $BPGS -gt 10
# small ratio change does not change pg_num
ceph osd pool set meta0 target_size_ratio 7
ceph osd pool set bulk0 target_size_ratio 2
sleep 60
APGS2=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
BPGS2=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
test $APGS -eq $APGS2
test $BPGS -eq $BPGS2
# target_size
ceph osd pool set meta0 target_size_bytes 1000000000000000
ceph osd pool set bulk0 target_size_bytes 1000000000000000
ceph osd pool set meta0 target_size_ratio 0
ceph osd pool set bulk0 target_size_ratio 0
wait_for 60 "ceph health detail | grep POOL_TARGET_SIZE_BYTES_OVERCOMMITTED"
ceph osd pool set meta0 target_size_bytes 1000
ceph osd pool set bulk0 target_size_bytes 1000
ceph osd pool set meta0 target_size_ratio 1
wait_for 60 "ceph health detail | grep POOL_HAS_TARGET_SIZE_BYTES_AND_RATIO"
# test autoscale warn
ceph osd pool create warn0 1 --autoscale-mode=warn
wait_for 120 "ceph health detail | grep POOL_TOO_FEW_PGS"
ceph osd pool create warn1 256 --autoscale-mode=warn
wait_for 120 "ceph health detail | grep POOL_TOO_MANY_PGS"
ceph osd pool rm meta0 meta0 --yes-i-really-really-mean-it
ceph osd pool rm bulk0 bulk0 --yes-i-really-really-mean-it
ceph osd pool rm bulk1 bulk1 --yes-i-really-really-mean-it
ceph osd pool rm bulk2 bulk2 --yes-i-really-really-mean-it
ceph osd pool rm warn0 warn0 --yes-i-really-really-mean-it
ceph osd pool rm warn1 warn1 --yes-i-really-really-mean-it
# test pool starvation where PGs not distributed evenly
ceph osd pool create data0
ceph osd pool create data1
ceph osd pool create data2
ceph osd pool set data0 target_size_ratio 0.3333
ceph osd pool set data1 target_size_ratio 0.3333
ceph osd pool set data2 target_size_ratio 0.3333
sleep 30
TARGET_PG_1=$(ceph osd pool get data0 pg_num| grep -Eo '[0-9]{1,4}')
wait_for 300 "ceph osd pool get data1 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data2 pg_num | grep $TARGET_PG_1"
ceph osd pool rm data0 data0 --yes-i-really-really-mean-it
ceph osd pool rm data1 data1 --yes-i-really-really-mean-it
ceph osd pool rm data2 data2 --yes-i-really-really-mean-it
# test edge case for exact budget
ceph config set global mon_target_pg_per_osd 128
ceph osd pool create data0
ceph osd pool create data1
ceph osd pool set data0 target_size_ratio 0.5
ceph osd pool set data1 target_size_ratio 0.5
POOL_SIZE_1=$(ceph osd pool get data0 size| grep -Eo '[0-9]{1,4}')
POOL_SIZE_2=$(ceph osd pool get data1 size| grep -Eo '[0-9]{1,4}')
PG_LEFT=$(($NUM_OSDS * 128))
NUM_POOLS=$(ceph osd pool ls | wc -l)
TARGET_PG_1=$(power2_floor $((($PG_LEFT)/($NUM_POOLS)/($POOL_SIZE_1))))
TARGET_PG_2=$(power2_floor $((($PG_LEFT)/($NUM_POOLS)/($POOL_SIZE_2))))
wait_for 300 "ceph osd pool get data0 pg_num | grep $TARGET_PG_1"
wait_for 300 "ceph osd pool get data1 pg_num | grep $TARGET_PG_2"
ceph osd pool rm data0 data0 --yes-i-really-really-mean-it
ceph osd pool rm data1 data1 --yes-i-really-really-mean-it
test_autoscaler_basic || return 1
test_pool_starvation || return 1
test_exact_budget || return 1
test_overlapping_roots || exit 1
echo OK