ceph/qa/workunits/mon/pg_autoscaler.sh

   1 #!/bin/bash -ex
   2
   3 NUM_OSDS=$(ceph osd ls | wc -l)
   4 if [ $NUM_OSDS -lt 6 ]; then
   5     echo "test requires at least 6 OSDs"
   6     exit 1
   7 fi
   8
   9 NUM_POOLS=$(ceph osd pool ls | wc -l)
  10 if [ $NUM_POOLS -gt 0 ]; then
  11     echo "test requires no preexisting pools"
  12     exit 1
  13 fi
  14
  15 function wait_for() {
  16     local sec=$1
  17     local cmd=$2
  18
  19     while true ; do
  20         if bash -c "$cmd" ; then
  21             break
  22         fi
  23         sec=$(( $sec - 1 ))
  24         if [ $sec -eq 0 ]; then
  25             echo failed
  26             return 1
  27         fi
  28         sleep 1
  29     done
  30     return 0
  31 }
  32
  33 function power2() { echo "x=l($1)/l(2); scale=0; 2^((x+0.5)/1)" | bc -l;}
  34
  35 function eval_actual_expected_val() {
  36     local actual_value=$1
  37     local expected_value=$2
  38     if [[ $actual_value = $expected_value ]]
  39     then
  40      echo "Success: " $actual_value "=" $expected_value
  41     else
  42       echo "Error: " $actual_value "!=" $expected_value
  43       exit 1
  44     fi
  45 }
  46
  47 # enable
  48 ceph config set mgr mgr/pg_autoscaler/sleep_interval 60
  49 ceph mgr module enable pg_autoscaler
  50 # ceph config set global osd_pool_default_pg_autoscale_mode on
  51
  52 # pg_num_min
  53 ceph osd pool create meta0 16
  54 ceph osd pool create bulk0 16 --bulk
  55 ceph osd pool create bulk1 16 --bulk
  56 ceph osd pool create bulk2 16 --bulk
  57 ceph osd pool set meta0 pg_autoscale_mode on
  58 ceph osd pool set bulk0 pg_autoscale_mode on
  59 ceph osd pool set bulk1 pg_autoscale_mode on
  60 ceph osd pool set bulk2 pg_autoscale_mode on
  61 # set pool size
  62 ceph osd pool set meta0 size 2
  63 ceph osd pool set bulk0 size 2
  64 ceph osd pool set bulk1 size 2
  65 ceph osd pool set bulk2 size 2
  66
  67 # get num pools again since we created more pools
  68 NUM_POOLS=$(ceph osd pool ls | wc -l)
  69
  70 # get bulk flag of each pool through the command ceph osd pool autoscale-status
  71 BULK_FLAG_1=$(ceph osd pool autoscale-status | grep 'meta0' | grep -o -m 1 'True\|False' || true)
  72 BULK_FLAG_2=$(ceph osd pool autoscale-status | grep 'bulk0' | grep -o -m 1 'True\|False' || true)
  73 BULK_FLAG_3=$(ceph osd pool autoscale-status | grep 'bulk1' | grep -o -m 1 'True\|False' || true)
  74 BULK_FLAG_4=$(ceph osd pool autoscale-status | grep 'bulk2' | grep -o -m 1 'True\|False' || true)
  75
  76 # evaluate the accuracy of ceph osd pool autoscale-status specifically the `BULK` column
  77
  78 eval_actual_expected_val $BULK_FLAG_1 'False'
  79 eval_actual_expected_val $BULK_FLAG_2 'True'
  80 eval_actual_expected_val $BULK_FLAG_3 'True'
  81 eval_actual_expected_val $BULK_FLAG_4 'True'
  82
  83 # This part of this code will now evaluate the accuracy of the autoscaler
  84
  85 # get pool size
  86 POOL_SIZE_1=$(ceph osd pool get meta0 size| grep -Eo '[0-9]{1,4}')
  87 POOL_SIZE_2=$(ceph osd pool get bulk0 size| grep -Eo '[0-9]{1,4}')
  88 POOL_SIZE_3=$(ceph osd pool get bulk1 size| grep -Eo '[0-9]{1,4}')
  89 POOL_SIZE_4=$(ceph osd pool get bulk2 size| grep -Eo '[0-9]{1,4}')
  90
  91 # Calculate target pg of each pools
  92 # First Pool is a non-bulk so we do it first.
  93 # Since the Capacity ratio = 0 we first meta pool remains the same pg_num
  94
  95 TARGET_PG_1=$(ceph osd pool get meta0 pg_num| grep -Eo '[0-9]{1,4}')
  96 PG_LEFT=$NUM_OSDS*100
  97 NUM_POOLS_LEFT=$NUM_POOLS-1
  98 # Rest of the pool is bulk and even pools so pretty straight forward
  99 # calculations.
 100 TARGET_PG_2=$(power2 $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_2))))
 101 TARGET_PG_3=$(power2 $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_3))))
 102 TARGET_PG_4=$(power2 $((($PG_LEFT)/($NUM_POOLS_LEFT)/($POOL_SIZE_4))))
 103
 104 # evaluate target_pg against pg num of each pools
 105 wait_for 300 "ceph osd pool get meta0 pg_num | grep $TARGET_PG_1"
 106 wait_for 300 "ceph osd pool get bulk0 pg_num | grep $TARGET_PG_2"
 107 wait_for 300 "ceph osd pool get bulk1 pg_num | grep $TARGET_PG_3"
 108 wait_for 300 "ceph osd pool get bulk2 pg_num | grep $TARGET_PG_4"
 109
 110 # target ratio
 111 ceph osd pool set meta0 target_size_ratio 5
 112 ceph osd pool set bulk0 target_size_ratio 1
 113 sleep 60
 114 APGS=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
 115 BPGS=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
 116 test $APGS -gt 100
 117 test $BPGS -gt 10
 118
 119 # small ratio change does not change pg_num
 120 ceph osd pool set meta0 target_size_ratio 7
 121 ceph osd pool set bulk0 target_size_ratio 2
 122 sleep 60
 123 APGS2=$(ceph osd dump -f json-pretty | jq '.pools[0].pg_num_target')
 124 BPGS2=$(ceph osd dump -f json-pretty | jq '.pools[1].pg_num_target')
 125 test $APGS -eq $APGS2
 126 test $BPGS -eq $BPGS2
 127
 128 # target_size
 129 ceph osd pool set meta0 target_size_bytes 1000000000000000
 130 ceph osd pool set bulk0 target_size_bytes 1000000000000000
 131 ceph osd pool set meta0 target_size_ratio 0
 132 ceph osd pool set bulk0 target_size_ratio 0
 133 wait_for 60 "ceph health detail | grep POOL_TARGET_SIZE_BYTES_OVERCOMMITTED"
 134
 135 ceph osd pool set meta0 target_size_bytes 1000
 136 ceph osd pool set bulk0 target_size_bytes 1000
 137 ceph osd pool set meta0 target_size_ratio 1
 138 wait_for 60 "ceph health detail | grep POOL_HAS_TARGET_SIZE_BYTES_AND_RATIO"
 139
 140 ceph osd pool rm meta0 meta0 --yes-i-really-really-mean-it
 141 ceph osd pool rm bulk0 bulk0 --yes-i-really-really-mean-it
 142 ceph osd pool rm bulk1 bulk1 --yes-i-really-really-mean-it
 143 ceph osd pool rm bulk2 bulk2 --yes-i-really-really-mean-it
 144
 145 echo OK
 146