summaryrefslogtreecommitdiff
path: root/lib/gitlab/database/batch_count.rb
diff options
context:
space:
mode:
authorGitLab Bot <gitlab-bot@gitlab.com>2020-08-20 18:42:06 +0000
committerGitLab Bot <gitlab-bot@gitlab.com>2020-08-20 18:42:06 +0000
commit6e4e1050d9dba2b7b2523fdd1768823ab85feef4 (patch)
tree78be5963ec075d80116a932011d695dd33910b4e /lib/gitlab/database/batch_count.rb
parent1ce776de4ae122aba3f349c02c17cebeaa8ecf07 (diff)
downloadgitlab-ce-6e4e1050d9dba2b7b2523fdd1768823ab85feef4.tar.gz
Add latest changes from gitlab-org/gitlab@13-3-stable-ee
Diffstat (limited to 'lib/gitlab/database/batch_count.rb')
-rw-r--r--lib/gitlab/database/batch_count.rb23
1 files changed, 19 insertions, 4 deletions
diff --git a/lib/gitlab/database/batch_count.rb b/lib/gitlab/database/batch_count.rb
index ab069ce1da1..1762b81b7d8 100644
--- a/lib/gitlab/database/batch_count.rb
+++ b/lib/gitlab/database/batch_count.rb
@@ -16,6 +16,7 @@
# batch_count(::Clusters::Cluster.aws_installed.enabled, :cluster_id)
# batch_distinct_count(::Project, :creator_id)
# batch_distinct_count(::Project.with_active_services.service_desk_enabled.where(time_period), start: ::User.minimum(:id), finish: ::User.maximum(:id))
+# batch_sum(User, :sign_in_count)
module Gitlab
module Database
module BatchCount
@@ -27,6 +28,10 @@ module Gitlab
BatchCounter.new(relation, column: column).count(mode: :distinct, batch_size: batch_size, start: start, finish: finish)
end
+ def batch_sum(relation, column, batch_size: nil, start: nil, finish: nil)
+ BatchCounter.new(relation, column: nil, operation: :sum, operation_args: [column]).count(batch_size: batch_size, start: start, finish: finish)
+ end
+
class << self
include BatchCount
end
@@ -35,6 +40,7 @@ module Gitlab
class BatchCounter
FALLBACK = -1
MIN_REQUIRED_BATCH_SIZE = 1_250
+ DEFAULT_SUM_BATCH_SIZE = 1_000
MAX_ALLOWED_LOOPS = 10_000
SLEEP_TIME_IN_SECONDS = 0.01 # 10 msec sleep
ALLOWED_MODES = [:itself, :distinct].freeze
@@ -43,13 +49,16 @@ module Gitlab
DEFAULT_DISTINCT_BATCH_SIZE = 10_000
DEFAULT_BATCH_SIZE = 100_000
- def initialize(relation, column: nil)
+ def initialize(relation, column: nil, operation: :count, operation_args: nil)
@relation = relation
@column = column || relation.primary_key
+ @operation = operation
+ @operation_args = operation_args
end
def unwanted_configuration?(finish, batch_size, start)
- batch_size <= MIN_REQUIRED_BATCH_SIZE ||
+ (@operation == :count && batch_size <= MIN_REQUIRED_BATCH_SIZE) ||
+ (@operation == :sum && batch_size < DEFAULT_SUM_BATCH_SIZE) ||
(finish - start) / batch_size >= MAX_ALLOWED_LOOPS ||
start > finish
end
@@ -60,7 +69,7 @@ module Gitlab
check_mode!(mode)
# non-distinct have better performance
- batch_size ||= mode == :distinct ? DEFAULT_DISTINCT_BATCH_SIZE : DEFAULT_BATCH_SIZE
+ batch_size ||= batch_size_for_mode_and_operation(mode, @operation)
start = actual_start(start)
finish = actual_finish(finish)
@@ -91,11 +100,17 @@ module Gitlab
def batch_fetch(start, finish, mode)
# rubocop:disable GitlabSecurity/PublicSend
- @relation.select(@column).public_send(mode).where(between_condition(start, finish)).count
+ @relation.select(@column).public_send(mode).where(between_condition(start, finish)).send(@operation, *@operation_args)
end
private
+ def batch_size_for_mode_and_operation(mode, operation)
+ return DEFAULT_SUM_BATCH_SIZE if operation == :sum
+
+ mode == :distinct ? DEFAULT_DISTINCT_BATCH_SIZE : DEFAULT_BATCH_SIZE
+ end
+
def between_condition(start, finish)
return @column.between(start..(finish - 1)) if @column.is_a?(Arel::Attributes::Attribute)