Commit af6752b0 authored by Kamil Trzciński's avatar Kamil Trzciński

Merge branch 'feature/gb/refactor-register-build-metrics' into 'master'

Refactor runner build request / registration metrics

See merge request gitlab-org/gitlab!54953
parents d4fd5bd2 cdfdf9cf
...@@ -6,15 +6,11 @@ module Ci ...@@ -6,15 +6,11 @@ module Ci
class RegisterJobService class RegisterJobService
attr_reader :runner attr_reader :runner
JOB_QUEUE_DURATION_SECONDS_BUCKETS = [1, 3, 10, 30, 60, 300, 900, 1800, 3600].freeze
JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET = 5.freeze
METRICS_SHARD_TAG_PREFIX = 'metrics_shard::'
DEFAULT_METRICS_SHARD = 'default'
Result = Struct.new(:build, :build_json, :valid?) Result = Struct.new(:build, :build_json, :valid?)
def initialize(runner) def initialize(runner)
@runner = runner @runner = runner
@metrics = ::Gitlab::Ci::Queue::Metrics.new(runner)
end end
# rubocop: disable CodeReuse/ActiveRecord # rubocop: disable CodeReuse/ActiveRecord
...@@ -48,7 +44,7 @@ module Ci ...@@ -48,7 +44,7 @@ module Ci
next unless result next unless result
if result.valid? if result.valid?
register_success(result.build) @metrics.register_success(result.build)
return result return result
else else
...@@ -58,7 +54,7 @@ module Ci ...@@ -58,7 +54,7 @@ module Ci
end end
end end
register_failure @metrics.register_failure
Result.new(nil, nil, valid) Result.new(nil, nil, valid)
end end
# rubocop: enable CodeReuse/ActiveRecord # rubocop: enable CodeReuse/ActiveRecord
...@@ -189,48 +185,6 @@ module Ci ...@@ -189,48 +185,6 @@ module Ci
builds builds
end end
def register_failure
failed_attempt_counter.increment
attempt_counter.increment
end
def register_success(job)
labels = { shared_runner: runner.instance_type?,
jobs_running_for_project: jobs_running_for_project(job),
shard: DEFAULT_METRICS_SHARD }
if runner.instance_type?
shard = runner.tag_list.sort.find { |name| name.starts_with?(METRICS_SHARD_TAG_PREFIX) }
labels[:shard] = shard.gsub(METRICS_SHARD_TAG_PREFIX, '') if shard
end
job_queue_duration_seconds.observe(labels, Time.current - job.queued_at) unless job.queued_at.nil?
attempt_counter.increment
end
# rubocop: disable CodeReuse/ActiveRecord
def jobs_running_for_project(job)
return '+Inf' unless runner.instance_type?
# excluding currently started job
running_jobs_count = job.project.builds.running.where(runner: Ci::Runner.instance_type)
.limit(JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET + 1).count - 1
running_jobs_count < JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET ? running_jobs_count : "#{JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET}+"
end
# rubocop: enable CodeReuse/ActiveRecord
def failed_attempt_counter
@failed_attempt_counter ||= Gitlab::Metrics.counter(:job_register_attempts_failed_total, "Counts the times a runner tries to register a job")
end
def attempt_counter
@attempt_counter ||= Gitlab::Metrics.counter(:job_register_attempts_total, "Counts the times a runner tries to register a job")
end
def job_queue_duration_seconds
@job_queue_duration_seconds ||= Gitlab::Metrics.histogram(:job_queue_duration_seconds, 'Request handling execution time', {}, JOB_QUEUE_DURATION_SECONDS_BUCKETS)
end
def pre_assign_runner_checks def pre_assign_runner_checks
{ {
missing_dependency_failure: -> (build, _) { !build.has_valid_build_dependencies? }, missing_dependency_failure: -> (build, _) { !build.has_valid_build_dependencies? },
......
# frozen_string_literal: true
module Gitlab
module Ci
module Queue
class Metrics
extend Gitlab::Utils::StrongMemoize
QUEUE_DURATION_SECONDS_BUCKETS = [1, 3, 10, 30, 60, 300, 900, 1800, 3600].freeze
METRICS_SHARD_TAG_PREFIX = 'metrics_shard::'
DEFAULT_METRICS_SHARD = 'default'
JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET = 5.freeze
attr_reader :runner
def initialize(runner)
@runner = runner
end
def register_failure
self.class.failed_attempt_counter.increment
self.class.attempt_counter.increment
end
def register_success(job)
labels = { shared_runner: runner.instance_type?,
jobs_running_for_project: jobs_running_for_project(job),
shard: DEFAULT_METRICS_SHARD }
if runner.instance_type?
shard = runner.tag_list.sort.find { |name| name.starts_with?(METRICS_SHARD_TAG_PREFIX) }
labels[:shard] = shard.gsub(METRICS_SHARD_TAG_PREFIX, '') if shard
end
self.class.job_queue_duration_seconds.observe(labels, Time.current - job.queued_at) unless job.queued_at.nil?
self.class.attempt_counter.increment
end
# rubocop: disable CodeReuse/ActiveRecord
def jobs_running_for_project(job)
return '+Inf' unless runner.instance_type?
# excluding currently started job
running_jobs_count = job.project.builds.running.where(runner: ::Ci::Runner.instance_type)
.limit(JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET + 1).count - 1
running_jobs_count < JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET ? running_jobs_count : "#{JOBS_RUNNING_FOR_PROJECT_MAX_BUCKET}+"
end
# rubocop: enable CodeReuse/ActiveRecord
def self.failed_attempt_counter
strong_memoize(:failed_attempt_counter) do
name = :job_register_attempts_failed_total
comment = 'Counts the times a runner tries to register a job'
Gitlab::Metrics.counter(name, comment)
end
end
def self.attempt_counter
strong_memoize(:attempt_counter) do
name = :job_register_attempts_total
comment = 'Counts the times a runner tries to register a job'
Gitlab::Metrics.counter(name, comment)
end
end
def self.job_queue_duration_seconds
strong_memoize(:job_queue_duration_seconds) do
name = :job_queue_duration_seconds
comment = 'Request handling execution time'
labels = {}
buckets = QUEUE_DURATION_SECONDS_BUCKETS
Gitlab::Metrics.histogram(name, comment, labels, buckets)
end
end
end
end
end
end
...@@ -590,22 +590,14 @@ module Ci ...@@ -590,22 +590,14 @@ module Ci
before do before do
allow(Time).to receive(:now).and_return(current_time) allow(Time).to receive(:now).and_return(current_time)
# Stub defaults for any metrics other than the ones we're testing
allow(Gitlab::Metrics).to receive(:counter)
.with(any_args)
.and_return(Gitlab::Metrics::NullMetric.instance)
allow(Gitlab::Metrics).to receive(:histogram)
.with(any_args)
.and_return(Gitlab::Metrics::NullMetric.instance)
# Stub tested metrics # Stub tested metrics
allow(Gitlab::Metrics).to receive(:counter) allow(Gitlab::Ci::Queue::Metrics)
.with(:job_register_attempts_total, anything) .to receive(:attempt_counter)
.and_return(attempt_counter) .and_return(attempt_counter)
allow(Gitlab::Metrics).to receive(:histogram)
.with(:job_queue_duration_seconds, anything, anything, anything) allow(Gitlab::Ci::Queue::Metrics)
.and_return(job_queue_duration_seconds) .to receive(:job_queue_duration_seconds)
.and_return(job_queue_duration_seconds)
project.update!(shared_runners_enabled: true) project.update!(shared_runners_enabled: true)
pending_job.update!(created_at: current_time - 3600, queued_at: current_time - 1800) pending_job.update!(created_at: current_time - 3600, queued_at: current_time - 1800)
...@@ -655,7 +647,7 @@ module Ci ...@@ -655,7 +647,7 @@ module Ci
context 'when shared runner is used' do context 'when shared runner is used' do
let(:runner) { create(:ci_runner, :instance, tag_list: %w(tag1 tag2)) } let(:runner) { create(:ci_runner, :instance, tag_list: %w(tag1 tag2)) }
let(:expected_shared_runner) { true } let(:expected_shared_runner) { true }
let(:expected_shard) { Ci::RegisterJobService::DEFAULT_METRICS_SHARD } let(:expected_shard) { ::Gitlab::Ci::Queue::Metrics::DEFAULT_METRICS_SHARD }
let(:expected_jobs_running_for_project_first_job) { 0 } let(:expected_jobs_running_for_project_first_job) { 0 }
let(:expected_jobs_running_for_project_third_job) { 2 } let(:expected_jobs_running_for_project_third_job) { 2 }
...@@ -694,7 +686,7 @@ module Ci ...@@ -694,7 +686,7 @@ module Ci
context 'when specific runner is used' do context 'when specific runner is used' do
let(:runner) { create(:ci_runner, :project, projects: [project], tag_list: %w(tag1 metrics_shard::shard_tag tag2)) } let(:runner) { create(:ci_runner, :project, projects: [project], tag_list: %w(tag1 metrics_shard::shard_tag tag2)) }
let(:expected_shared_runner) { false } let(:expected_shared_runner) { false }
let(:expected_shard) { Ci::RegisterJobService::DEFAULT_METRICS_SHARD } let(:expected_shard) { ::Gitlab::Ci::Queue::Metrics::DEFAULT_METRICS_SHARD }
let(:expected_jobs_running_for_project_first_job) { '+Inf' } let(:expected_jobs_running_for_project_first_job) { '+Inf' }
let(:expected_jobs_running_for_project_third_job) { '+Inf' } let(:expected_jobs_running_for_project_third_job) { '+Inf' }
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment