Commit 3986aa16 authored by Dmytro Zaporozhets's avatar Dmytro Zaporozhets

Merge branch...

Merge branch '218560-allow-generic-endpoint-to-receive-alerts-from-external-prometheus' into 'master'

Allow generic endpoint to receive alerts from external Prometheus

See merge request gitlab-org/gitlab!32676
parents e657bc52 ac284638
......@@ -29,12 +29,22 @@ module Projects
end
def notify_service
notify_service_class.new(project, current_user, notification_payload)
end
def notify_service_class
# We are tracking the consolidation of these services in
# https://gitlab.com/groups/gitlab-org/-/epics/3360
# to get rid of this workaround.
if Projects::Prometheus::Alerts::NotifyService.processable?(notification_payload)
Projects::Prometheus::Alerts::NotifyService
else
Projects::Alerting::NotifyService
.new(project, current_user, notification_payload)
end
end
def notification_payload
params.permit![:notification]
@notification_payload ||= params.permit![:notification]
end
end
end
......
......@@ -7,9 +7,19 @@ module Projects
include Gitlab::Utils::StrongMemoize
include IncidentManagement::Settings
# This set of keys identifies a payload as a valid Prometheus
# payload and thus processable by this service. See also
# https://prometheus.io/docs/alerting/configuration/#webhook_config
REQUIRED_PAYLOAD_KEYS = %w[
version groupKey status receiver groupLabels commonLabels
commonAnnotations externalURL alerts
].to_set.freeze
SUPPORTED_VERSION = '4'
def execute(token)
return bad_request unless valid_payload_size?
return unprocessable_entity unless valid_version?
return unprocessable_entity unless self.class.processable?(params)
return unauthorized unless valid_alert_manager_token?(token)
process_prometheus_alerts
......@@ -20,6 +30,14 @@ module Projects
ServiceResponse.success
end
def self.processable?(params)
# Workaround for https://gitlab.com/gitlab-org/gitlab/-/issues/220496
return false unless params
REQUIRED_PAYLOAD_KEYS.subset?(params.keys.to_set) &&
params['version'] == SUPPORTED_VERSION
end
private
def valid_payload_size?
......@@ -42,12 +60,10 @@ module Projects
params['alerts']
end
def valid_version?
params['version'] == '4'
end
def valid_alert_manager_token?(token)
valid_for_manual?(token) || valid_for_managed?(token)
valid_for_manual?(token) ||
valid_for_alerts_endpoint?(token) ||
valid_for_managed?(token)
end
def valid_for_manual?(token)
......@@ -61,6 +77,13 @@ module Projects
end
end
def valid_for_alerts_endpoint?(token)
return false unless project.alerts_service_activated?
# Here we are enforcing the existence of the token
compare_token(token, project.alerts_service.token)
end
def valid_for_managed?(token)
prometheus_application = available_prometheus_application(project)
return false unless prometheus_application
......
---
title: Allow generic endpoint to receive alerts from external Prometheus
merge_request: 32676
author:
type: added
......@@ -18,6 +18,9 @@ create an issue with the payload in the body of the issue. You can always
The entire payload will be posted in the issue discussion as a comment
authored by the GitLab Alert Bot.
NOTE: **Note**
In GitLab versions 13.1 and greater, you can configure [External Prometheus instances](prometheus.md#external-prometheus-instances) to use this endpoint.
## Setting up generic alerts
To set up the generic alerts integration:
......
......@@ -981,6 +981,9 @@ receivers:
In order for GitLab to associate your alerts with an [environment](../../../ci/environments/index.md), you need to configure a `gitlab_environment_name` label on the alerts you set up in Prometheus. The value of this should match the name of your Environment in GitLab.
NOTE: **Note**
In GitLab versions 13.1 and greater, you can configure your manually configured Prometheus server to use the [Generic alerts integration](generic_alerts.md).
### Taking action on incidents **(ULTIMATE)**
>- [Introduced](https://gitlab.com/gitlab-org/gitlab/-/issues/4925) in [GitLab Ultimate](https://about.gitlab.com/pricing/) 11.11.
......
......@@ -3,6 +3,8 @@
require 'spec_helper'
RSpec.describe Projects::Prometheus::Alerts::NotifyService do
include PrometheusHelpers
let_it_be(:project, reload: true) { create(:project) }
let(:service) { described_class.new(project, nil, payload) }
......@@ -94,7 +96,7 @@ RSpec.describe Projects::Prometheus::Alerts::NotifyService do
context 'with valid payload' do
let(:alert_firing) { create(:prometheus_alert, project: project) }
let(:alert_resolved) { create(:prometheus_alert, project: project) }
let(:payload_raw) { payload_for(firing: [alert_firing], resolved: [alert_resolved]) }
let(:payload_raw) { prometheus_alert_payload(firing: [alert_firing], resolved: [alert_resolved]) }
let(:payload) { ActionController::Parameters.new(payload_raw).permit! }
let(:payload_alert_firing) { payload_raw['alerts'].first }
let(:token) { 'token' }
......@@ -135,51 +137,4 @@ RSpec.describe Projects::Prometheus::Alerts::NotifyService do
end
end
end
private
def payload_for(firing: [], resolved: [])
status = firing.any? ? 'firing' : 'resolved'
alerts = firing + resolved
alert_name = alerts.first.title
prometheus_metric_id = alerts.first.prometheus_metric_id.to_s
alerts_map = \
firing.map { |alert| map_alert_payload('firing', alert) } +
resolved.map { |alert| map_alert_payload('resolved', alert) }
# See https://prometheus.io/docs/alerting/configuration/#%3Cwebhook_config%3E
{
'version' => '4',
'receiver' => 'gitlab',
'status' => status,
'alerts' => alerts_map,
'groupLabels' => {
'alertname' => alert_name
},
'commonLabels' => {
'alertname' => alert_name,
'gitlab' => 'hook',
'gitlab_alert_id' => prometheus_metric_id
},
'commonAnnotations' => {},
'externalURL' => '',
'groupKey' => "{}:{alertname=\'#{alert_name}\'}"
}
end
def map_alert_payload(status, alert)
{
'status' => status,
'labels' => {
'alertname' => alert.title,
'gitlab' => 'hook',
'gitlab_alert_id' => alert.prometheus_metric_id.to_s
},
'annotations' => {},
'startsAt' => '2018-09-24T08:57:31.095725221Z',
'endsAt' => '0001-01-01T00:00:00Z',
'generatorURL' => 'http://prometheus-prometheus-server-URL'
}
end
end
......@@ -7,29 +7,23 @@ RSpec.describe Projects::Alerting::NotificationsController do
let_it_be(:environment) { create(:environment, project: project) }
describe 'POST #create' do
let(:service_response) { ServiceResponse.success }
let(:notify_service) { instance_double(Projects::Alerting::NotifyService, execute: service_response) }
around do |example|
ForgeryProtection.with_forgery_protection { example.run }
end
shared_examples 'process alert payload' do |notify_service_class|
let(:service_response) { ServiceResponse.success }
let(:notify_service) { instance_double(notify_service_class, execute: service_response) }
before do
allow(Projects::Alerting::NotifyService).to receive(:new).and_return(notify_service)
allow(notify_service_class).to receive(:new).and_return(notify_service)
end
def make_request(body = {})
post :create, params: project_params, body: body.to_json, as: :json
def make_request
post :create, params: project_params, body: payload.to_json, as: :json
end
context 'when notification service succeeds' do
let(:payload) do
{
title: 'Alert title',
hosts: 'https://gitlab.com'
}
end
let(:permitted_params) { ActionController::Parameters.new(payload).permit! }
it 'responds with ok' do
......@@ -39,9 +33,9 @@ RSpec.describe Projects::Alerting::NotificationsController do
end
it 'does not pass excluded parameters to the notify service' do
make_request(payload)
make_request
expect(Projects::Alerting::NotifyService)
expect(notify_service_class)
.to have_received(:new)
.with(project, nil, permitted_params)
end
......@@ -86,6 +80,23 @@ RSpec.describe Projects::Alerting::NotificationsController do
end
end
context 'generic alert payload' do
it_behaves_like 'process alert payload', Projects::Alerting::NotifyService do
let(:payload) { { title: 'Alert title' } }
end
end
context 'Prometheus alert payload' do
include PrometheusHelpers
it_behaves_like 'process alert payload', Projects::Prometheus::Alerts::NotifyService do
let(:payload) { prometheus_alert_payload }
end
end
end
private
def project_params(opts = {})
opts.reverse_merge(namespace_id: project.namespace, project_id: project)
end
......
......@@ -45,9 +45,13 @@ FactoryBot.define do
end
factory :alerts_service do
active
project
type { 'AlertsService' }
trait :active do
active { true }
end
trait :inactive do
active { false }
......
......@@ -3,6 +3,8 @@
require 'spec_helper'
describe Projects::Prometheus::Alerts::NotifyService do
include PrometheusHelpers
let_it_be(:project, reload: true) { create(:project) }
let(:service) { described_class.new(project, nil, payload) }
......@@ -92,9 +94,10 @@ describe Projects::Prometheus::Alerts::NotifyService do
end
context 'with valid payload' do
let(:alert_firing) { create(:prometheus_alert, project: project) }
let(:alert_resolved) { create(:prometheus_alert, project: project) }
let(:payload_raw) { payload_for(firing: [alert_firing], resolved: [alert_resolved]) }
let_it_be(:alert_firing) { create(:prometheus_alert, project: project) }
let_it_be(:alert_resolved) { create(:prometheus_alert, project: project) }
let_it_be(:cluster) { create(:cluster, :provided_by_user, projects: [project]) }
let(:payload_raw) { prometheus_alert_payload(firing: [alert_firing], resolved: [alert_resolved]) }
let(:payload) { ActionController::Parameters.new(payload_raw).permit! }
let(:payload_alert_firing) { payload_raw['alerts'].first }
let(:token) { 'token' }
......@@ -116,9 +119,7 @@ describe Projects::Prometheus::Alerts::NotifyService do
with_them do
before do
cluster = create(:cluster, :provided_by_user,
projects: [project],
enabled: cluster_enabled)
cluster.update!(enabled: cluster_enabled)
if status
create(:clusters_applications_prometheus, status,
......@@ -179,6 +180,39 @@ describe Projects::Prometheus::Alerts::NotifyService do
end
end
context 'with generic alerts integration' do
using RSpec::Parameterized::TableSyntax
where(:alerts_service, :token, :result) do
:active | :valid | :success
:active | :invalid | :failure
:active | nil | :failure
:inactive | :valid | :failure
nil | nil | :failure
end
with_them do
let(:valid) { project.alerts_service.token }
let(:invalid) { 'invalid token' }
let(:token_input) { public_send(token) if token }
before do
if alerts_service
create(:alerts_service, alerts_service, project: project)
end
end
case result = params[:result]
when :success
it_behaves_like 'notifies alerts'
when :failure
it_behaves_like 'no notifications', http_status: :unauthorized
else
raise "invalid result: #{result.inspect}"
end
end
end
context 'alert emails' do
before do
create(:prometheus_service, project: project)
......@@ -227,7 +261,7 @@ describe Projects::Prometheus::Alerts::NotifyService do
context 'with multiple firing alerts and resolving alerts' do
let(:payload_raw) do
payload_for(firing: [alert_firing, alert_firing], resolved: [alert_resolved])
prometheus_alert_payload(firing: [alert_firing, alert_firing], resolved: [alert_resolved])
end
it 'processes Prometheus alerts' do
......@@ -258,7 +292,7 @@ describe Projects::Prometheus::Alerts::NotifyService do
context 'multiple firing alerts' do
let(:payload_raw) do
payload_for(firing: [alert_firing, alert_firing], resolved: [])
prometheus_alert_payload(firing: [alert_firing, alert_firing], resolved: [])
end
it_behaves_like 'processes incident issues', 2
......@@ -266,7 +300,7 @@ describe Projects::Prometheus::Alerts::NotifyService do
context 'without firing alerts' do
let(:payload_raw) do
payload_for(firing: [], resolved: [alert_resolved])
prometheus_alert_payload(firing: [], resolved: [alert_resolved])
end
it_behaves_like 'processes incident issues', 1
......@@ -284,24 +318,17 @@ describe Projects::Prometheus::Alerts::NotifyService do
end
context 'with invalid payload' do
context 'without version' do
context 'when payload is not processable' do
let(:payload) { {} }
it_behaves_like 'no notifications', http_status: :unprocessable_entity
before do
allow(described_class).to receive(:processable?).with(payload)
.and_return(false)
end
context 'when version is not "4"' do
let(:payload) { { 'version' => '5' } }
it_behaves_like 'no notifications', http_status: :unprocessable_entity
end
context 'with missing alerts' do
let(:payload) { { 'version' => '4' } }
it_behaves_like 'no notifications', http_status: :unauthorized
end
context 'when the payload is too big' do
let(:payload) { { 'the-payload-is-too-big' => true } }
let(:deep_size_object) { instance_double(Gitlab::Utils::DeepSize, valid?: false) }
......@@ -328,50 +355,39 @@ describe Projects::Prometheus::Alerts::NotifyService do
end
end
private
def payload_for(firing: [], resolved: [])
status = firing.any? ? 'firing' : 'resolved'
alerts = firing + resolved
alert_name = alerts.first.title
prometheus_metric_id = alerts.first.prometheus_metric_id.to_s
alerts_map = \
firing.map { |alert| map_alert_payload('firing', alert) } +
resolved.map { |alert| map_alert_payload('resolved', alert) }
# See https://prometheus.io/docs/alerting/configuration/#%3Cwebhook_config%3E
{
'version' => '4',
'receiver' => 'gitlab',
'status' => status,
'alerts' => alerts_map,
'groupLabels' => {
'alertname' => alert_name
},
'commonLabels' => {
'alertname' => alert_name,
'gitlab' => 'hook',
'gitlab_alert_id' => prometheus_metric_id
},
'commonAnnotations' => {},
'externalURL' => '',
'groupKey' => "{}:{alertname=\'#{alert_name}\'}"
}
end
def map_alert_payload(status, alert)
{
'status' => status,
'labels' => {
'alertname' => alert.title,
'gitlab' => 'hook',
'gitlab_alert_id' => alert.prometheus_metric_id.to_s
},
'annotations' => {},
'startsAt' => '2018-09-24T08:57:31.095725221Z',
'endsAt' => '0001-01-01T00:00:00Z',
'generatorURL' => 'http://prometheus-prometheus-server-URL'
}
describe '.processable?' do
let(:valid_payload) { prometheus_alert_payload }
subject { described_class.processable?(payload) }
context 'with valid payload' do
let(:payload) { valid_payload }
it { is_expected.to eq(true) }
context 'containing unrelated keys' do
let(:payload) { valid_payload.merge('unrelated' => 'key') }
it { is_expected.to eq(true) }
end
end
context 'with invalid payload' do
where(:missing_key) do
described_class::REQUIRED_PAYLOAD_KEYS.to_a
end
with_them do
let(:payload) { valid_payload.except(missing_key) }
it { is_expected.to eq(false) }
end
end
context 'with unsupported version' do
let(:payload) { valid_payload.merge('version' => '5') }
it { is_expected.to eq(false) }
end
end
end
......@@ -236,4 +236,51 @@ module PrometheusHelpers
]
}
end
def prometheus_alert_payload(firing: [], resolved: [])
status = firing.any? ? 'firing' : 'resolved'
alerts = firing + resolved
alert_name = alerts.first&.title || ''
prometheus_metric_id = alerts.first&.prometheus_metric_id&.to_s
alerts_map = \
firing.map { |alert| prometheus_map_alert_payload('firing', alert) } +
resolved.map { |alert| prometheus_map_alert_payload('resolved', alert) }
# See https://prometheus.io/docs/alerting/configuration/#%3Cwebhook_config%3E
{
'version' => '4',
'receiver' => 'gitlab',
'status' => status,
'alerts' => alerts_map,
'groupLabels' => {
'alertname' => alert_name
},
'commonLabels' => {
'alertname' => alert_name,
'gitlab' => 'hook',
'gitlab_alert_id' => prometheus_metric_id
},
'commonAnnotations' => {},
'externalURL' => '',
'groupKey' => "{}:{alertname=\'#{alert_name}\'}"
}
end
private
def prometheus_map_alert_payload(status, alert)
{
'status' => status,
'labels' => {
'alertname' => alert.title,
'gitlab' => 'hook',
'gitlab_alert_id' => alert.prometheus_metric_id.to_s
},
'annotations' => {},
'startsAt' => '2018-09-24T08:57:31.095725221Z',
'endsAt' => '0001-01-01T00:00:00Z',
'generatorURL' => 'http://prometheus-prometheus-server-URL'
}
end
end
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