Merge branch '235161_send_sql_queries_to_sentry' into 'master'

Send sql queries to sentry

See merge request gitlab-org/gitlab!45724

Gemfile

@@ -307,6 +307,9 @@ gem 'rack-attack', '~> 6.3.0'
 # Sentry integration
 gem 'sentry-raven', '~> 3.0'
 
+# PostgreSQL query parsing
+gem 'pg_query', '~> 1.2'
+
 gem 'premailer-rails', '~> 1.10.3'
 
 # LabKit: Tracing and Correlation

Gemfile.lock

@@ -828,6 +828,7 @@ GEM
     peek (1.1.0)
       railties (>= 4.0.0)
     pg (1.2.3)
+    pg_query (1.2.0)
     png_quantizator (0.2.1)
     po_to_json (1.0.1)
       json (>= 1.6.0)
@@ -1424,6 +1425,7 @@ DEPENDENCIES
   parallel (~> 1.19)
   peek (~> 1.1)
   pg (~> 1.1)
+  pg_query (~> 1.2)
   png_quantizator (~> 0.2.1)
   premailer-rails (~> 1.10.3)
   prometheus-client-mmap (~> 0.12.0)

lib/gitlab/error_tracking.rb

@@ -123,6 +123,7 @@ module Gitlab
         end
 
         extra = sanitize_request_parameters(extra)
+        inject_sql_query_into_extra(exception, extra)
 
         if sentry && Raven.configuration.server
           Raven.capture_exception(exception, tags: default_tags, extra: extra)
@@ -149,6 +150,12 @@ module Gitlab
         filter.filter(parameters)
       end
 
+      def inject_sql_query_into_extra(exception, extra)
+        return unless exception.is_a?(ActiveRecord::StatementInvalid)
+
+        extra[:sql] = PgQuery.normalize(exception.sql.to_s)
+      end
+
       def sentry_dsn
         return unless Rails.env.production? || Rails.env.development?
         return unless Gitlab.config.sentry.enabled

spec/lib/gitlab/error_tracking_spec.rb

@@ -198,47 +198,39 @@ RSpec.describe Gitlab::ErrorTracking do
   end
 
   describe '.track_exception' do
-    it 'calls Raven.capture_exception' do
-      expected_extras = {
-        some_other_info: 'info',
-        issue_url: issue_url
-      }
+    let(:extra) { { issue_url: issue_url, some_other_info: 'info' } }
 
-      expected_tags = {
-        correlation_id: 'cid'
-      }
+    subject(:track_exception) { described_class.track_exception(exception, extra) }
 
-      expect(Raven).to receive(:capture_exception)
-                         .with(exception,
-                          tags: a_hash_including(expected_tags),
-                          extra: a_hash_including(expected_extras))
-
-      described_class.track_exception(
-        exception,
-        issue_url: issue_url,
-        some_other_info: 'info'
-      )
+    before do
+      allow(Raven).to receive(:capture_exception).and_call_original
+      allow(Gitlab::ErrorTracking::Logger).to receive(:error)
+    end
+
+    it 'calls Raven.capture_exception' do
+      track_exception
+
+      expect(Raven).to have_received(:capture_exception)
+                   .with(exception,
+                         tags: a_hash_including(correlation_id: 'cid'),
+                         extra: a_hash_including(some_other_info: 'info', issue_url: issue_url))
     end
 
     it 'calls Gitlab::ErrorTracking::Logger.error with formatted payload' do
-      expect(Gitlab::ErrorTracking::Logger).to receive(:error)
-        .with(a_hash_including(*expected_payload_includes))
+      track_exception
 
-      described_class.track_exception(
-        exception,
-        issue_url: issue_url,
-        some_other_info: 'info'
-      )
+      expect(Gitlab::ErrorTracking::Logger).to have_received(:error)
+                                           .with(a_hash_including(*expected_payload_includes))
     end
 
     context 'with filterable parameters' do
       let(:extra) { { test: 1, my_token: 'test' } }
 
       it 'filters parameters' do
-        expect(Gitlab::ErrorTracking::Logger).to receive(:error).with(
-          hash_including({ 'extra.test' => 1, 'extra.my_token' => '[FILTERED]' }))
+        track_exception
 
-        described_class.track_exception(exception, extra)
+        expect(Gitlab::ErrorTracking::Logger).to have_received(:error)
+                                             .with(hash_including({ 'extra.test' => 1, 'extra.my_token' => '[FILTERED]' }))
       end
     end
 
@@ -247,44 +239,58 @@ RSpec.describe Gitlab::ErrorTracking do
       let(:exception) { double(message: 'bang!', sentry_extra_data: extra_info, backtrace: caller) }
 
       it 'includes the extra data from the exception in the tracking information' do
-        expect(Raven).to receive(:capture_exception)
-          .with(exception, a_hash_including(extra: a_hash_including(extra_info)))
+        track_exception
 
-        described_class.track_exception(exception)
+        expect(Raven).to have_received(:capture_exception)
+                     .with(exception, a_hash_including(extra: a_hash_including(extra_info)))
       end
     end
 
     context 'the exception implements :sentry_extra_data, which returns nil' do
       let(:exception) { double(message: 'bang!', sentry_extra_data: nil, backtrace: caller) }
+      let(:extra) { { issue_url: issue_url } }
 
       it 'just includes the other extra info' do
-        extra_info = { issue_url: issue_url }
-        expect(Raven).to receive(:capture_exception)
-          .with(exception, a_hash_including(extra: a_hash_including(extra_info)))
+        track_exception
 
-        described_class.track_exception(exception, extra_info)
+        expect(Raven).to have_received(:capture_exception)
+                     .with(exception, a_hash_including(extra: a_hash_including(extra)))
       end
     end
 
     context 'with sidekiq args' do
-      it 'ensures extra.sidekiq.args is a string' do
-        extra = { sidekiq: { 'class' => 'PostReceive', 'args' => [1, { 'id' => 2, 'name' => 'hello' }, 'some-value', 'another-value'] } }
+      context 'when the args does not have anything sensitive' do
+        let(:extra) { { sidekiq: { 'class' => 'PostReceive', 'args' => [1, { 'id' => 2, 'name' => 'hello' }, 'some-value', 'another-value'] } } }
 
-        expect(Gitlab::ErrorTracking::Logger).to receive(:error).with(
-          hash_including({ 'extra.sidekiq' => { 'class' => 'PostReceive', 'args' => ['1', '{"id"=>2, "name"=>"hello"}', 'some-value', 'another-value'] } }))
+        it 'ensures extra.sidekiq.args is a string' do
+          track_exception
 
-        described_class.track_exception(exception, extra)
+          expect(Gitlab::ErrorTracking::Logger).to have_received(:error).with(
+            hash_including({ 'extra.sidekiq' => { 'class' => 'PostReceive', 'args' => ['1', '{"id"=>2, "name"=>"hello"}', 'some-value', 'another-value'] } }))
+        end
       end
 
-      it 'filters sensitive arguments before sending' do
-        extra = { sidekiq: { 'class' => 'UnknownWorker', 'args' => ['sensitive string', 1, 2] } }
+      context 'when the args has sensitive information' do
+        let(:extra) { { sidekiq: { 'class' => 'UnknownWorker', 'args' => ['sensitive string', 1, 2] } } }
+
+        it 'filters sensitive arguments before sending' do
+          track_exception
+
+          expect(sentry_event.dig('extra', 'sidekiq', 'args')).to eq(['[FILTERED]', 1, 2])
+          expect(Gitlab::ErrorTracking::Logger).to have_received(:error).with(
+            hash_including('extra.sidekiq' => { 'class' => 'UnknownWorker', 'args' => ['[FILTERED]', '1', '2'] }))
+        end
+      end
+    end
 
-        expect(Gitlab::ErrorTracking::Logger).to receive(:error).with(
-          hash_including('extra.sidekiq' => { 'class' => 'UnknownWorker', 'args' => ['[FILTERED]', '1', '2'] }))
+    context 'when the error is kind of an `ActiveRecord::StatementInvalid`' do
+      let(:exception) { ActiveRecord::StatementInvalid.new(sql: 'SELECT "users".* FROM "users" WHERE "users"."id" = 1 AND "users"."foo" = $1') }
 
-        described_class.track_exception(exception, extra)
+      it 'injects the normalized sql query into extra' do
+        track_exception
 
-        expect(sentry_event.dig('extra', 'sidekiq', 'args')).to eq(['[FILTERED]', 1, 2])
+        expect(Raven).to have_received(:capture_exception)
+          .with(exception, a_hash_including(extra: a_hash_including(sql: 'SELECT "users".* FROM "users" WHERE "users"."id" = $2 AND "users"."foo" = $1')))
       end
     end
   end