Merge branch 'ml-update-gitaly-cluster-e2e-tests' into 'master'

Update Gitaly Cluster E2E tests

See merge request gitlab-org/gitlab!45734

qa/qa/service/praefect_manager.rb

@@ -206,14 +206,14 @@ module QA
 
       def wait_for_new_primary_node(node)
         QA::Runtime::Logger.info("Wait until #{node} is the primary node")
-        with_praefect_log do |log|
+        with_praefect_log(max_duration: 120) do |log|
           break true if log['msg'] == 'primary node changed' && log['newPrimary'] == node
         end
       end
 
       def wait_for_new_primary
         QA::Runtime::Logger.info("Wait until a new primary node is selected")
-        with_praefect_log do |log|
+        with_praefect_log(max_duration: 120) do |log|
           break true if log['msg'] == 'primary node changed'
         end
       end
@@ -406,8 +406,8 @@ module QA
         end
       end
 
-      def with_praefect_log
-        wait_until_shell_command("docker exec #{@praefect} bash -c 'tail -n 1 /var/log/gitlab/praefect/current'") do |line|
+      def with_praefect_log(**kwargs)
+        wait_until_shell_command("docker exec #{@praefect} bash -c 'tail -n 1 /var/log/gitlab/praefect/current'", **kwargs) do |line|
           QA::Runtime::Logger.debug(line.chomp)
           yield JSON.parse(line)
         end

qa/qa/specs/features/api/3_create/gitaly/automatic_failover_and_recovery_spec.rb

@@ -2,7 +2,7 @@
 
 module QA
   RSpec.describe 'Create' do
-    context 'Gitaly automatic failover and manual recovery', :orchestrated, :gitaly_cluster, quarantine: { issue: 'https://gitlab.com/gitlab-org/gitlab/-/issues/238953', type: :flaky } do
+    context 'Gitaly automatic failover and recovery', :orchestrated, :gitaly_cluster, quarantine: { issue: 'https://gitlab.com/gitlab-org/gitlab/-/issues/238953', type: :flaky } do
       # Variables shared between contexts. They're used and shared between
       # contexts so they can't be `let` variables.
       praefect_manager = Service::PraefectManager.new
@@ -66,17 +66,13 @@ module QA
       end
 
       context 'when recovering from dataloss after failover' do
-        it 'allows reconciliation', quarantine: { issue: 'https://gitlab.com/gitlab-org/gitlab/-/issues/238187', type: :stale }, testcase: 'https://gitlab.com/gitlab-org/quality/testcases/-/issues/977' do
+        it 'automatically reconciles', quarantine: { issue: 'https://gitlab.com/gitlab-org/gitlab/-/issues/238187', type: :stale }, testcase: 'https://gitlab.com/gitlab-org/quality/testcases/-/issues/977' do
           # Start the old primary node again
           praefect_manager.start_primary_node
           praefect_manager.wait_for_health_check_current_primary_node
 
-          # Confirm dataloss (i.e., inconsistent nodes)
-          expect(praefect_manager.replicated?(project.id)).to be false
-
-          # Reconcile nodes to recover from dataloss
-          praefect_manager.reconcile_nodes
-          praefect_manager.wait_for_replication(project.id)
+          # Confirm automatic reconciliation
+          expect(praefect_manager.replicated?(project.id)).to be true
 
           # Confirm that all commits are available after reconciliation
           expect(project.commits.map { |commit| commit[:message].chomp })