Add ExclusiveLease guards for RepositoryCheck::{DispatchWorker,BatchWorker}

We saw in production that DispatchWorker was running about twice an hour,
which would schedule twice as many jobs as it should.

For some reason, BatchWorker was running 1000 times per hour, possibly
due to Sidekiq RSS kills that caused these jobs to restart.

Adding an ExclusiveLease prevents these jobs from running more
than they should.

Relates to https://gitlab.com/gitlab-com/infrastructure/issues/4526

2 files changed, 29 insertions, 4 deletions

diff --git a/app/workers/repository_check/batch_worker.rb b/app/workers/repository_check/batch_worker.rb
index 051382a08a9..07559ea479b 100644
--- a/app/workers/repository_check/batch_worker.rb
+++ b/app/workers/repository_check/batch_worker.rb
@@ -4,9 +4,11 @@ module RepositoryCheck
   class BatchWorker
     include ApplicationWorker
     include RepositoryCheckQueue
+    include ExclusiveLeaseGuard
 
     RUN_TIME = 3600
     BATCH_SIZE = 10_000
+    LEASE_TIMEOUT = 1.hour
 
     attr_reader :shard_name
 
@@ -16,6 +18,20 @@ module RepositoryCheck
       return unless Gitlab::CurrentSettings.repository_checks_enabled
       return unless Gitlab::ShardHealthCache.healthy_shard?(shard_name)
 
+      try_obtain_lease do
+        perform_repository_checks
+      end
+    end
+
+    def lease_timeout
+      LEASE_TIMEOUT
+    end
+
+    def lease_key
+      "repository_check_batch_worker:#{shard_name}"
+    end
+
+    def perform_repository_checks
       start = Time.now
 
       # This loop will break after a little more than one hour ('a little
@@ -26,7 +42,7 @@ module RepositoryCheck
       project_ids.each do |project_id|
         break if Time.now - start >= RUN_TIME
 
-        next unless try_obtain_lease(project_id)
+        next unless try_obtain_lease_for_project(project_id)
 
         SingleRepositoryWorker.new.perform(project_id)
       end
@@ -60,7 +76,7 @@ module RepositoryCheck
       Project.where(repository_storage: shard_name)
     end
 
-    def try_obtain_lease(id)
+    def try_obtain_lease_for_project(id)
       # Use a 24-hour timeout because on servers/projects where 'git fsck' is
       # super slow we definitely do not want to run it twice in parallel.
       Gitlab::ExclusiveLease.new(
diff --git a/app/workers/repository_check/dispatch_worker.rb b/app/workers/repository_check/dispatch_worker.rb
index 891a273afd7..96634f09a15 100644
--- a/app/workers/repository_check/dispatch_worker.rb
+++ b/app/workers/repository_check/dispatch_worker.rb
@@ -3,13 +3,22 @@ module RepositoryCheck
     include ApplicationWorker
     include CronjobQueue
     include ::EachShardWorker
+    include ExclusiveLeaseGuard
+
+    LEASE_TIMEOUT = 1.hour
 
     def perform
       return unless Gitlab::CurrentSettings.repository_checks_enabled
 
-      each_eligible_shard do |shard_name|
-        RepositoryCheck::BatchWorker.perform_async(shard_name)
+      try_obtain_lease do
+        each_eligible_shard do |shard_name|
+          RepositoryCheck::BatchWorker.perform_async(shard_name)
+        end
       end
     end
+
+    def lease_timeout
+      LEASE_TIMEOUT
+    end
   end
 end