Merge branch 'add-new-arg-to-git-rev-list-call' into 'master'

Pass the --in-commit-order arg to `git-rev-list`

Closes #45062

See merge request gitlab-org/gitlab-ce!19494

3 files changed, 28 insertions, 6 deletions

diff --git a/changelogs/unreleased/add-new-arg-to-git-rev-list-call.yml b/changelogs/unreleased/add-new-arg-to-git-rev-list-call.yml
new file mode 100644
index 00000000000..86680b6b117
--- /dev/null
+++ b/changelogs/unreleased/add-new-arg-to-git-rev-list-call.yml
@@ -0,0 +1,5 @@
+---
+title: Improve performance of LFS integrity check
+merge_request: 19494
+author:
+type: performance
diff --git a/lib/gitlab/git/lfs_changes.rb b/lib/gitlab/git/lfs_changes.rb
index b230289e7bf..f3cc388ea41 100644
--- a/lib/gitlab/git/lfs_changes.rb
+++ b/lib/gitlab/git/lfs_changes.rb
@@ -30,7 +30,7 @@ module Gitlab
 
       def git_new_pointers(object_limit, not_in)
         @new_pointers ||= begin
-          rev_list.new_objects(not_in: not_in, require_path: true) do |object_ids|
+          rev_list.new_objects(rev_list_params(not_in: not_in)) do |object_ids|
             object_ids = object_ids.take(object_limit) if object_limit
 
             Gitlab::Git::Blob.batch_lfs_pointers(@repository, object_ids)
@@ -39,13 +39,12 @@ module Gitlab
       end
 
       def git_all_pointers
-        params = { require_path: true }
-
-        if Gitlab::Git.version >= Gitlab::VersionInfo.parse('2.16.0')
+        params = {}
+        if rev_list_supports_new_options?
           params[:options] = ["--filter=blob:limit=#{Gitlab::Git::Blob::LFS_POINTER_MAX_SIZE}"]
         end
 
-        rev_list.all_objects(params) do |object_ids|
+        rev_list.all_objects(rev_list_params(params)) do |object_ids|
           Gitlab::Git::Blob.batch_lfs_pointers(@repository, object_ids)
         end
       end
@@ -53,6 +52,23 @@ module Gitlab
       def rev_list
         Gitlab::Git::RevList.new(@repository, newrev: @newrev)
       end
+
+      # We're passing the `--in-commit-order` arg to ensure we don't wait
+      # for git to traverse all commits before returning pointers.
+      # This is required in order to improve the performance of LFS integrity check
+      def rev_list_params(params = {})
+        params[:options] ||= []
+        params[:options] << "--in-commit-order" if rev_list_supports_new_options?
+        params[:require_path] = true
+
+        params
+      end
+
+      def rev_list_supports_new_options?
+        return @option_supported if defined?(@option_supported)
+
+        @option_supported = Gitlab::Git.version >= Gitlab::VersionInfo.parse('2.16.0')
+      end
     end
   end
 end
diff --git a/lib/gitlab/git/rev_list.rb b/lib/gitlab/git/rev_list.rb
index 33b641a090b..4e661eceffb 100644
--- a/lib/gitlab/git/rev_list.rb
+++ b/lib/gitlab/git/rev_list.rb
@@ -27,9 +27,10 @@ module Gitlab
       #
       # When given a block it will yield objects as a lazy enumerator so
       # the caller can limit work done instead of processing megabytes of data
-      def new_objects(require_path: nil, not_in: nil, &lazy_block)
+      def new_objects(options: [], require_path: nil, not_in: nil, &lazy_block)
         opts = {
           including: newrev,
+          options: options,
           excluding: not_in.nil? ? :all : not_in,
           require_path: require_path
         }