From 396a2bbf7521ab5045d0eeef5a7e9ae4b63db16d Mon Sep 17 00:00:00 2001
From: James Liu <jliu@gitlab.com>
Date: Mon, 21 Jul 2025 12:33:40 +1000
Subject: [PATCH 1/4] benchmark: Upgrade Python packages

The existing versions of PyYAML and cffi cause build issues on Fedora
42. Upgrade them, along with ansible and ansible-core as they're quite
outdtaed.
---
 _support/benchmarking/requirements.txt | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/_support/benchmarking/requirements.txt b/_support/benchmarking/requirements.txt
index 07d2c59e208..566d2dddd3e 100644
--- a/_support/benchmarking/requirements.txt
+++ b/_support/benchmarking/requirements.txt
@@ -1,10 +1,10 @@
-ansible==7.2.0
-ansible-core==2.14.2
-cffi==1.15.1
+ansible==11.8.0
+ansible-core==2.18.7
+cffi==1.17.1
 cryptography==39.0.1
 Jinja2==3.1.2
 MarkupSafe==2.1.2
 packaging==23.0
 pycparser==2.21
-PyYAML==6.0
+PyYAML==6.0.2
 resolvelib==0.8.1
-- 
GitLab


From f83a0dcd553426e3b59ff9abeaf78b5216690f40 Mon Sep 17 00:00:00 2001
From: James Liu <jliu@gitlab.com>
Date: Tue, 22 Jul 2025 09:29:39 +1000
Subject: [PATCH 2/4] benchmark: Update Python and Terraform versions

---
 _support/benchmarking/.tool-versions | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/_support/benchmarking/.tool-versions b/_support/benchmarking/.tool-versions
index db57b0f1554..b9911e7b29d 100644
--- a/_support/benchmarking/.tool-versions
+++ b/_support/benchmarking/.tool-versions
@@ -1,2 +1,2 @@
-terraform 1.3.9
-python 3.11.2
+terraform 1.12.2
+python 3.13.5
-- 
GitLab


From ba76cdf7c5d352cc2d36817b7d47d287a986985a Mon Sep 17 00:00:00 2001
From: James Liu <jliu@gitlab.com>
Date: Wed, 23 Jul 2025 10:42:18 +1000
Subject: [PATCH 3/4] benchmark: Install gitlab-shell as the `git` user

Git complains about ownership issues in the repository unless we first
switch to the `git` user, which cloned the gitlab-shell source
previously.
---
 _support/benchmarking/roles/gitaly/tasks/setup_gitlab_shell.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/_support/benchmarking/roles/gitaly/tasks/setup_gitlab_shell.yml b/_support/benchmarking/roles/gitaly/tasks/setup_gitlab_shell.yml
index 21143a729b7..3b752ccd88d 100644
--- a/_support/benchmarking/roles/gitaly/tasks/setup_gitlab_shell.yml
+++ b/_support/benchmarking/roles/gitaly/tasks/setup_gitlab_shell.yml
@@ -41,6 +41,7 @@
     chdir: /src/gitlab-shell
   environment:
     PATH: '/usr/local/go/bin:{{ ansible_env.PATH }}'
+  become_user: git
 
 - name: Create gitlab-shell secret
   file:
-- 
GitLab


From 8accdf5b35ae0161dc888cce91d4569b4e0cd91d Mon Sep 17 00:00:00 2001
From: James Liu <jliu@gitlab.com>
Date: Thu, 24 Jul 2025 12:22:21 +1000
Subject: [PATCH 4/4] benchmark: Add Terraform resources to create
 git-repositories disk

The existing documentation and Terraform plan reference a GCP disk image
called "git-repos", which contains a collection of cloned repos that are
used for benchmarking. The disk image appears to have been created
manually.

In order to simplify the process of recreating the disk image, and to
allow the list of repositories to be easily customised, add Terraform
resources to spin up a temporary GCP VM which clones a configurable list
of repositories onto a secondary disk. The secondary disk is turned into
a disk image which is later mounted onto the Gitaly host VM.
---
 _support/benchmarking/.gitignore            |  3 +
 _support/benchmarking/config.yml.example    |  8 +++
 _support/benchmarking/setup-repositories.sh | 30 +++++++++
 _support/benchmarking/terraform/main.tf     | 74 ++++++++++++++++++---
 4 files changed, 107 insertions(+), 8 deletions(-)
 create mode 100644 _support/benchmarking/setup-repositories.sh

diff --git a/_support/benchmarking/.gitignore b/_support/benchmarking/.gitignore
index 7b1864e6511..7d6f92c819f 100644
--- a/_support/benchmarking/.gitignore
+++ b/_support/benchmarking/.gitignore
@@ -2,6 +2,9 @@
 /terraform/*
 !/terraform/main.tf
 !/terraform/locals.tf
+/terraform-prepare/*
+!/terraform-prepare/main.tf
+!/terraform-prepare/locals.tf
 /group_vars/all.yml
 /results/*
 config.yml
diff --git a/_support/benchmarking/config.yml.example b/_support/benchmarking/config.yml.example
index 30ba48f2ec0..92a241799b8 100644
--- a/_support/benchmarking/config.yml.example
+++ b/_support/benchmarking/config.yml.example
@@ -13,3 +13,11 @@ benchmark_zone: "us-central1-a"
 use_regional_disk: false
 # One of these zones must be the zone for benchmark_zone
 regional_disk_replica_zones: ["us-central1-a", "us-central1-b"]
+
+# The list of repositories which will be cloned into Gitaly's storage disk.
+repositories:
+  - "https://gitlab.com/gitlab-org/git.git"
+  - "https://gitlab.com/gitlab-org/gitlab"
+  - "https://gitlab.com/linux-kernel/linux.git"
+  - "https://github.com/Homebrew/homebrew-core.git"
+  - "https://chromium.googlesource.com/chromium/src"
diff --git a/_support/benchmarking/setup-repositories.sh b/_support/benchmarking/setup-repositories.sh
new file mode 100644
index 00000000000..0db5ff942a6
--- /dev/null
+++ b/_support/benchmarking/setup-repositories.sh
@@ -0,0 +1,30 @@
+#!/bin/bash
+
+# This script is templated and will be executed when creating the
+# `repository-setup-vm` Terraform resource.
+
+set -euxo pipefail
+
+disk_name="/dev/disk/by-id/google-repositories"
+mountpoint="/mnt/disks/repositories"
+
+# Disks mounted to a GCP VM are prefixed with "google-"
+# https://cloud.google.com/compute/docs/disks/format-mount-disk-linux#format_linux
+#
+# TODO make this configurable so we can test other filesystems, like btrfs.
+mkfs.ext4 -m 0 -E lazy_itable_init=0,lazy_journal_init=0,discard "$${disk_name}"
+mkdir -p "$${mountpoint}"
+mount -o discard,defaults "$${disk_name}" "$${mountpoint}"
+chmod a+w "$${mountpoint}"
+
+echo "repositories disk successfully mounted as an ext4 volume at $${mountpoint}}"
+
+cd "$${mountpoint}"
+
+%{for repo_url in repositories}
+if ! git clone --bare "${repo_url}"; then
+echo "WARNING: Failed to clone ${repo_url}, continuing with other repositories" >&2
+fi
+%{endfor}
+
+shutdown -h now
diff --git a/_support/benchmarking/terraform/main.tf b/_support/benchmarking/terraform/main.tf
index 2dc52860003..7103feafe58 100644
--- a/_support/benchmarking/terraform/main.tf
+++ b/_support/benchmarking/terraform/main.tf
@@ -13,15 +13,77 @@ provider "google" {
   zone    = local.config.benchmark_zone
 }
 
-data "google_compute_disk" "repository-disk" {
-  name    = "git-repos"
-  project = local.config.project
+# Temporary disk from which we create the `git-repositories-<hash>` disk image from.
+resource "google_compute_disk" "prepare_repos" {
+  name = format("%s-prepare-repos-disk", var.gitaly_benchmarking_instance_name)
+  type = "pd-standard"
+  size = 100
+}
+
+# Temporary VM which clones and prepares the `git-repositories-<hash>` disk.
+resource "google_compute_instance" "prepare_repos" {
+  name         = format("%s-prepare-repos-vm", var.gitaly_benchmarking_instance_name)
+  machine_type = "n2d-standard-2"
+  zone         = local.config.benchmark_zone
+
+  boot_disk {
+    initialize_params {
+      image = local.config.os_image
+    }
+  }
+
+  attached_disk {
+    source      = google_compute_disk.prepare_repos.self_link
+    device_name = "repositories"
+  }
+
+  network_interface {
+    network = "default"
+    access_config {}
+  }
+
+  # If the script fails, you can SSH into the VM manually to debug the script:
+  # https://cloud.google.com/compute/docs/instances/startup-scripts/linux
+  metadata = {
+    startup-script = templatefile("${path.module}/../setup-repositories.sh", {
+      repositories = local.config.repositories
+    })
+  }
+}
+
+# Waits for the temporary setup VM to shut down, signalling the repos have been cloned.
+resource "null_resource" "prepare_repos_wait" {
+  provisioner "local-exec" {
+    command = <<-EOF
+      timeout 7200 bash -c '
+        while true; do
+          STATUS=$(gcloud compute instances describe ${google_compute_instance.prepare_repos.name} \
+            --zone=${google_compute_instance.prepare_repos.zone} \
+            --format="value(status)" \
+            --project=${local.config.project})
+          if [ "$STATUS" = "TERMINATED" ]; then
+            break
+          fi
+          sleep 30
+        done
+      '
+    EOF
+  }
+
+  depends_on = [google_compute_instance.prepare_repos]
+}
+
+resource "google_compute_image" "repos" {
+  name        = "git-repos"
+  source_disk = google_compute_disk.prepare_repos.self_link
+
+  depends_on = [null_resource.prepare_repos_wait]
 }
 
 resource "google_compute_disk" "repository-disk" {
   name  = format("%s-repository-disk", var.gitaly_benchmarking_instance_name)
   type  = local.config.repository_disk_type
-  image = format("projects/%s/global/images/git-repositories", local.config.project)
+  image = google_compute_image.repos.self_link
 }
 
 resource "google_compute_region_disk" "repository-region-disk" {
@@ -69,10 +131,6 @@ resource "google_compute_instance" "gitaly" {
   }
 
   tags = ["gitaly"]
-
-  lifecycle {
-    ignore_changes = [attached_disk]
-  }
 }
 
 resource "google_compute_instance" "client" {
-- 
GitLab