GitLab Runner搭配VirtualBox执行器无法处理并发作业的问题求助

我需要通过预配置的Windows 10基础镜像，使用VirtualBox执行器运行2个及以上并发作业。VirtualBox会创建该基础镜像的两个实例（xyz-concurrent-0和xyz-concurrent-1）并同时启动，但第一个作业运行正常，第二个作业始终失败，报错如下：

正常作业日志（第一个启动的作业） #

Running with gitlab-runner 16.10.0 (81ab07f6)
  on Runner_1 93V7Dm8Bk, system ID: s_a1f2f83ac892
Preparing the "virtualbox" executor
01:55
Using VirtualBox version 7.0.14r161095 executor...
Creating new VM...
Creating default snapshot...
Waiting for VM to become responsive...
Starting SSH command...
Preparing environment
00:02
Running on WIN10 via 
EVH11716WS...
Getting source from Git repository
00:12
Fetching changes with git depth set to 20...
Initialized empty Git repository in C:/Users/gitlab-user/builds/myname/roli-dummy-project/.git/
Created fresh repository.
warning: auto-detection of host provider took too long (>2000ms)
warning: see https://aka.ms/gcm/autodetect for more information.
fatal: Unable to persist credentials with the 'wincredman' credential store.
See https://aka.ms/gcm/credstores for more information.
Checking out 6cce2e68 as detached HEAD (ref is main)...
git-lfs/3.4.1 (GitHub; windows amd64; go 1.20.11; git 0898dcbc)
Skipping Git submodules setup
Executing "step_script" stage of the job script
00:32
$ powershell -File echo.ps1
WIN10
WIN10
WIN10
Cleaning up project directory and file based variables
00:01
Job succeeded

失败作业日志（第二个启动的作业） #

Running with gitlab-runner 16.10.0 (81ab07f6)
  on Runner_1 93V7Dm8Bk, system ID: s_a1f2f83ac892
Preparing the "virtualbox" executor
00:25
Using VirtualBox version 7.0.14r161095 executor...
Creating new VM...
ERROR: Preparation failed: VBoxManageOutput error: vboxmanage.exe: error: Locking of attached media failed. A possible reason is that one of the media is attached to a running VM
vboxmanage.exe: error: Details: code VBOX_E_INVALID_OBJECT_STATE (0x80bb0007), component SessionMachine, interface IMachine
Will be retried in 3s ...
Using VirtualBox version 7.0.14r161095 executor...
Restoring VM from snapshot...
Previous VM failed. Deleting, because VBoxManageOutput error: 
Creating new VM...
ERROR: Preparation failed: VBoxManageOutput error: vboxmanage.exe: error: Locking of attached media failed. A possible reason is that one of the media is attached to a running VM
vboxmanage.exe: error: Details: code VBOX_E_INVALID_OBJECT_STATE (0x80bb0007), component SessionMachine, interface IMachine
Will be retried in 3s ...
Using VirtualBox version 7.0.14r161095 executor...
Restoring VM from snapshot...
Previous VM failed. Deleting, because VBoxManageOutput error: 
Creating new VM...
ERROR: Preparation failed: VBoxManageOutput error: vboxmanage.exe: error: Locking of attached media failed. A possible reason is that one of the media is attached to a running VM
vboxmanage.exe: error: Details: code VBOX_E_INVALID_OBJECT_STATE (0x80bb0007), component SessionMachine, interface IMachine
Will be retried in 3s ...
ERROR: Job failed (system failure): VBoxManageOutput error: vboxmanage.exe: error: Locking of attached media failed. A possible reason is that one of the media is attached to a running VM
vboxmanage.exe: error: Details: code VBOX_E_INVALID_OBJECT_STATE (0x80bb0007), component SessionMachine, interface IMachine

已尝试的配置 #

配置1：单个Runner允许2个并发作业 #

concurrent = 2
check_interval = 0
shutdown_timeout = 0

[session_server]
  session_timeout = 1800

[[runners]]
  name = "Runner_1"
  url = "myurl"
  id = 1
  token = "mytoken"
  token_obtained_at = 2024-04-19T10:36:11Z
  token_expires_at = 0001-01-01T00:00:00Z
  executor = "virtualbox"
  shell = "powershell"
  [runners.cache]
    MaxUploadedArchiveSize = 0
  [runners.ssh]
    user = "gitlab-user"
    identity_file = "C:\\Users\\myuser\\.ssh\\id_ed25519"
    disable_strict_host_key_checking = true
  [runners.virtualbox]
    base_name = "Win10_prepare"
    base_folder = ""
    disable_snapshots = false
    start_type = ""

配置2：两个Runner各允许2个并发作业 #

concurrent = 2
check_interval = 0
shutdown_timeout = 0

[session_server]
  session_timeout = 1800

[[runners]]
  name = "Runner_1"
  url = "myurl"
  id = 1
  token = "mytoken"
  token_obtained_at = 2024-04-19T10:36:11Z
  token_expires_at = 0001-01-01T00:00:00Z
  executor = "virtualbox"
  shell = "powershell"
  [runners.cache]
    MaxUploadedArchiveSize = 0
  [runners.ssh]
    user = "gitlab-user"
    identity_file = "C:\\Users\\myuser\\.ssh\\id_ed25519"
    disable_strict_host_key_checking = true
  [runners.virtualbox]
    base_name = "Win10_prepare"
    base_folder = ""
    disable_snapshots = false
    start_type = ""

[[runners]]
  name = "Runner_2"
  url = "myurl"
  id = 2
  token = "mytoken"
  token_obtained_at = 2024-04-19T10:36:11Z
  token_expires_at = 0001-01-01T00:00:00Z
  executor = "virtualbox"
  shell = "powershell"
  [runners.cache]
    MaxUploadedArchiveSize = 0
  [runners.ssh]
    user = "gitlab-user"
    identity_file = "C:\\Users\\myuser\\.ssh\\id_ed25519"
    disable_strict_host_key_checking = true
  [runners.virtualbox]
    base_name = "Win10_prepare"
    base_folder = ""
    disable_snapshots = false
    start_type = ""

两种配置均出现相同问题。

解决方案 #

1. 强制创建完整克隆而非链接克隆 #

GitLab Runner默认使用VirtualBox的链接克隆，多个并发VM会共享同一基础磁盘文件，导致锁定冲突。修改Runner配置，添加clone_type = "full"参数：

[runners.virtualbox]
  base_name = "Win10_prepare"
  base_folder = ""
  disable_snapshots = false
  start_type = ""
  clone_type = "full"

完整克隆会生成独立的磁盘文件，避免共享锁定问题，但会占用更多磁盘空间。

2. 检查基础镜像的磁盘设置 #

• 打开VirtualBox管理器，选中Win10_prepare镜像，进入设置→存储
• 右键虚拟磁盘→属性，确保磁盘未被标记为“不可共享”，或直接设置为可共享
• 停止所有该镜像的运行实例，重新创建干净的快照，确保基础镜像处于完全关闭状态

3. 验证Runner权限与宿主机资源 #

• 确保GitLab Runner服务运行的用户拥有VirtualBox安装目录及虚拟磁盘文件目录的读写权限
• 宿主机需分配足够的CPU核心、内存资源给每个VM，资源不足可能触发磁盘锁定异常

4. 禁用快照（可选） #

如果不需要基于快照恢复，可以尝试禁用快照功能，直接从基础镜像克隆：

[runners.virtualbox]
  disable_snapshots = true

内容的提问来源于stack exchange，提问作者Roli Mayer