运维 – 📔独行看客的笔记本

「记录」群晖DSM7 SSO客户端对接Authing服务

薛定谔的nil2022年11月10日 0 Comments

为了方便老师登陆各种院内系统，跟学院申请购买了Authing的统一身份认证服务。其他购买的系统都可以让开发方对接，唯一麻烦的就是群晖的DSM，只能自己来了。

最终目标是想实现每一个用户，网页登陆群晖（及任何一个学院购买的系统）的时候直接微信扫码登陆，无需输入用户名密码。群晖相较于其他系统还有一个不同点，就是SMB服务必须输入用户名密码，所以要么从Authing直接获取带密码的用户信息，要么就得让从Authing登陆的用户能自己改密码。

关于群晖DSM的SSO客户端功能的资料非常少，群晖官方也是几句就把用户打发了，而且群晖公司从销售到技术都透着一股傲然于世的气息（虽然我打心眼里觉得他家NAS做的确实很好，但这种自以为已经能拳打EMC、脚踩HPE的态度实在让人想踩死他）。找遍全网只找到这篇资料比较靠谱，而且这篇资料的最终目标和我出奇的匹配，甚至我怀疑他也是和Authing对接。鉴于现在的局域网环境，我还是把原文截图转发过来。

虽然上面的资料和我的目标几乎完全相同，但这篇文章并没有详细描述所有在操作过程中可能遇到的坑，因此我觉得有必要记录一下我的踩坑过程。

1、群晖DSM7不支持用从SSO服务器获取的用户信息自动注册本地用户

这也是上述资料第一段所说的。详细解释就是：一个在Authing里设置好的用户，在登陆DSM之后，DSM并不会为这个用户自动创建一个本地用户，更不要提把这个本地用户和Authing的用户进行关联。这会导致一系列的麻烦，比如这个用户怎么设置DSM里共享文件夹的访问权限、这个用户怎么使用SMB服务等，这些问题很容易被忽略但又很致命（我一开始就没注意资料里第一段写了啥，都弄完才发现很重要）。

解决方法：先把DSM加入Authing的LDAP域，让DSM获取到Authing里配置好的用户、群组等信息，然后同时启用SSO客户端，并用一个唯一值进行用户关联。

2、群晖DSM本地用户和域用户冲突

当DSM本地和Authing用户池里同时有一个用户名为“张三”的用户时，会发生冲突。“张三”用户使用SSO方式进行登陆时，DSM会提示用户名密码错误。具体原因大约只有群晖工程师能解释，但我猜测是DSM没有将域用户和本地用户进行区分，导致用SSO登陆的重名用户匹配到本地用户鉴权信息上，最后鉴权失败。

解决方法：保证用户池里所有的用户名都不会出现在本地用户里。

3、配置DSM的LDAP客户端

配置LDAP客户端的最后一步是检测，会提示“不支持Samba架构”。这是由于Authing的LDAP实现问题，必须开启DSM的“启用 GIFS 明文密码验证”功能，才能通过检测。此处建议不开启，因为开启之后有安全风险。所以点详情，然后跳过。

之后还会提示另一个问题，同样方式处理，跳过。最后完成所有检测之后，点击确定。

4、在Authing添加自定义应用

往下找到认证配置，修改登陆回调地址！然后再点击其他配置！

【注意】在其他配置中只修改id_token签名算法！！！其余保持不变！完成后保存

5、配置DSM的SSO客户端

到此，就完成了所有操作，可以尝试使用Authing登陆DSM了！

「记录」使用PCI设备直通技术在vSphere中部署基于Ubuntu Server 1804的GPU服务器

薛定谔的nil2021年1月24日 0 Comments

大约在两年前我撰写了一篇文章记录我为实验室部署GPU云的核心过程。现在给学院部署GPU云，本以为可以照着原来的文章很顺利就可以装完，结果却搞得心态爆炸，一度怀疑人生，所以很有必要再写一篇记录一下。

环境：
实体机：曙光天阔 W780-G20
操作系统： VMware ESXi, 6.7.0, 14320388
管理平台： VMware vCenter Server Appliance, 6.7.0, 15129973
GPU：NVIDIA GP100GL [Tesla P100 PCIe 16GB] ×8

先简单说一下部署环境和上次的区别：第一是平台不再是商业大厂DELL EMC，很多文档资料缺失；第二是底层esxi和管理平台vcenter都升级到了6.7；第三是虚拟机准备安装Ubuntu Server 1804，之前的16在21年来看有点过老。最关键的是第四点：显卡质量和数量都上了一个巨大台阶，原来单台机器最多3卡（一个泰坦X，两个泰坦XP），遇到的显存最大的卡应该就是p40。当时在安装p40的时候就遇到奇怪的问题，为此还写了一篇记录文章。这次遇到的核心问题和p40那次应该是同一个，都是高级显卡显存过大，需要一些特定的底层配置参数，vmware专门有博客文章（下简称博文）进行了介绍，但是我一开始寻找解决方案的方向有很大偏差，绕了一大圈才找到。

0. 准备工作

0.1. 主机BIOS设置

根据博文内容，需要在运行虚拟GPU服务器的esxi主机的BIOS设置中查找并开启类似这样的选项：“above 4G decoding”、“memory mapped I/O above 4GB”、“PCI 64-bit resource handing above 4G”

0.2. 配置PCI设备直通

在esxi主机中配置好显卡的设备直通

0.3. 上传系统安装镜像

将ubuntu server 1804安装镜像上传到主机数据存储

1. 创建虚拟机

⚠️注意：必须在vsphere client中创建虚拟机，不可以在本地workstation创建虚拟机安装后再上传！

首先按照常规方法创建虚拟机，创建后不要启动！按以下步骤进行预先配置。

1.1. 设置虚拟机引导方式为EFI

在虚拟机页面，“操作”→“编辑设置”→“虚拟机选项”→“引导选项”→“固件”，选择“EFI”。

1.2. 设置虚拟机高级配置参数

在刚才1.1的“引导选项”下方找到“高级”，点击“配置参数”右侧的“编辑配置”，进入高级参数配置窗口。

在窗口中点击“添加配置参数”，然后添加2条：

pciPassthru.use64bitMMIO = "TRUE"
pciPassthru.64bitMMIOSizeGB = "[⚠️此处按需填写⚠️]"

1 2	pciPassthru.use64bitMMIO = "TRUE" pciPassthru.64bitMMIOSizeGB = "[⚠️此处按需填写⚠️]"

⚠️注意：“pciPassthru.64bitMMIOSizeGB”条目的值需要根据虚拟机具体直通的显卡型号和数量进行计算后得出！

博文中提供了一种简单估算方法：GPU个数 × 一块GPU的显存大小（GB为单位），然后将结果向上取整到下一个2的整数次幂。

例如，要给这个虚拟机直通两个p100显卡，则值应为：2 × 16 = 32，向上取整到下一个2的整数次幂，得到64。但如果是直通v100显卡，64则只能直通一块，因为v100单卡显存就是32G，想要直通两块就需要改为128。

⚠️注意：与通常情况不同的是不可以添加“hypervisor.cpuid.v0 = FALSE”参数！

1.3. 修改虚拟机兼容性

在虚拟机页面，“操作”→“兼容性”→“升级虚拟机兼容性”，打开“配置虚拟机兼容性窗口”。

将虚拟机兼容性改为：ESXi 6.7 Update 2 及更高版本

2. 安装操作系统

现在可以进行第一次开机，并按照提示安装操作系统。安装后，顺便进行基本的系统配置，例如换源、配置网络、软件包升级等。

3. 为虚拟机添加显卡直通

3.1. 启用内存预留

在虚拟机页面，“操作”→“编辑设置”→“虚拟硬件”→“内存”，勾选“预留所有客户机内存（全部锁定）”。

3.2. 添加显卡设备

在虚拟机页面，“操作”→“编辑设置”→“虚拟硬件”→“添加新设备”→“PCI 设备”，然后在PCI设备选项中选择需要添加的显卡。

4. 安装NVIDIA驱动和CUDA

4.1. 检查GPU

lspci | grep -i nvidia

1	lspci \| grep -i nvidia

4.2. 禁用nouveau

# 若有输出则表示nouveau正在加载，需要禁用。
lsmod | grep nouveau

sudo nano /etc/modprobe.d/blacklist-nouveau.conf
# 文件中添加以下内容
blacklist nouveau  
options nouveau modeset=0

# “4.15.0-134”可能会根据机器差异发送变化，需要到具体路径下确认。
cd /lib/modules/4.15.0-134-generic/kernel/drivers/gpu/drm/nouveau 
sudo rm -rf nouveau.ko 
sudo rm -rf nouveau.ko.org

sudo nano /etc/modprobe.d/blacklist.conf
# 文件末尾追加以下内容
blacklist rivafb
blacklist vga16fb
blacklist nouveau
blacklist nvidiafb
blacklist rivatv

# 更新并重启
sudo update-initramfs -u
sudo reboot

# 若有输出则表示nouveau正在加载，需要禁用。

lsmod | grep nouveau

sudo nano /etc/modprobe.d/blacklist-nouveau.conf

# 文件中添加以下内容

blacklist nouveau

options nouveau modeset=0

# “4.15.0-134”可能会根据机器差异发送变化，需要到具体路径下确认。

cd /lib/modules/4.15.0-134-generic/kernel/drivers/gpu/drm/nouveau

sudo rm -rf nouveau.ko

sudo rm -rf nouveau.ko.org

sudo nano /etc/modprobe.d/blacklist.conf

# 文件末尾追加以下内容

blacklist rivafb

blacklist vga16fb

blacklist nouveau

blacklist nvidiafb

blacklist rivatv

# 更新并重启

sudo update-initramfs -u

sudo reboot

4.3. 安装依赖包

sudo apt-get install linux-headers-$(uname -r)
sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

1 2	sudo apt-get install linux-headers-$(uname -r) sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

4.4. 安装NVIDIA驱动

默认已经下载好驱动文件

sudo apt-get remove --purge nvidia-*

sudo chmod +x NVIDIA-Linux-x86_64-440.118.02.run
sudo ./NVIDIA-Linux-x86_64-440.118.02.run --no-x-check --no-nouveau-check --no-opengl-files

# 随后会进入一个简陋的彩色界面，根据它的指示安装就可以了。

# 重启，检查驱动是否已正常运行
sudo reboot
nvidia-smi
# 应该显示相应的显卡运行情况

sudo apt-get remove --purge nvidia-*

sudo chmod +x NVIDIA-Linux-x86_64-440.118.02.run

sudo ./NVIDIA-Linux-x86_64-440.118.02.run --no-x-check --no-nouveau-check --no-opengl-files

# 随后会进入一个简陋的彩色界面，根据它的指示安装就可以了。

# 重启，检查驱动是否已正常运行

sudo reboot

nvidia-smi

# 应该显示相应的显卡运行情况

4.5. 安装CUDA

sudo chmod +x ./cuda_10.2.89_440.33.01_linux.run

# 安装CUDA
sudo sh ./cuda_10.2.89_440.33.01_linux.run

# 经过一小段时间的等待，会出现用户许可协议，输入accept然后回车继续。
# 之后会有安装选项，是一个文字模拟的选项界面。注意如果已经安装驱动，就不需要再次安装了！

# 安装完成后应该显示如下信息
===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-10.2/
Samples:  Installed in /home/ubuntu/

Please make sure that
 -   PATH includes /usr/local/cuda-10.2/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-10.2/lib64, or, add /usr/local/cuda-10.2/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-10.2/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.2/doc/pdf for detailed information on setting up CUDA.
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 440.00 is required for CUDA 10.2 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run --silent --driver

Logfile is /var/log/cuda-installer.log
# 安装完成后应该还需要配置环境变量等操作，但我不需要在宿主机运行cuda，因此略过。

# 重启，检查CUDA是否已正常运行
sudo reboot
cd  /usr/local/cuda-10.0/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
# 应该返回以下信息
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 2 CUDA Capable device(s)

Device 0: "Tesla P100-PCIE-16GB"
  CUDA Driver Version / Runtime Version          10.2 / 10.2
  CUDA Capability Major/Minor version number:    6.0
  Total amount of global memory:                 16281 MBytes (17071734784 bytes)
  (56) Multiprocessors, ( 64) CUDA Cores/MP:     3584 CUDA Cores
  GPU Max Clock rate:                            1329 MHz (1.33 GHz)
  Memory Clock rate:                             715 Mhz
  Memory Bus Width:                              4096-bit
  L2 Cache Size:                                 4194304 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Enabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 11 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 1: "Tesla P100-PCIE-16GB"
  CUDA Driver Version / Runtime Version          10.2 / 10.2
  CUDA Capability Major/Minor version number:    6.0
  Total amount of global memory:                 16281 MBytes (17071734784 bytes)
  (56) Multiprocessors, ( 64) CUDA Cores/MP:     3584 CUDA Cores
  GPU Max Clock rate:                            1329 MHz (1.33 GHz)
  Memory Clock rate:                             715 Mhz
  Memory Bus Width:                              4096-bit
  L2 Cache Size:                                 4194304 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 2 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Enabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 19 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
> Peer access from Tesla P100-PCIE-16GB (GPU0) -> Tesla P100-PCIE-16GB (GPU1) : No
> Peer access from Tesla P100-PCIE-16GB (GPU1) -> Tesla P100-PCIE-16GB (GPU0) : No

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.2, CUDA Runtime Version = 10.2, NumDevs = 2
Result = PASS

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

sudo chmod +x ./cuda_10.2.89_440.33.01_linux.run

# 安装CUDA

sudo sh ./cuda_10.2.89_440.33.01_linux.run

# 经过一小段时间的等待，会出现用户许可协议，输入accept然后回车继续。

# 之后会有安装选项，是一个文字模拟的选项界面。注意如果已经安装驱动，就不需要再次安装了！

# 安装完成后应该显示如下信息

===========

= Summary =

===========

Driver: Not Selected

Toolkit: Installed in /usr/local/cuda-10.2/

Samples: Installed in /home/ubuntu/

Please make sure that

- PATH includes /usr/local/cuda-10.2/bin

- LD_LIBRARY_PATH includes /usr/local/cuda-10.2/lib64, or, add /usr/local/cuda-10.2/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-10.2/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.2/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 440.00 is required for CUDA 10.2 functionality to work.

To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:

sudo <CudaInstaller>.run --silent --driver

Logfile is /var/log/cuda-installer.log

# 安装完成后应该还需要配置环境变量等操作，但我不需要在宿主机运行cuda，因此略过。

# 重启，检查CUDA是否已正常运行

sudo reboot

cd /usr/local/cuda-10.0/samples/1_Utilities/deviceQuery

sudo make

./deviceQuery

# 应该返回以下信息

./deviceQuery Starting...

CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 2 CUDA Capable device(s)

Device 0: "Tesla P100-PCIE-16GB"

CUDA Driver Version / Runtime Version 10.2 / 10.2

CUDA Capability Major/Minor version number: 6.0

Total amount of global memory: 16281 MBytes (17071734784 bytes)

(56) Multiprocessors, ( 64) CUDA Cores/MP: 3584 CUDA Cores

GPU Max Clock rate: 1329 MHz (1.33 GHz)

Memory Clock rate: 715 Mhz

Memory Bus Width: 4096-bit

L2 Cache Size: 4194304 bytes

Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)

Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers

Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers

Total amount of constant memory: 65536 bytes

Total amount of shared memory per block: 49152 bytes

Total number of registers available per block: 65536

Warp size: 32

Maximum number of threads per multiprocessor: 2048

Maximum number of threads per block: 1024

Max dimension size of a thread block (x,y,z): (1024, 1024, 64)

Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)

Maximum memory pitch: 2147483647 bytes

Texture alignment: 512 bytes

Concurrent copy and kernel execution: Yes with 2 copy engine(s)

Run time limit on kernels: No

Integrated GPU sharing Host Memory: No

Support host page-locked memory mapping: Yes

Alignment requirement for Surfaces: Yes

Device has ECC support: Enabled

Device supports Unified Addressing (UVA): Yes

Device supports Compute Preemption: Yes

Supports Cooperative Kernel Launch: Yes

Supports MultiDevice Co-op Kernel Launch: Yes

Device PCI Domain ID / Bus ID / location ID: 0 / 11 / 0

Compute Mode:

< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 1: "Tesla P100-PCIE-16GB"

CUDA Driver Version / Runtime Version 10.2 / 10.2

CUDA Capability Major/Minor version number: 6.0

Total amount of global memory: 16281 MBytes (17071734784 bytes)

(56) Multiprocessors, ( 64) CUDA Cores/MP: 3584 CUDA Cores

GPU Max Clock rate: 1329 MHz (1.33 GHz)

Memory Clock rate: 715 Mhz

Memory Bus Width: 4096-bit

L2 Cache Size: 4194304 bytes

Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)

Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers

Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers

Total amount of constant memory: 65536 bytes

Total amount of shared memory per block: 49152 bytes

Total number of registers available per block: 65536

Warp size: 32

Maximum number of threads per multiprocessor: 2048

Maximum number of threads per block: 1024

Max dimension size of a thread block (x,y,z): (1024, 1024, 64)

Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)

Maximum memory pitch: 2147483647 bytes

Texture alignment: 512 bytes

Concurrent copy and kernel execution: Yes with 2 copy engine(s)

Run time limit on kernels: No

Integrated GPU sharing Host Memory: No

Support host page-locked memory mapping: Yes

Alignment requirement for Surfaces: Yes

Device has ECC support: Enabled

Device supports Unified Addressing (UVA): Yes

Device supports Compute Preemption: Yes

Supports Cooperative Kernel Launch: Yes

Supports MultiDevice Co-op Kernel Launch: Yes

Device PCI Domain ID / Bus ID / location ID: 0 / 19 / 0

Compute Mode:

< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

> Peer access from Tesla P100-PCIE-16GB (GPU0) -> Tesla P100-PCIE-16GB (GPU1) : No

> Peer access from Tesla P100-PCIE-16GB (GPU1) -> Tesla P100-PCIE-16GB (GPU0) : No

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.2, CUDA Runtime Version = 10.2, NumDevs = 2

Result = PASS

「记录」解决Ubuntu Server 18.04因Intel微码修复无法启动问题

薛定谔的nil2021年1月21日 0 Comments

最近在重建学院GPU云，使用原来实验室云的双层虚拟化架构。原来实验室云使用的是基于ubuntu16的GPU服务器，21年再使用16版本就有点老了，准备换成ubuntu18，但是遇到一个奇怪的问题，在此记录一下。

问题的表现形式如下：

环境：
实体机：曙光天阔 W780-G20
操作系统： VMware ESXi, 6.7.0, 14320388
管理平台： VMware vCenter Server Appliance, 6.7.0, 15129973
虚拟机兼容性： ESXi 6.7 Update 2
虚拟机高级配置附加参数：
hypervisor.cpuid.v0 = FALSE
pciPassthru.use64bitMMIO = TRUE
虚拟机操作系统版本：Ubuntu Server 18.04.5 LTS
截止2021年1月20日，已更新所有最新包

当虚拟机CPU只给一个核心的时候，一切正常。当给2个或者更多核心的时候会出现无法启动的问题，显示如下：

经过排查发现问题触发的原因有两个：①虚拟机高级配置参数 hypervisor.cpuid.v0 = FALSE；②有2个或更多CPU核心

通过搜索找到两篇类似问题的文章：ESXi 6.7 ubuntu GPU直连踩坑记、VMware ESXi 6.7.0 update2 使用 GPU Passthrough 模式的坑。第一篇文章是通过更新esxi解决的，但是第一篇文章里更新后的esxi版本与我现在使用的是同一版本，所以解决方法对我不适用。第二篇文章将问题的根源指向了intel的漏洞修补微码软件包“intel-microcode”。

通过测试发现，无法启动的核心问题就是由“intel-microcode”导致的，该软件包是intel为了修补幽灵、熔断漏洞的补丁，应该是在启动之前执行某种防护操作。我测试时首先尝试卸载该软件包，并同时卸载依赖包，结果导致部分核心包被卸载，无法启动。之后测试只卸载该软件包，不卸载依赖，结果问题解决，确定问题核心就是该软件包导致。

为了解决问题，我首先想到的是在安装vsphere过程中被提示可能受intel的漏洞影响，所以我考虑可能是虚拟化平台的宿主机没有进行漏洞修补，而ubuntu虚拟机进行了漏洞修补，出现兼容性问题。为此我查阅了vsphere的文档，确定默认状态下vsphere6.7确实是不启用漏洞修补程序的。同时，我也发现启动修补程序会导致严重的性能损失，不适合我现在的场景。

之后，我考虑既然虚拟化平台没有进行漏洞修补，那关掉ubuntu虚拟机的漏洞修补应该可以解决问题。于是我阅读了第二篇文章中提到的讨论：Intel-microcode package upgrade in ubuntu 18.04 leads to unbootable system。根据这个讨论，可以通过更改grub参数的方式禁用微码修补，但是尝试了讨论中的解决方案和关联的解决方案，发现并没有效果。最终，我决定按照讨论中的最终解决方案，更换旧版本的二进制软件包。

解决方案：

# 查看已安装包的版本号
sudo apt list --installed | grep intel-microcode

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

intel-microcode/bionic-updates,bionic-security,now 3.20201110.0ubuntu0.18.04.2 amd64 [installed,automatic]

# 查看软件源中该软件包的所有支持版本
apt-cache madison Ubuntu Server
intel-microcode | 3.20201110.0ubuntu0.18.04.2 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic-updates/main amd64 Packages
intel-microcode | 3.20201110.0ubuntu0.18.04.2 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic-security/main amd64 Packages
intel-microcode | 3.20180312.0~ubuntu18.04.1 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic/main amd64 Packages
# 删除原有软件包
sudo apt remove intel-microcode
# 安装旧版软件包
sudo apt install intel-microcode=3.20180312.0~ubuntu18.04.1
# 禁止该软件包自动更新
sudo apt-mark hold intel-microcode

# 查看已安装包的版本号

sudo apt list --installed | grep intel-microcode

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

intel-microcode/bionic-updates,bionic-security,now 3.20201110.0ubuntu0.18.04.2 amd64 [installed,automatic]

# 查看软件源中该软件包的所有支持版本

apt-cache madison Ubuntu Server

intel-microcode | 3.20201110.0ubuntu0.18.04.2 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic-updates/main amd64 Packages

intel-microcode | 3.20201110.0ubuntu0.18.04.2 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic-security/main amd64 Packages

intel-microcode | 3.20180312.0~ubuntu18.04.1 | https://mirrors.tuna.tsinghua.edu.cn/ubuntu bionic/main amd64 Packages

# 删除原有软件包

sudo apt remove intel-microcode

# 安装旧版软件包

sudo apt install intel-microcode=3.20180312.0~ubuntu18.04.1

# 禁止该软件包自动更新

sudo apt-mark hold intel-microcode

[记录]配置Docker出现的错误

薛定谔的nil2019年5月22日 0 Comments

　　为了方便管理实验室的GPU资源，上了多层虚拟化，在部署和运维Docker节点的过程中踩了一些坑，在这记录一下，以便日后查阅。

1、Docker服务无法启动

sudo service docker start

# systemctl status docker.service
systemd[1]: Failed to start Docker Application Container Engine.
systemd[1]: docker.service: Unit entered failed state.
systemd[1]: docker.service: Failed with result 'start-limit-hit'.

# journalctl -xe
dockerd[3285]: unable to configure the Docker daemon with file etc/docker/daemon.json: the following directives are specified both as a flag and in the configuration file: hosts: (from flag: [fd://], from file: [unix:///var/run/docker.sock tcp://0.0.0.0:56379])
systemd[1]: docker.service: Main process exited, code=exited, status=1/FAILURE
systemd[1]: Failed to start Docker Application Container Engine.

sudo service docker start

# systemctl status docker.service

systemd[1]: Failed to start Docker Application Container Engine.

systemd[1]: docker.service: Unit entered failed state.

systemd[1]: docker.service: Failed with result 'start-limit-hit'.

# journalctl -xe

dockerd[3285]: unable to configure the Docker daemon with file etc/docker/daemon.json: the following directives are specified both as a flag and in the configuration file: hosts: (from flag: [fd://], from file: [unix:///var/run/docker.sock tcp://0.0.0.0:56379])

systemd[1]: docker.service: Main process exited, code=exited, status=1/FAILURE

systemd[1]: Failed to start Docker Application Container Engine.

问题定位： Docker启动的时候传入了命令行参数，同时也指定了配置文件，两个配置发生了冲突。

解决方法：

sudo nano /lib/systemd/system/docker.service

# ExecStart=/usr/bin/dockerd -H fd://
# 改为
# ExecStart=/usr/bin/dockerd
# # -H fd://

sudo systemctl daemon-reload
sudo service docker restart

sudo nano /lib/systemd/system/docker.service

# ExecStart=/usr/bin/dockerd -H fd://

# 改为

# ExecStart=/usr/bin/dockerd

# # -H fd://

sudo systemctl daemon-reload

sudo service docker restart

[记录]英伟达Tesla K80显卡直通出现未知错误

薛定谔的nil2019年5月21日 0 Comments

　　实验室添置了新显卡，超微塔式服务器没地方了，只好把旧显卡换到戴尔刀片里，之后在做显卡直通的时候遇到一些莫名其妙的问题，折腾了好久才解决，在这记录一下。

环境：
实体机：DELL PowerEdge R730
操作系统： VMware ESXi, 6.5.0, 6765664
管理平台： VMware vCenter Server Appliance, 6.7.0.21000, 11726888
设备： Nvidia Tesla K80

　　问题是这样的：在主机上开虚拟机，并直通其他PCI设备（usb控制器），一切正常，且虚拟机正常启动，正常识别设备。给虚拟机直通Nvidia Tesla K80，虚拟机无法开机，报未知异常，如下图：

　　尝试了很多国内外常见的解决方案都无法生效。怀疑是跟显卡型号有关，最后在官方社区找到一个出现完全相同问题的讨论：Passing through Tesla k80 Issue…。下面有一个官方人员的回答是：

Re: Passing through Tesla k80 Issue…
A previous version of this post included advice to add two VMX file entries (efi.legacyBoot.enabled and efi.bootOrder) as part of the solution. These two settings should NOT be used. Instead, following the directions below.
——–
You should be able to pass a single GPU (that is, half of a K80) to a VM running on ESX 6 by creating an EFI-bootable VM, doing an EFI installation of your guest OS, and then adding the following to the VM’s VMX file.
pciPassthru.use64bitMMIO=”TRUE”
Trying to pass more than one of these GPUs into the same VM will currently hit a platform memory limit and the VM will fail to boot. (NOTE: This limit has been removed in ESX 6.5).
A smaller card like the K2 does not have this issue: GPGPU Blog Entry
If the above does not work for you, send me email directly at “simons at vmware dot com”. In either case, please share your experience with others on the thread.
And if you have any other questions about running HPC applications in a VMware environment, I’d be happy to hear from you directly.
If you are interested in learning more of what we’ve been doing related to HPC, you can check out our HPC entries on the VMware CTO blog site here: HPC Blog Entries

Josh Simons
High Performance Computing
Office of the CTO
VMware, Inc.

　　关键是要添加以下高级参数：

pciPassthru.use64bitMMIO = "TRUE"

1	pciPassthru.use64bitMMIO = "TRUE"

「记录」在vSphere中基于Ubuntu Server 1604部署Docker，并安装CUDA，构建多层虚拟化深度学习开发环境

薛定谔的nil2019年2月25日 0 Comments

首先准备好Ubuntu Server 1604的虚拟机，只设定基本功能，不添加显卡直通等特殊功能，以便于基础环境完成后进行快照。

安装好系统后，进行ssh端口设置，apt换源，apt upgrade等基本操作，之后关机打快照。

1、显卡直通

1.1、添加显卡直通

勾选“预留所有客户机内存（全部锁定）”，并添加显卡PCI设备，确定修改。不在第一步直通显卡的话，之后的驱动安装可能有问题！

1.2、修改虚拟机硬件配置参数

切换到“虚拟机硬件”选项卡。展开“高级”，修改“配置参数”。在打开的“配置参数”对话框中，点击“添加配置参数”，填入“
hypervisor.cpuid.v0 = FALSE”，确定修改。完成后务必再次进入“配置参数”，检查参数确实已经添加并保存！

1.3、启动虚拟机并配置基本环境

配置网络、换源、更新软件包等

完成基本配置后建议对虚拟机生成快照！

2、安装显卡驱动

2.1、下载驱动

可以在 NVIDIA驱动下载或者 Geforce显卡驱动下载页面搜索并下载需要的驱动程序，也可以通过直接拼接资源链接：
http://cn.download.nvidia.com/XFree86/Linux-x86_64/[驱动版本号]/NVIDIA-Linux-x86_64-[驱动版本号].run，进行下载。比如我要下载430.14版本号的Linux驱动，可以直接在服务器运行以下代码：

cd ~/downloads
wget https://cn.download.nvidia.cn/XFree86/Linux-x86_64/430.14/NVIDIA-Linux-x86_64-430.14.run

1 2	cd ~/downloads wget https://cn.download.nvidia.cn/XFree86/Linux-x86_64/430.14/NVIDIA-Linux-x86_64-430.14.run

2.2、检查GPU

应该显示相应的PCI设备

lspci | grep -i nvidia

1	lspci \| grep -i nvidia

2.3、禁用nouveau

# 若有输出则表示nouveau正在加载，需要禁用。
lsmod | grep nouveau

cd /etc/modprobe.d
sudo touch blacklist-nouveau.conf
sudo nano blacklist-nouveau.conf
# 文件中添加以下内容
blacklist nouveau  
options nouveau modeset=0

# “4.4.0-83”可能会根据机器差异发送变化，需要到具体路径下确认。
cd /lib/modules/4.4.0-83-generic/kernel/drivers/gpu/drm/nouveau 
sudo rm -rf nouveau.ko 
sudo rm -rf nouveau.ko.org

sudo nano /etc/modprobe.d/blacklist.conf
# 文件末尾追加以下内容
blacklist rivafb
blacklist vga16fb
blacklist nouveau
blacklist nvidiafb
blacklist rivatv

# 更新并重启
sudo update-initramfs -u
sudo reboot

# 若有输出则表示nouveau正在加载，需要禁用。

lsmod | grep nouveau

cd /etc/modprobe.d

sudo touch blacklist-nouveau.conf

sudo nano blacklist-nouveau.conf

# 文件中添加以下内容

blacklist nouveau

options nouveau modeset=0

# “4.4.0-83”可能会根据机器差异发送变化，需要到具体路径下确认。

cd /lib/modules/4.4.0-83-generic/kernel/drivers/gpu/drm/nouveau

sudo rm -rf nouveau.ko

sudo rm -rf nouveau.ko.org

sudo nano /etc/modprobe.d/blacklist.conf

# 文件末尾追加以下内容

blacklist rivafb

blacklist vga16fb

blacklist nouveau

blacklist nvidiafb

blacklist rivatv

# 更新并重启

sudo update-initramfs -u

sudo reboot

2.3、安装显卡驱动

sudo apt-get remove --purge nvidia-*
cd ~/downloads
sudo chmod +x NVIDIA-Linux-x86_64-430.14.run
sudo ./NVIDIA-Linux-x86_64-430.14.run --no-x-check --no-nouveau-check --no-opengl-files
# --no-x-check 安装驱动时关闭X服务
# --no-nouveau-check 安装驱动时禁用nouveau
# --no-opengl-files 只安装驱动文件，不安装OpenGL文件

# 随后会进入一个简陋的彩色界面，根据它的指示安装就可以了。

# 重启，检查驱动是否已正常运行
sudo reboot
nvidia-smi
# 应该显示相应的显卡运行情况

sudo apt-get remove --purge nvidia-*

cd ~/downloads

sudo chmod +x NVIDIA-Linux-x86_64-430.14.run

sudo ./NVIDIA-Linux-x86_64-430.14.run --no-x-check --no-nouveau-check --no-opengl-files

# --no-x-check 安装驱动时关闭X服务

# --no-nouveau-check 安装驱动时禁用nouveau

# --no-opengl-files 只安装驱动文件，不安装OpenGL文件

# 随后会进入一个简陋的彩色界面，根据它的指示安装就可以了。

# 重启，检查驱动是否已正常运行

sudo reboot

nvidia-smi

# 应该显示相应的显卡运行情况

安装完驱动建议对虚拟机生成快照！

然后开机安装docker，去清华源帮助文档里找清华源安装方法。安装好之后配置docker远程管理，参考“docker 配置 TLS 认证开启远程访问”和“Docker 守护进程+远程连接+安全访问+启动冲突解决办法 (完整收藏版)”。配置好docker之后关机打快照。

~~在ESXi中添加显卡直通，并启动。~~

2020年6月28日补充

2.4、安装CUDA

本处使用的CUDA安装文件是文章在19年6月初次完成时下载的，因此安装方法也是当时的安装方法。根据从英伟达官网的CUDA安装文档来看，最新的推荐安装方式似乎变成了用apt包管理器安装。但是根据最新的安装文档，我在操作过程中出现了两个问题：首先是所有的下载资源（包括软件源证书和软件包）都无法在中国大陆正常下载；另一个是安装后server系统会出现GUI界面。因此，我决定使用去年的安装包和安装方法。

cd ~/downloads
sudo chmod +x ./cuda_10.0.130_410.48_linux.run
# 安装依赖包
sudo apt-get install linux-headers-$(uname -r)
sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev
# 安装CUDA
sudo sh ./cuda_10.0.130_410.48_linux.run

# 经过一小段时间的等待，会出现一个非常长的用户许可协议，按住回车键直到协议走完，不用担心按过头，最后需要输入accept才会继续。
# 之后会有安装选项，类似下面的内容。注意如果已经安装驱动，就不需要再次安装了！
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?
(y)es/(n)o/(q)uit: n

Install the CUDA 10.0 Toolkit?
(y)es/(n)o/(q)uit: y

Enter Toolkit Location
 [ default is /usr/local/cuda-10.0 ]: 

Do you want to install a symbolic link at /usr/local/cuda?
(y)es/(n)o/(q)uit: y

Install the CUDA 10.0 Samples?
(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location
 [ default is /home/ubuntu ]: 

Installing the CUDA Toolkit in /usr/local/cuda-10.0 ...
Installing the CUDA Samples in /home/ubuntu ...
Copying samples to /home/ubuntu/NVIDIA_CUDA-10.0_Samples now...
Finished copying samples.

# 安装完成后应该显示如下信息
===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-10.0
Samples:  Installed in /home/ubuntu

Please make sure that
 -   PATH includes /usr/local/cuda-10.0/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 384.00 is required for CUDA 10.0 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run -silent -driver

Logfile is /tmp/cuda_install_6983.log
# 安装完成后应该还需要配置环境变量等操作，但我不需要在宿主机运行cuda，因此略过。

# 重启，检查CUDA是否已正常运行
sudo reboot
cd  /usr/local/cuda-10.0/samples/1_Utilities/deviceQuery
sudo make
./deviceQuery
# 应该返回以下信息
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 2 CUDA Capable device(s)

Device 0: "GeForce RTX 2080 Ti"
  CUDA Driver Version / Runtime Version          10.2 / 10.0
  CUDA Capability Major/Minor version number:    7.5
  Total amount of global memory:                 11019 MBytes (11554717696 bytes)
  (68) Multiprocessors, ( 64) CUDA Cores/MP:     4352 CUDA Cores
  GPU Max Clock rate:                            1545 MHz (1.54 GHz)
  Memory Clock rate:                             7000 Mhz
  Memory Bus Width:                              352-bit
  L2 Cache Size:                                 5767168 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1024
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 3 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 3 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 1: "GeForce RTX 2080 Ti"
  CUDA Driver Version / Runtime Version          10.2 / 10.0
  CUDA Capability Major/Minor version number:    7.5
  Total amount of global memory:                 11019 MBytes (11554717696 bytes)
  (68) Multiprocessors, ( 64) CUDA Cores/MP:     4352 CUDA Cores
  GPU Max Clock rate:                            1545 MHz (1.54 GHz)
  Memory Clock rate:                             7000 Mhz
  Memory Bus Width:                              352-bit
  L2 Cache Size:                                 5767168 bytes
  Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)
  Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers
  Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1024
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and kernel execution:          Yes with 3 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device supports Compute Preemption:            Yes
  Supports Cooperative Kernel Launch:            Yes
  Supports MultiDevice Co-op Kernel Launch:      Yes
  Device PCI Domain ID / Bus ID / location ID:   0 / 4 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >
> Peer access from GeForce RTX 2080 Ti (GPU0) -> GeForce RTX 2080 Ti (GPU1) : No
> Peer access from GeForce RTX 2080 Ti (GPU1) -> GeForce RTX 2080 Ti (GPU0) : No

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.2, CUDA Runtime Version = 10.0, NumDevs = 2
Result = PASS

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

cd ~/downloads

sudo chmod +x ./cuda_10.0.130_410.48_linux.run

# 安装依赖包

sudo apt-get install linux-headers-$(uname -r)

sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

# 安装CUDA

sudo sh ./cuda_10.0.130_410.48_linux.run

# 经过一小段时间的等待，会出现一个非常长的用户许可协议，按住回车键直到协议走完，不用担心按过头，最后需要输入accept才会继续。

# 之后会有安装选项，类似下面的内容。注意如果已经安装驱动，就不需要再次安装了！

Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?

(y)es/(n)o/(q)uit: n

Install the CUDA 10.0 Toolkit?

(y)es/(n)o/(q)uit: y

Enter Toolkit Location

[ default is /usr/local/cuda-10.0 ]:

Do you want to install a symbolic link at /usr/local/cuda?

(y)es/(n)o/(q)uit: y

Install the CUDA 10.0 Samples?

(y)es/(n)o/(q)uit: y

Enter CUDA Samples Location

[ default is /home/ubuntu ]:

Installing the CUDA Toolkit in /usr/local/cuda-10.0 ...

Installing the CUDA Samples in /home/ubuntu ...

Copying samples to /home/ubuntu/NVIDIA_CUDA-10.0_Samples now...

Finished copying samples.

# 安装完成后应该显示如下信息

===========

= Summary =

===========

Driver: Not Selected

Toolkit: Installed in /usr/local/cuda-10.0

Samples: Installed in /home/ubuntu

Please make sure that

- PATH includes /usr/local/cuda-10.0/bin

- LD_LIBRARY_PATH includes /usr/local/cuda-10.0/lib64, or, add /usr/local/cuda-10.0/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-10.0/bin

Please see CUDA_Installation_Guide_Linux.pdf in /usr/local/cuda-10.0/doc/pdf for detailed information on setting up CUDA.

***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 384.00 is required for CUDA 10.0 functionality to work.

To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:

sudo <CudaInstaller>.run -silent -driver

Logfile is /tmp/cuda_install_6983.log

# 安装完成后应该还需要配置环境变量等操作，但我不需要在宿主机运行cuda，因此略过。

# 重启，检查CUDA是否已正常运行

sudo reboot

cd /usr/local/cuda-10.0/samples/1_Utilities/deviceQuery

sudo make

./deviceQuery

# 应该返回以下信息

./deviceQuery Starting...

CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 2 CUDA Capable device(s)

Device 0: "GeForce RTX 2080 Ti"

CUDA Driver Version / Runtime Version 10.2 / 10.0

CUDA Capability Major/Minor version number: 7.5

Total amount of global memory: 11019 MBytes (11554717696 bytes)

(68) Multiprocessors, ( 64) CUDA Cores/MP: 4352 CUDA Cores

GPU Max Clock rate: 1545 MHz (1.54 GHz)

Memory Clock rate: 7000 Mhz

Memory Bus Width: 352-bit

L2 Cache Size: 5767168 bytes

Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)

Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers

Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers

Total amount of constant memory: 65536 bytes

Total amount of shared memory per block: 49152 bytes

Total number of registers available per block: 65536

Warp size: 32

Maximum number of threads per multiprocessor: 1024

Maximum number of threads per block: 1024

Max dimension size of a thread block (x,y,z): (1024, 1024, 64)

Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)

Maximum memory pitch: 2147483647 bytes

Texture alignment: 512 bytes

Concurrent copy and kernel execution: Yes with 3 copy engine(s)

Run time limit on kernels: No

Integrated GPU sharing Host Memory: No

Support host page-locked memory mapping: Yes

Alignment requirement for Surfaces: Yes

Device has ECC support: Disabled

Device supports Unified Addressing (UVA): Yes

Device supports Compute Preemption: Yes

Supports Cooperative Kernel Launch: Yes

Supports MultiDevice Co-op Kernel Launch: Yes

Device PCI Domain ID / Bus ID / location ID: 0 / 3 / 0

Compute Mode:

< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

Device 1: "GeForce RTX 2080 Ti"

CUDA Driver Version / Runtime Version 10.2 / 10.0

CUDA Capability Major/Minor version number: 7.5

Total amount of global memory: 11019 MBytes (11554717696 bytes)

(68) Multiprocessors, ( 64) CUDA Cores/MP: 4352 CUDA Cores

GPU Max Clock rate: 1545 MHz (1.54 GHz)

Memory Clock rate: 7000 Mhz

Memory Bus Width: 352-bit

L2 Cache Size: 5767168 bytes

Maximum Texture Dimension Size (x,y,z) 1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)

Maximum Layered 1D Texture Size, (num) layers 1D=(32768), 2048 layers

Maximum Layered 2D Texture Size, (num) layers 2D=(32768, 32768), 2048 layers

Total amount of constant memory: 65536 bytes

Total amount of shared memory per block: 49152 bytes

Total number of registers available per block: 65536

Warp size: 32

Maximum number of threads per multiprocessor: 1024

Maximum number of threads per block: 1024

Max dimension size of a thread block (x,y,z): (1024, 1024, 64)

Max dimension size of a grid size (x,y,z): (2147483647, 65535, 65535)

Maximum memory pitch: 2147483647 bytes

Texture alignment: 512 bytes

Concurrent copy and kernel execution: Yes with 3 copy engine(s)

Run time limit on kernels: No

Integrated GPU sharing Host Memory: No

Support host page-locked memory mapping: Yes

Alignment requirement for Surfaces: Yes

Device has ECC support: Disabled

Device supports Unified Addressing (UVA): Yes

Device supports Compute Preemption: Yes

Supports Cooperative Kernel Launch: Yes

Supports MultiDevice Co-op Kernel Launch: Yes

Device PCI Domain ID / Bus ID / location ID: 0 / 4 / 0

Compute Mode:

< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

> Peer access from GeForce RTX 2080 Ti (GPU0) -> GeForce RTX 2080 Ti (GPU1) : No

> Peer access from GeForce RTX 2080 Ti (GPU1) -> GeForce RTX 2080 Ti (GPU0) : No

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 10.2, CUDA Runtime Version = 10.0, NumDevs = 2

Result = PASS

安装完CUDA建议对虚拟机生成快照！

3、安装Docker

3.1、安装Docker

根据清华开源镜像站的帮助文档，安装Docker。

3.2、安装nvidia-docker

cd ~
# 添加软件源
curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

sudo apt-get update
sudo apt-get install -y nvidia-docker2

# 修改docker配置
sudo pkill -SIGHUP dockerd
sudo nano /etc/docker/daemon.json
# 配置文件样例，$端口号是你设定的端口号
{
    "hosts": ["unix:///var/run/docker.sock", "tcp://0.0.0.0:$端口号"],
    "registry-mirrors": ["https://******.mirror.aliyuncs.com"],
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}
# 主配置文件的启动参数和上述配置有冲突，重启docker会导致如下报错：
Job for docker.service failed because the control process exited with error code. See "systemctl status docker.service" and "journalctl -xe" for details.
# 因此，同时修改主配置文件
sudo nano /lib/systemd/system/docker.service
# 修改ExecStart参数为如下样式：
ExecStart=/usr/bin/dockerd
# -H fd:// --containerd=/run/containerd/containerd.sock

# 刷新配置，重启docker
systemctl daemon-reload
sudo systemctl restart docker

cd ~

# 添加软件源

curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -

distribution=$(. /etc/os-release;echo $ID$VERSION_ID)

curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

sudo apt-get update

sudo apt-get install -y nvidia-docker2

# 修改docker配置

sudo pkill -SIGHUP dockerd

sudo nano /etc/docker/daemon.json

# 配置文件样例，$端口号是你设定的端口号

{

"hosts": ["unix:///var/run/docker.sock", "tcp://0.0.0.0:$端口号"],

"registry-mirrors": ["https://******.mirror.aliyuncs.com"],

"default-runtime": "nvidia",

"runtimes": {

"nvidia": {

"path": "nvidia-container-runtime",

"runtimeArgs": []

}

# 主配置文件的启动参数和上述配置有冲突，重启docker会导致如下报错：

Job for docker.service failed because the control process exited with error code. See "systemctl status docker.service" and "journalctl -xe" for details.

# 因此，同时修改主配置文件

sudo nano /lib/systemd/system/docker.service

# 修改ExecStart参数为如下样式：

ExecStart=/usr/bin/dockerd

# -H fd:// --containerd=/run/containerd/containerd.sock

# 刷新配置，重启docker

systemctl daemon-reload

sudo systemctl restart docker

安装并配置好docker建议对虚拟机生成快照！

4、配置TLS远程管理，及Portainer接入

配置TLS远程管理是在Docker服务器（运行docker守护进程的服务器）进行配置，因此下面的$HOST是Docker服务器的IP！

4.1、修改openssl配置的CA部分

ubuntu 16.04 下 openssl 配置文件位置：/usr/lib/ssl/openssl.cnf，其他系统可参考

sudo su
nano /usr/lib/ssl/openssl.cnf
# 修改以下部分，路径仅作示例
[ CA_default ]
dir = /etc/openssl

# 创建相关目录和文件
mkdir /etc/openssl
cd /etc/openssl
mkdir -p {certs,private,tls,crl,newcerts}
# 注意serial要有数字，不然会报错
echo 00 > serial
touch index.txt

sudo su

nano /usr/lib/ssl/openssl.cnf

# 修改以下部分，路径仅作示例

[ CA_default ]

dir = /etc/openssl

# 创建相关目录和文件

mkdir /etc/openssl

cd /etc/openssl

mkdir -p {certs,private,tls,crl,newcerts}

# 注意serial要有数字，不然会报错

echo 00 > serial

touch index.txt

4.2、生成私钥并自签证书

# 生成CA私钥，必须设定一个CA私钥密码，该密码很重要，之后的步骤要输入
openssl genrsa -out private/cakey.pem -des 4096
# 返回如下内容
Generating RSA private key, 4096 bit long modulus
..............++
.........................................................................................................................++
e is 65537 (0x10001)
Enter pass phrase for private/cakey.pem:
Verifying - Enter pass phrase for private/cakey.pem:

# 用CA自签证书
openssl req -new -x509 -key private/cakey.pem -days 3650 -out cacert.pem
# 返回如下内容
# 执行后首先会要求输入CA私钥密码
Enter pass phrase for private/cakey.pem:
# 之后会要求填写一些信息，其中Common Name为必填项！$HOST是Docker服务器（运行docker守护进程的服务器）的IP
You are about to be asked to enter information that will be incorporated
into your certificate request.
What you are about to enter is what is called a Distinguished Name or a DN.
There are quite a few fields but you can leave some blank
For some fields there will be a default value,
If you enter '.', the field will be left blank.
-----
Country Name (2 letter code) [AU]:CN
State or Province Name (full name) [Some-State]:Beijing
Locality Name (eg, city) []:Beijing
Organization Name (eg, company) [Internet Widgits Pty Ltd]:BISTU
Organizational Unit Name (eg, section) []:PCIJLab
Common Name (e.g. server FQDN or YOUR name) []:$HOST
Email Address []:

# 生成CA私钥，必须设定一个CA私钥密码，该密码很重要，之后的步骤要输入

openssl genrsa -out private/cakey.pem -des 4096

# 返回如下内容

Generating RSA private key, 4096 bit long modulus

..............++

.........................................................................................................................++

e is 65537 (0x10001)

Enter pass phrase for private/cakey.pem:

Verifying - Enter pass phrase for private/cakey.pem:

# 用CA自签证书

openssl req -new -x509 -key private/cakey.pem -days 3650 -out cacert.pem

# 返回如下内容

# 执行后首先会要求输入CA私钥密码

Enter pass phrase for private/cakey.pem:

# 之后会要求填写一些信息，其中Common Name为必填项！$HOST是Docker服务器（运行docker守护进程的服务器）的IP

You are about to be asked to enter information that will be incorporated

into your certificate request.

What you are about to enter is what is called a Distinguished Name or a DN.

There are quite a few fields but you can leave some blank

For some fields there will be a default value,

If you enter '.', the field will be left blank.

-----

Country Name (2 letter code) [AU]:CN

State or Province Name (full name) [Some-State]:Beijing

Locality Name (eg, city) []:Beijing

Organization Name (eg, company) [Internet Widgits Pty Ltd]:BISTU

Organizational Unit Name (eg, section) []:PCIJLab

Common Name (e.g. server FQDN or YOUR name) []:$HOST

Email Address []:

4.3、颁发证书

# 生成要颁发证书的密钥文件
openssl genrsa -out private/docker.key 4096
# 返回如下内容
Generating RSA private key, 4096 bit long modulus
...................++
...................................................++
e is 65537 (0x10001)

# 生成证书请求
openssl req -new -key private/docker.key -days 3650 -out docker.csr
# 需要填写请求信息，其中Common Name为必填项！$HOST是Docker服务器（运行docker守护进程的服务器）的IP
You are about to be asked to enter information that will be incorporated
into your certificate request.
What you are about to enter is what is called a Distinguished Name or a DN.
There are quite a few fields but you can leave some blank
For some fields there will be a default value,
If you enter '.', the field will be left blank.
-----
Country Name (2 letter code) [AU]:CN
State or Province Name (full name) [Some-State]:Beijing
Locality Name (eg, city) []:Beijing
Organization Name (eg, company) [Internet Widgits Pty Ltd]:BISTU
Organizational Unit Name (eg, section) []:PCIJLab
Common Name (e.g. server FQDN or YOUR name) []:$HOST
Email Address []:
# 下面的密码和公司名都可以留空
Please enter the following 'extra' attributes
to be sent with your certificate request
A challenge password []:
An optional company name []:

# 颁发证书
openssl ca -in docker.csr -out certs/docker.crt -days 3650
# 会要求输入CA私钥密码
Using configuration from /usr/lib/ssl/openssl.cnf
Enter pass phrase for /etc/openssl/private/cakey.pem:
# 之后会显示相关信息，并要求确认
Check that the request matches the signature
Signature ok
Certificate Details:
        Serial Number: 0 (0x0)
        Validity
            Not Before: Jul  8 10:07:47 2020 GMT
            Not After : Jul  6 10:07:47 2030 GMT
        Subject:
            countryName               = CN
            stateOrProvinceName       = Beijing
            organizationName          = BISTU
            organizationalUnitName    = PCIJLab
            commonName                = $HOST
        X509v3 extensions:
            X509v3 Basic Constraints: 
                CA:FALSE
            Netscape Comment: 
                OpenSSL Generated Certificate
            X509v3 Subject Key Identifier: 
                9A:21:68:DD:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF
            X509v3 Authority Key Identifier: 
                keyid:19:D5:EF:76:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF

Certificate is to be certified until Jul  6 10:07:47 2030 GMT (3650 days)
Sign the certificate? [y/n]:y


1 out of 1 certificate requests certified, commit? [y/n]y
Write out database with 1 new entries
Data Base Updated

# 生成要颁发证书的密钥文件

openssl genrsa -out private/docker.key 4096

# 返回如下内容

Generating RSA private key, 4096 bit long modulus

...................++

...................................................++

e is 65537 (0x10001)

# 生成证书请求

openssl req -new -key private/docker.key -days 3650 -out docker.csr

# 需要填写请求信息，其中Common Name为必填项！$HOST是Docker服务器（运行docker守护进程的服务器）的IP

You are about to be asked to enter information that will be incorporated

into your certificate request.

What you are about to enter is what is called a Distinguished Name or a DN.

There are quite a few fields but you can leave some blank

For some fields there will be a default value,

If you enter '.', the field will be left blank.

-----

Country Name (2 letter code) [AU]:CN

State or Province Name (full name) [Some-State]:Beijing

Locality Name (eg, city) []:Beijing

Organization Name (eg, company) [Internet Widgits Pty Ltd]:BISTU

Organizational Unit Name (eg, section) []:PCIJLab

Common Name (e.g. server FQDN or YOUR name) []:$HOST

Email Address []:

# 下面的密码和公司名都可以留空

Please enter the following 'extra' attributes

to be sent with your certificate request

A challenge password []:

An optional company name []:

# 颁发证书

openssl ca -in docker.csr -out certs/docker.crt -days 3650

# 会要求输入CA私钥密码

Using configuration from /usr/lib/ssl/openssl.cnf

Enter pass phrase for /etc/openssl/private/cakey.pem:

# 之后会显示相关信息，并要求确认

Check that the request matches the signature

Signature ok

Certificate Details:

Serial Number: 0 (0x0)

Validity

Not Before: Jul 8 10:07:47 2020 GMT

Not After : Jul 6 10:07:47 2030 GMT

Subject:

countryName = CN

stateOrProvinceName = Beijing

organizationName = BISTU

organizationalUnitName = PCIJLab

commonName = $HOST

X509v3 extensions:

X509v3 Basic Constraints:

CA:FALSE

Netscape Comment:

OpenSSL Generated Certificate

X509v3 Subject Key Identifier:

9A:21:68:DD:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF

X509v3 Authority Key Identifier:

keyid:19:D5:EF:76:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF:FF

Certificate is to be certified until Jul 6 10:07:47 2030 GMT (3650 days)

Sign the certificate? [y/n]:y

1 out of 1 certificate requests certified, commit? [y/n]y

Write out database with 1 new entries

Data Base Updated

4.4、配置Docker使用TLS认证

# 转存生成的秘钥和证书
cd /etc/docker/
mkdir ssl
cd /etc/docker/ssl
cp /etc/openssl/cacert.pem ./ca.pem
cp /etc/openssl/newcerts/00.pem ./cert.pem
cp /etc/openssl/private/docker.key ./key.pem

# 修改配置文件
nano /etc/docker/daemon.json
# 配置文件样例，$端口号是你设定的端口号
{
    "tlsverify": true,
    "tlscacert": "/etc/docker/ssl/ca.pem",
    "tlscert": "/etc/docker/ssl/cert.pem",
    "tlskey": "/etc/docker/ssl/key.pem",
    "hosts": ["unix:///var/run/docker.sock", "tcp://0.0.0.0:$端口号"],
    "registry-mirrors": ["https://******.mirror.aliyuncs.com"],
    "default-shm-size": "32G",
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}
# 重启Docker
systemctl restart docker

# 转存生成的秘钥和证书

cd /etc/docker/

mkdir ssl

cd /etc/docker/ssl

cp /etc/openssl/cacert.pem ./ca.pem

cp /etc/openssl/newcerts/00.pem ./cert.pem

cp /etc/openssl/private/docker.key ./key.pem

# 修改配置文件

nano /etc/docker/daemon.json

# 配置文件样例，$端口号是你设定的端口号

{

"tlsverify": true,

"tlscacert": "/etc/docker/ssl/ca.pem",

"tlscert": "/etc/docker/ssl/cert.pem",

"tlskey": "/etc/docker/ssl/key.pem",

"hosts": ["unix:///var/run/docker.sock", "tcp://0.0.0.0:$端口号"],

"registry-mirrors": ["https://******.mirror.aliyuncs.com"],

"default-shm-size": "32G",

"default-runtime": "nvidia",

"runtimes": {

"nvidia": {

"path": "nvidia-container-runtime",

"runtimeArgs": []

}

# 重启Docker

systemctl restart docker

4.5、 Portainer接入

如果已经添加过服务器，只更新TLS认证配置，则只需要上传相应证书：

将/etc/docker/ssl/目录内的cert.pem和key.pem文件拷贝到本地，上传至Portainer。

如果是新服务器，按照下面步骤进行添加：

配置好TLS远程管理后建议对虚拟机生成快照！

参考资料

docker 配置 TLS 认证开启远程访问

Docker 守护进程+远程连接+安全访问+启动冲突解决办法 (完整收藏版)

0. 准备工作

0.1. 主机BIOS设置

0.2. 配置PCI设备直通

0.3. 上传系统安装镜像

1. 创建虚拟机

1.1. 设置虚拟机引导方式为EFI

1.2. 设置虚拟机高级配置参数

1.3. 修改虚拟机兼容性

2. 安装操作系统

3. 为虚拟机添加显卡直通

3.1. 启用内存预留

3.2. 添加显卡设备

4. 安装NVIDIA驱动和CUDA

4.1. 检查GPU

4.2. 禁用nouveau

4.3. 安装依赖包

4.4. 安装NVIDIA驱动

4.5. 安装CUDA

解决方案：

1、 显卡直通

1.1、 添加显卡直通

1.2、 修改虚拟机硬件配置参数

1.3、 启动虚拟机并配置基本环境

2、 安装显卡驱动

2.1、 下载驱动

2.2、 检查GPU

2.3、 禁用nouveau

2.3、 安装显卡驱动

2020年6月28日补充

2.4、安装CUDA

3、 安装Docker

3.1、 安装Docker

3.2、 安装nvidia-docker

4、 配置TLS远程管理，及Portainer接入

4.1、 修改openssl配置的CA部分

4.2、 生成私钥并自签证书

4.3、 颁发证书

4.4、 配置Docker使用TLS认证

4.5、 Portainer接入

参考资料

其他操作

1、显卡直通

1.1、添加显卡直通

1.2、修改虚拟机硬件配置参数

1.3、启动虚拟机并配置基本环境

2、安装显卡驱动

2.1、下载驱动

2.2、检查GPU

2.3、禁用nouveau

2.3、安装显卡驱动

3、安装Docker

3.1、安装Docker

3.2、安装nvidia-docker

4、配置TLS远程管理，及Portainer接入

4.1、修改openssl配置的CA部分

4.2、生成私钥并自签证书

4.3、颁发证书

4.4、配置Docker使用TLS认证