Issue while Installing of Media Server Studio Essential on CentOS

[Kapil_D_]

Hi,

I am trying to install intel media server studio essential 2015 R5 on the latest version/build of CentOS (3.10.0.229.4.2). I am following the steps provided in the getting_started_guide, but after the "rpm -Uvh kernel-3.10.*.rpm" step, I am not able to reboot the system.  During reboot, I am getting the following error: "file /vmlinuz-3.10....... not found. Error: you need to load the kernel first". I repeated the process 3 times and got the same issue. It seems that  there is an issue with the kernel installation step. 

Could anyone tell me how to fix this problem? Specifically, I am following these steps:

Processor type: Intel Core i7-4790K

1. OS installation: Installed CentOS 7 with "Development and Creative Workstation" base environment and the following add-ons: additional development, compatibility libraries, development tools, hardware monitoring utilities, office suite and productivity, and Python. This installs kernel 3.10.0.*.x86_64 of the CentOS with root and "admin" user.

2. Installing SDK UMD:

# su

# ./install_sdk_UMD_CentOS.sh : successful

# mkdir /MSS

# chown admin:admin /MSS

# exit

3. Building the kernel:

# cp build_kernel_rpm_CentOS.sh /MSS

# cd /MSS

Made following changes in build_kernel_rpm_CentOS.sh file:

INTEL_MSS_RELEASE_NUM=5 (original 4)

CENTOS7_LATEST_KER_BID=229.4.2 (original 229.1.2)

I also tried without making any change to this file, but the end result is still the same: not able to reboot!

# ./build_kernel_rpm_CentOS.sh ; the build step seems to be successful 

4. Installing the newly built kernel:

# su

# cd /MSS/rpmbuild/RPMS/x86_64 : there are 3 rpm files in this directory:

kernel-3.10.*.rpm,  kernel-devel.*.rpm, kernel-headers.*.rpm

# rpm -Uvh kernel-3.10.*rpm : this command (as suggested in the start guide) probably installs only the first one. I tried to install all three also in my third attempt, but the end result is the same

# reboot

 While rebooting, I get stuck with the following error: "file /vmlinuz-3.10....... not found. Error: you need to load the kernel first".

Please let me know what is the possible cause of this problem and how can I resolve it.

Thanks and Regards,

Kapil

[Surbhi_M_Intel]

Hi Kapil, 

For Media Sever Studio R5, the official supported is CentOS 7.1 & SLES and I believe you have CentOS 7.1 installed with "Development and Creative Workstation" base environment.Can you please check the o/p of the following cmd-  lspci –nn and ensure the o/p matches with the one mentioned in the getting started started guide.

In the original kernel mode installation script I see CENTOS7_LATEST_KER_BID=229.1.2 (instead of 229.4.2). Thanks for pointing at R4 in the installation script, there is a mismatch in numbering but that shouldn't affect the installation. Please check if /opt/intel/mediasdk was populated after user mode installation. 

For kernel mode installation, only this kernel-3.10.*.rpm needs to be installed. and please keep in mind that kernel rpm should be run from a short path(mentioned in the installation guide). kernel installation takes a while in general to install as compared to user mode, if it was quick then it might not be fully installed. Can you please check that? You can check if kernel is installed by this cmd - uname -r, the o/p should have MSS R4 mentioned(for e.g. 3.10.0-229.1.2.36242.MSSr4.el7.centos.x86_64). If you see that then problem will be in booting up with the right kernel. There is lot of help available online to do so like using GRUB boot loader. Generally, at the time of booting up you have an option to choose kernel. 

If you still experience an issue, I would suggest to do fresh CentOS7.1 & MSS installation. 
Hope that helps. 
 

Thanks,
Surbhi

[Kapil_D_]

Hi Surbhi,

Thanks a lot for the quick response!

Based on your suggestion, I installed the CentOS 7 again, hoping that it would minimize any future issues but I am not able to install the SDK. 

Please find below the details of the step that I followed this time.

- Chose "Development and Creative Workstation" base environment and only "development tools" add-on during the OS installation. The base kernel name is now 3.10.0-229.el7.x86_64. 

- Checked VGA adapter by lspci -nn -s 00:02.0 and I see the graphics device from Intel. The output of the command is

"00:02.0 VGA compatible controller [0300]: Intel Corporation Xeon E3-1200 v3/4th Gen Core Processor Integrated Graphics Controller [8086:0412] (rev 06)"

So, I think this part is good since I have Haswell i7-4790K (Haswell processor) with 4600 HD graphics. Now the installation part:

===================================================================

User Mode Installation step: running ./install_sdk_UMD_CentOS.sh file

After the user mode installation step: the /opt/intel/mediasdk has following directories: doc, include, lib, lib64, opensource, plugins, and tools. So, I assume that this step went fine.

===================================================================

Kernel building step: running ./build_kernel_rpm_CentOS.sh file from /MSS directory

This time I kept everything to default, i.e.: INTEL_MSS_RELEASE_NUM=4 and CENTOS7_LATEST_KER_BID=229.1.2. This step took more than 30 mins and ended with the following message: 

+exit 0

/MSS

INFO... Please find kernel rpm under ./rpmbuild/RPMS folder.

So, I think this step went successful. I see 3 rpm files kernel-3.10.0-229.1.2.36242.MSSr4

===================================================================

Installing the kernel step:

# cd /MSS/rpmbuild/RPMS/x86_64
# rpm -Uvh kernel-3.10.*.rpm

I have the same same directory path as suggested in the getting_started_guide. I believe it is short enough. I see the following output of the command:

"[root@ultra4 x86_64]# rpm -Uvh kernel-3.10.*.rpm
Preparing...                          ################################# [100%]
Updating / installing...
   1:kernel-3.10.0-229.1.2.36242.MSSr4################################# [ 50%]
Cleaning up / removing...
   2:kernel-3.10.0-229.el7            warning: file /lib/modules/3.10.0-229.el7.x86_64/weak-updates: remove failed: No such file or directory
warning: file /lib/modules/3.10.0-229.el7.x86_64/modules.softdep: remove failed: No such file or directory
warning: file /lib/modules/3.10.0-229.el7.x86_64/modules.devname: remove failed: No such file or directory
################################# [100%]"

This step took less than 1 min and finished with the above warnings (no error though). Now, when I run uname -r command, I see 3.10.0-229.el7.x86_64 version. This is before rebooting the system. Do you think this step is an issue?

===================================================================

I haven't rebooted the system this time because I am afraid that the kernel is not installed properly and the system will not boot. Could you help me with how to proceed from here?

Thanks,

Kapil

PS: Earlier, I set the variable CENTOS7_LATEST_KER_BID to 229.4.2 instead of 229.1.2 because I thought that the base kernel version might have changed to 229.4.2 after the 2nd command in ./install_sdk_UMD_CentOS.sh file: (yum -y -t install kernel-headers kernel-devel). In particular, I notice that this yum command installs 3.10.0-229.4.2.el7 version of kernel-devel package and also updates the kernel-headers package to  3.10.0-229.4.2.el7 version. As I mentioned above, this time I reinstalled everything with the exact same commands mentioned in the guide.

[Kapil_D_]

Hi Surbhi,

Just wanted to tell you that after the "rpm -Uvh kernel-3.10.*.rpm" step, I see that following directory is created.

ls /lib/modules/3.10.0-229.1.2.36242.MSSr4.el7.centos.x86_64/

But this was created earlier too because when I reboot, I see "3.10.0-229.1.2.36242.MSSr4.el7.centos.x86_64" in the choices during the kernel selection step. But, when I choose this kernel, I end up getting the above mentioned error, i.e.: "file /vmlinuz-3.10....... not found. Error: you need to load the kernel first"

Thanks,

Kapil

[Surbhi_M_Intel]

Are you using CentOS 7 or CentOS 7.1?

-Surbhi

[Kapil_D_]

I believe its 7.1. Sorry, I mentioned 7 in the above thread. Actually, on the default desktop it has a logo saying CentOS 7, instead of 7.1, thats why I said 7.

I downloaded the CentOS ISO from: http://isoredirect.centos.org/centos/7/isos/x86_64/CentOS-7-x86_64-DVD-1503-01.iso

When I check the installed OS version on my system and it says:

==============================================

[root@ultra4 MSS]#  cat /etc/*release*
CentOS Linux release 7.1.1503 (Core)
Derived from Red Hat Enterprise Linux 7.1 (Source)
NAME="CentOS Linux"
VERSION="7 (Core)"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="7"
PRETTY_NAME="CentOS Linux 7 (Core)"
ANSI_COLOR="0;31"
CPE_NAME="cpe:/o:centos:centos:7"
HOME_URL="https://www.centos.org/"
BUG_REPORT_URL="https://bugs.centos.org/"

CENTOS_MANTISBT_PROJECT="CentOS-7"
CENTOS_MANTISBT_PROJECT_VERSION="7"
REDHAT_SUPPORT_PRODUCT="centos"
REDHAT_SUPPORT_PRODUCT_VERSION="7"

CentOS Linux release 7.1.1503 (Core)
CentOS Linux release 7.1.1503 (Core)
cpe:/o:centos:centos:7

==============================================

Thanks,

Kapil

[Kapil_D_]

Hi Surbhi,

Please let me know if you have come across a solution to my problem. Also, it will be great if you could point me to a document or instructions (other than getting_started_guide) that has more details on the installation process.

Thanks,

Kapil

[Surbhi_M_Intel]

Hi Kapil,

I wasn't able to replicate the issue you have reported but I have escalated it. I will get back to you as soon as I have an answer. 
It seems either kernel installation script didn't successfully installed the kernel or may be the loader(grub) was failed. 

Thanks,
Surbhi

[Kapil_D_]

Thanks a lot Surbhi!

If it helps, here is the exact error that I see during the reboot:

"error: /vimlinuz-3.10.0.229.1.2.36242.MSSr4.el7.centos.x86_64 has invalid signature.

error: you need to load the kernel first"

So, as you said, most probably the kernel installation step is not happening successfully.

Since I rebooted the system and crashed the OS again, I will have to reinstall the OS again:-) I will do it in the mean time. I am wondering if there is a way to install the newly built kernel alongside the old one. That way I will not have to install the OS again:)

I am wondering if I can use rpm -ivh kernel.*rpm command to install all all three rpm files instead of using rpm -Uvh kernel-3.10.*.rpm, which is probably overwriting/updating some of the files in the existing kernel?

Separately, I want to ask you whether after installing the media server studio essential, I will be able to write and execute any OpenCL program (implemented using OpenCL 1.2 APIs and specifications) on both GPU and CPU devices?

In addition to the sample programs provided in the sample kit, I would also like to evaluate some standard benchmarks, like Rodinia benchmarks (http://www.cs.virginia.edu/~skadron/wiki/rodinia/index.php/Rodinia:Accelerating_Compute-Intensive_Applications_with_Accelerators), SHOC (https://github.com/vetter/shoc/wiki), OpenDwarfs (https://github.com/opendwarfs), and Parboil (http://impact.crhc.illinois.edu/Parboil/parboil.aspx). I am hoping that I will be able to run most of typical OpenCL benchmarks. 

Please let me know if my understanding is not correct.

Thanks again,

Kapil

[Kapil_D_]

OK. So, I have a good news.

Finally, I am able to boot from the newly built kernel by doing the following changes in the BIOS settings: under the Boot/secure boot menu, the default OS type was "Windows WEFI mode". I changed it to "Other OS". May be you want to include this in the installation guide...

Now, the uname -a reports the correct version of the kernel.

I would like to get your comments on my other questions about running OpenCL benchmarks from different sources on both CPU and GPU devices.

Thanks,

Kapil

[Surbhi_M_Intel]

Hi Kapil, 

Great it worked! Thanks for the suggestion, we will include this as BKM for the installation. 
Can you please start an another thread for your OpenCL question, so that someone else can learn about the installation here & won't get confused with OpenCL discussion?

Thanks,
Surbhi 

[Kapil_D_]

Sure.

Thanks again for the help!

-Kapil