Issue of installing cuda toolkit on Redhat 6.5

I followed strictly nvidia guidelines as When I execute the comman:sudo yum install cuda, I had a fatal error saying that jre 1.7 is required, however I DID install java 1.7 and can display correct java version info using ‘java -version’ command. I have struggled for the whole day… Anybody can help me? Thank you in advance.
btw, I used rpm installation method.
error msg:
—> Package xorg-x11-drv-nvidia-gl.x86_64 1:352.39-1.el6 will be installed

—> Package xorg-x11-drv-nvidia-libs.x86_64 1:352.39-1.el6 will be installed

–> Finished Dependency Resolution

Error: Package: cuda-visual-tools-7-5-7.5-18.x86_64 (cuda-7-5-local)

Requires: jre >= 1.7
You could try using --skip-broken to work around the problem
Found 6 pre-existing rpmdb problem(s), ‘yum check’ output follows:
java_cup-0.10k-5.el6.x86_64 has missing requires of java-gcj-compat
1:java_cup-0.10k-5.el6.x86_64 has missing requires of java-gcj-compat

jna-3.2.4-2.el6.x86_64 has missing requires of java
libvirt-java-0.4.9-1.el6.noarch has missing requires of java >= (‘0’, ‘1.5.0’, None)

sinjdoc-0.5-9.1.el6.x86_64 has missing requires of java-gcj-compat >= (‘0’, ‘1.0.70’, None)

sinjdoc-0.5-9.1.el6.x86_64 has missing requires of java-gcj-compat >= (‘0’, ‘1.0.70’, None)

[admin@localhost bin]$ java -version
java version “1.7.0_80”

Java™ SE Runtime Environment (build 1.7.0_80-b15)

Java HotSpot™ 64-Bit Server VM (build 24.80-b11, mixed mode)

[admin@localhost bin]$

It’s very likely that the package manager wants to see the jre 1.7 PACKAGE installed, not just that you have jre 1.7 installed (using some other method).

You can fiddle with

yum list java*

to find out what packages are available. Then choose some appropriate ones to install.

Alternatively try using the runfile installer method. You may need to clean out the package manager install or start over with a clean load of RHEL 6.5

Thank you very much, txbob! You saved me! I successfully installed cuda toolkit 7.5 on RHEL 6.5 after following your advice to install another OpenJDK 1.7.

However after I reboot the server, it hanged in the boot phase. However I can enter CLI mode using CTRL + ALT + F2.

The gpu card is K40c, OS is RHEL 6.5, toolkit is 7.5 and cuda installation method is rpm.

Could anybody help me? Thanks!

btw, in CLI mode, I tested nvidia-msi can work properly. When I try startx command, it said: no device detected, no screens found. Should I blacklist some packages such as nouveau? Please advise, thank you in advance!

I tested lsmod | grep nouveau, nothing appeared, so I think nouveau has been blacklisted already…

just update the progress: the X-window can start after I renamed /etc/X11/xorg.conf. So far the issue was solved.

It’s because you used the package manager installation, and your X display was not hosted by the NVIDIA GPU.

To work around this, I recommend using the runfile installer method, and use the command line switch --no-openg-libs when running the runfile installer

note that the runfile installer and package manager methods are not compatible, so you should start with a clean OS load and choose one or the other

Thank you txbob! As the gpu server is off-line and it took me a LONG time to find and install many dependencies and meanwhile x window works well now, I decided not to reinstall cuda. However I will follow your advice when installing another server.