[elrepo] Messages during install of kmod-nvidia

[Phil Perry]

On 22/11/17 09:54, Kretschmer, Jens wrote:
> Hi,
> thank you for your response!

No problem. Please see my replies inline below.

> <redacted>[root]> rpm -qa kernel\*
> kernel-devel-3.10.0-693.5.2.el7.x86_64
> kernel-headers-3.10.0-693.5.2.el7.x86_64
> kernel-tools-3.10.0-693.5.2.el7.x86_64
> kernel-3.10.0-693.5.2.el7.x86_64
> kernel-tools-libs-3.10.0-693.5.2.el7.x86_64
> <redacted>[root]> find /lib/modules -name nvidia*\.ko
> /lib/modules/3.10.0-693.el7.x86_64/extra/nvidia/nvidia.ko
> /lib/modules/3.10.0-693.el7.x86_64/extra/nvidia/nvidia-drm.ko
> /lib/modules/3.10.0-693.el7.x86_64/extra/nvidia/nvidia-modeset.ko
> /lib/modules/3.10.0-693.el7.x86_64/extra/nvidia/nvidia-uvm.ko
> /lib/modules/3.10.0-693.5.2.el7.x86_64/weak-updates/nvidia/nvidia.ko
> /lib/modules/3.10.0-693.5.2.el7.x86_64/weak-updates/nvidia/nvidia-drm.ko
> /lib/modules/3.10.0-693.5.2.el7.x86_64/weak-updates/nvidia/nvidia-modeset.ko
> /lib/modules/3.10.0-693.5.2.el7.x86_64/weak-updates/nvidia/nvidia-uvm.ko

Great, that all looks good, exactly as expected.


> I received the following output when uninstalling and installing 
> kmod-nvidia. I did this procedure twice, with the same error messages. 
> We received the same error messages when doing the update to kmod-nvidia 
> 384.98-1.el7_4.elrepo and when trying to reinstall kmod-nvidia. So yes, 
> it is reproducible.

Thanks for performing that test.

<snip>

> It looks like your suspicion is correct:
> <redacted>[root]> grep kmod-nvidia yum.log
> Oct 23 09:45:58 Installed: kmod-nvidia-384.90-1.el7_4.elrepo.x86_64
> Nov 05 10:04:16 Updated: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 15 10:11:29 Installed: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 15 10:20:30 Installed: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 22 09:56:59 Erased: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 22 10:00:33 Installed: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 22 10:02:04 Erased: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> Nov 22 10:05:43 Installed: kmod-nvidia-384.98-1.el7_4.elrepo.x86_64
> <redacted>[root]> grep "dracut: Executing:" /var/log/messages*
> /var/log/messages:Nov 22 09:57:06 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.5.2.el7.x86_64.tmp 
> 3.10.0-693.5.2.el7.x86_64
> /var/log/messages:Nov 22 09:57:38 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.el7.x86_64.tmp 
> 3.10.0-693.el7.x86_64
> /var/log/messages:Nov 22 09:59:46 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.5.2.el7.x86_64.tmp 
> 3.10.0-693.5.2.el7.x86_64
> /var/log/messages:Nov 22 10:00:32 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.el7.x86_64.tmp 
> 3.10.0-693.el7.x86_64
> /var/log/messages:Nov 22 10:00:41 <redacted> dracut: Executing: 
> /usr/bin/dracut --add-drivers nvidia -f 
> /boot/initramfs-3.10.0-693.5.2.el7.x86_64.img 3.10.0-693.5.2.el7.x86_64
> /var/log/messages:Nov 22 10:02:11 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.5.2.el7.x86_64.tmp 
> 3.10.0-693.5.2.el7.x86_64
> /var/log/messages:Nov 22 10:02:43 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.el7.x86_64.tmp 
> 3.10.0-693.el7.x86_64
> /var/log/messages:Nov 22 10:04:58 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.5.2.el7.x86_64.tmp 
> 3.10.0-693.5.2.el7.x86_64
> /var/log/messages:Nov 22 10:05:43 <redacted> dracut: Executing: 
> /sbin/dracut -f /boot/initramfs-3.10.0-693.el7.x86_64.tmp 
> 3.10.0-693.el7.x86_64
> /var/log/messages:Nov 22 10:05:52 <redacted> dracut: Executing: 
> /usr/bin/dracut --add-drivers nvidia -f 
> /boot/initramfs-3.10.0-693.5.2.el7.x86_64.img 3.10.0-693.5.2.el7.x86_64

Yes, that looks pretty conclusive to me, except I'm unable to reproduce 
it on my system when I remove kernel-3.10.0-693.el7.x86_64 from the 
system. Lets see if we can figure out why it feels the need to run 
dracut on a kernel that is not installed on your system.

Please could you show me the versions installed for the following:

rpm -q kmod
kmod-20-15.el7_4.4.x86_64

rpm -q dracut
dracut-033-502.el7.x86_64

and the listing for the /boot directory (I'm looking for any reference 
to the offending kernel-3.10.0-693.el7.x86_64):

ls /boot

It would appear on your system that the weak-modules script is calling 
dracut on the kernel that the kmod package was built against, in this 
case kernel-3.10.0-693.el7.x86_64, which generates the errors/warnings 
above when said kernel isn't present on the system.

I'm pretty sure you can ignore the warnings on your system. The 
important thing is that the package is installed correctly on your 
installed kernel(s), and that is clearly the case, so the rest may be 
considered noise.

Installing kernel-3.10.0-693.el7.x86_64 should presumably silence these 
warnings. That might be a useful test to perform (after the test below).

Further, I wonder if I might ask you to try installing another one of 
our kmod packages on your system to see if it generates similar errors 
to determine if it is all kmod packages (hence likely a bug in 
weak-modules) or if it is something specific kmod-nvidia and/or your 
system. A suitable candidate would be kmod-hellop which is a simple 
"hello world" proof of principle test package located in the elrepo 
testing repository:

yum --enablerepo=elrepo-testing install kmod-hellop

I would expect this to generate similar errors on your system, albeit 
against kernel-3.10.0-123.el7.x86_64. Feel free to uninstall immediately 
afterwards.

Thanks

Phil

PS - We can rule out nvidia-x11-drv as the offending party as it calls 
dracut with the "--add-drivers nvidia" parameter and your logs clearly 
show this is only run against kernel-3.10.0-693.5.2.el7.x86_64, as intended.