Hi,
Just wondering about something in general: From time to time I get a yum error about trying to update something like an nvidia kernel module not being possible because there's no kernel available to go along with that version.
Trying several different mirrors doesn't help, and I can't be bothered to go trying them all out. I'll just wait a few days for things to settle.
But how have they compiled their module for a kernel that isn't available? Surely that's a bit chicken and eggish? Wouldn't they be compiling against a customised Fedora kernel, rather than one from somewhere else? And, if so, why don't we have the same one available to us?
Tim wrote:
Hi,
Just wondering about something in general: From time to time I get a yum error about trying to update something like an nvidia kernel module not being possible because there's no kernel available to go along with that version.
Trying several different mirrors doesn't help, and I can't be bothered to go trying them all out. I'll just wait a few days for things to settle.
But how have they compiled their module for a kernel that isn't available? Surely that's a bit chicken and eggish? Wouldn't they be compiling against a customised Fedora kernel, rather than one from somewhere else? And, if so, why don't we have the same one available to us?
I got that Nvidia update that failed when it could not find the needed kernel. I was not worried because I have switched to the Nvidia bash file which makes a kernel module when needed. It IS very slick. So I just watch for kernel updates and if one breaks Nvidia I will make it a module.
On Tue, 2007-11-06 at 15:46 -0700, Karl Larsen wrote:
I got that Nvidia update that failed when it could not find the needed kernel. I was not worried because I have switched to the Nvidia bash file which makes a kernel module when needed. It IS very slick. So I just watch for kernel updates and if one breaks Nvidia I will make it a module.
'twas more a general how and why, rather than getting Nvidia updated immediately...
Having said that, I don't really want to use the NVidia binary installer. I've been down the route before, and discovered what others had said about it: It messed up standard Xorg files to suit itself, making it very hard to stop using the NVidia driver if you had to (for problem solving, or using other video cards). I'm not talking about it customising xorg.conf, I mean it really did mess with other standard xorg files. It was quite a palaver to get a working Xorg again, afterwards.
Somewhere there's a page about why not to use the NVidia binary installer, that explained that all rather well, but I don't recall the address.
Tim wrote:
Hi,
Just wondering about something in general: From time to time I get a yum error about trying to update something like an nvidia kernel module not being possible because there's no kernel available to go along with that version.
Trying several different mirrors doesn't help, and I can't be bothered to go trying them all out. I'll just wait a few days for things to settle.
But how have they compiled their module for a kernel that isn't available? Surely that's a bit chicken and eggish? Wouldn't they be compiling against a customised Fedora kernel, rather than one from somewhere else? And, if so, why don't we have the same one available to us?
It is possible that they grabbed the kernel out of testing to build the module, so that it was ready when the kernel moved into updates. It is also possible that some of the mirrors were out of sync...
I forgot how many complaints I have seen on the list about a specific kernel module not being available when the new kernel comes out, so maybe this is an indication of the non-redhat repos trying to get around that problem...
Mikkel
Tim wrote:
Just wondering about something in general: From time to time I get a yum error about trying to update something like an nvidia kernel module not being possible because there's no kernel available to go along with that version.
Trying several different mirrors doesn't help, and I can't be bothered to go trying them all out. I'll just wait a few days for things to settle.
But how have they compiled their module for a kernel that isn't available? Surely that's a bit chicken and eggish? Wouldn't they be compiling against a customised Fedora kernel, rather than one from somewhere else? And, if so, why don't we have the same one available to us?
You *might* be experiencing this reported bug:
https://bugzilla.redhat.com/show_bug.cgi?id=330711
Summary: kernel pacakges that remain installed seem to be ignored in yum transaction check
On Wed, 2007-11-07 at 09:05 +1030, Tim wrote:
Hi,
Just wondering about something in general: From time to time I get a yum error about trying to update something like an nvidia kernel module not being possible because there's no kernel available to go along with that version.
Trying several different mirrors doesn't help, and I can't be bothered to go trying them all out. I'll just wait a few days for things to settle.
But how have they compiled their module for a kernel that isn't available? Surely that's a bit chicken and eggish? Wouldn't they be compiling against a customised Fedora kernel, rather than one from somewhere else? And, if so, why don't we have the same one available to us?
-- (This computer runs FC7, my others run FC4, FC5 & FC6, in case that's important to the thread.)
Don't send private replies to my address, the mailbox is ignored. I read messages from the public lists.
Next time this happens to you. Run a "yum clean all" then try updating again. This is what works for me.
On Tue, 2007-11-06 at 22:58 -0600, Adam Hough wrote:
Next time this happens to you. Run a "yum clean all" then try updating again. This is what works for me.
I do a yum clean metadata if the repos are a bit doolally, I'm not too keen on the shotgun approach, and having to re-get many megs over again.
In this case, things are working a day or two later.
Karl Larsen wrote:
Tim wrote:
Hi,
I got that Nvidia update that failed when it could not find the needed kernel. I was not worried because I have switched to the Nvidia bash file which makes a kernel module when needed. It IS very slick. So I just watch for kernel updates and if one breaks Nvidia I will make it a module.
Sounds like dkms that freshrpms uses. What I use now.