[SOLVED] I can't rename a kernel (package level) so where do I find old kernels with the regular names ?

[aegersz]

Hi,

My host won't allow me to boot from the latest kernel and he has an outdated list of kernels to choose from.

I tried to install an old one (on his list) but could only found it at the "vault" (embedded with "centos.plus") and the boot failed because it couldn't find the right kernel, as i suspected.

so i'd rather (yum) remove it, rename it somehow and then install it again.

i tried a simle rename but yum couldn't find it so is there any possible way to do it ?

a manual rename of the /boot and /lib files seems crazy as an effective alternative so do i have another option ?

[aegersz]

I discovered these two tools:

1. Extract cpio archive from RPM Package Manager (RPM) package
rpm2cpio name.rpm > output-file

2. A tool to build an rpm package from an installed package
rpmrebuild -ep name.rpm

whilst in SPEC file edit mode (make any changes here), open another windows and to to the directory to see all the files required for a rebuild
(and you may modify them). then go back to your original window and save the file. you will be prompted to continue and it will then tell your
where your new rpm is.

BUT it is too much trouble ... maybe there is a correctly name archive of this somewhere ?

[aegersz]

I only ask as my VPS Host has a specfic list of named kernels that can't be altered so I am FORCED into keeping the name the same. Centosplus just doesn't work.

any thoughts ? I will close this due to lack of interest if i don't hear from anybody soon.

it just means living with an old 2013 kernel until my list of yum preserved kernels catches up but it may take years

[avij]

I would migrate to some other VPS provider if I was you..

It may also be possible that the kernel is provided by your hoster, and whatever you do inside your own VPS won't affect the kernel at all, even if you find a kernel with a matching name and try to install it.

What kind of VPS is that? Please provide the output of uname -a (with the hostname edited out if you wish) and perhaps also virt-what (you may need to yum install virt-what first).

[aegersz]

not so. I asked him to move to a "later" kernel and renamed the /boot entries to match his latest (still quite old) but my system couldn't even start iptables because the /lib files and directories were missing as they were still pointing to the ones that i didn't know i had to rename. Too much trouble to fix so we simply fell back to the old one.

i can't supply uname info at this time but it's a very current CentOS 6.9 base system running a large list of external repos using most recent versions of postfix, mysql, php etc. it's a tiny 1 GB/30 GB SSD/2 NIC system and i have tuned it for the best possible performance based on the limited resources that i have. it's a free site running free everything (thank you Simple Machines and CentOS).

I will NOT change hosts because i love my host and he permits me to run a website that is EXTREMELY sensitive - it is https://forum.drugs-and-users.org

yes, he (my host) controls the kernels and all i have to do is ignore grub and ensure that i keep enough kernels (via yum.conf) that match his. Despite running the latest everything with a very current base, my kernel is stuck in 2013 but it is not Dirty COW vulnerable yet i have put in the recommended mitigation.

i have been working on the firewall, postfix and sshd hardening for years so nobody is going to hack me, take the database info and try to blackmail my member base or Law Enforcement can't get to the data without a subpoena and we give them no reason to do that by staying very close to the law !

[avij]

So that is a Digital Ocean machine. I have never used them, but perhaps you should read their fine manual for updating the kernel.

For Droplets that use the in-control panel legacy system, a special kernel called DigitalOcean GrubLoader is now available and recommended. This allows Droplets with the legacy management system to boot into internally installed kernels.

sounds interesting. Then there's a section titled "Booting to a Non-Default Kernel on Grub 1 Distributions", which sounds like it might work.

If you need additional assistance, please leave a comment below or open a support ticket in the control panel.

at the bottom of the page is also something that you should keep in mind. Their boot configuration is not standard CentOS, and Digital Ocean should know best how to proceed in this situation.

[aegersz]

I am not sure as the kernel management is handled by Singapore.

My host says that there is NO way to override the nominated kernels and i'm not going to pressure him any more.

I will just simply wait for our list of kernels to intersect at some point in time so am now saving as many as I can.

Thanks for your help !

[avij]

Ok, as you see fit.

[TrevorH]

I'd change hosting companies ASAP.

[aegersz]

i can't - it's like a marriage and i like them but thanks for the suggestion !