Having a "fun" time getting centos/rhel 8 to use grub
I have several Linodes that are now running RHEL8 and for the life of me am having issues getting grub2 to work.
I followed the guide here (https://www.linode.com/docs/guides/managing-the-kernel-on-a-linode/) to no luck. I can see that I am coming to the grub prompt via lish.
And yes, on each node the rhel kernel is indeed installed. I reinstalled it just in case the rpm scriptlets needed to do something I wasn't manually doing. I also reissued the grub2-mkconfig command as well.
Tossing my hands in the air on this one. Initially I thought that it was the one node in particular that was goofy and causing me grief, but now a second rig, and likely soon a third will be in the same boat.
What steps are required/what am I missing?
Thanks!
Andrew.
4 Replies
MBraley
Linode Staff
Hi Andrew,
That sounds like a very frustrating mess. Grub and I have had our many disagreements in the past as well.
There is some information that might be of help in this situation. Linode also has a guide that details the usage of Rescue and Rebuild. That should be able to provide a "backdoor" into the RHEL instance, using that, you can try to reconfigure grub. I pulled up the guide that you referenced and I wanted to highlight something that might be what you are facing (I can't see your system unfortunately!) There is a known problem with Asynchronous SCSI scans. I don't know if you pulled down a Linode kernel or a kernel from another mirror.
Also, I would imagine the process to be similar/the same for CentOS and Fedora, in the case they are not, I found Red Hat's Changing kernel command-line parameters for all boot entries provides some pretty in depth instructions.
Thanks for your time @MBraley!
To answer your comment about kernel and methods -- same across all RH family and clones. And the kernels used are either yours (currently in use) or the official RH one (preferred). I can use the offical Linode Latest kernel to get in and muck around, so the rescue isn't needed.
I don't have time RIGHT now to try the async scsi scan issue remedy, but that will be my first attempt.
TY,
Andrew.
MBraley
Linode Staff
Andrew,
Thanks for the reply. When you have the time, I would like to hear what you do end up finding. If the async scsi fixes work for you, or maybe something else is going on. It is interesting that this is occurring across all of your RH deployments.
Out of curiosity - do you typically take the more hands on approach with upgrading your system, or let the distribution handle it? Are there any other tweaks or changes that you have used to customize your installations? It would be nice to pinpoint this down to one specific cause.
Whenever you get around to it, feel free to drop back in with updates.
When it came time to upgrade my home server, I tried using the leapp tool RH put out, and it worked great. I had an old VM with CentOS on it that I cloned and tried their migration tool to move over to RHEL and updated that with leapp and it worked flawlessly.
On my 3 RH VPS nodes with you guys, I have moved from C7 to RHEL7 to RHEL8 using the aforementioned tools. Thankfully the leapp tool told me of potential issues and I was able to remedy them in order to have a clean upgrade.
If I can iron this issue out, I plan on getting everything upgraded over the next year to RHEL9, perhaps using Rocky/Alma and then the conversion tool (IIRC its called convert2rhel).
As far as tweaks/customizations -- not much anymore. Stock RHEL and its repos tend to take care of my needs quite nicely. And running most things with podman makes migration and management a breeze. One VPS is still running with a most ugly arrangement, that works and is secure, but nonetheless ugly. One of two things I need to get containerized. That being synapse, the matrix homeserver. The other being apache fully managed via containers -- httpd, reverse proxy and letsencrypt.
I suppose this would be a fantastic time to ask again for RHEL images to be available. Would save me alot of hassle and time.