SERIOUSLY BAD ADVICE IN APTITUDE

[BloodyIron]

Sorry for the caps title, this just really seems like a FUCKING HORRIBLE IDEA.

In my other thread I downgraded my environment from kernel 4.2.8-1-pve to 4.2.2-1-pve to address stability issues.

Now, when I did "apt-get purge pve-kernel-4.2.8-1-pve" it also removed proxmox-ve*, except not entirely.

Then aptitude said a whole bunch of packages were no longer needed, some pve related. Dumb me though this was okay, as I had upgraded from 3.4 to 4.2, and I thought "oh well maybe they changed what packages are really needed now, I'll autoremove these".

NOPE. HORRIBLE IDEA. My webgui for the cluster started dropping off the planet.

I did an apt-get install proxmox-ve again, and it might be okay, but I don't know yet.

EITHER WAY THIS IS SERIOUSLY BAD ADVICE. WHAT THE FUCK.

 

[BloodyIron]

Okay so this very issue has resulted in me having to rebuild my whole cluster, because when I just wanted to remove a specific kernel it removed the whole PVE environment with it. WHY I DONT KNOW, BUT ITS INFURIATING.

This is seriously NOT okay behavior! I just wanted to roll back the kernel and remove the newer one.

DEVS PLEASE DONT DO THIS!

 

[tom]

Sorry for the caps title, this just really seems like a FUCKING HORRIBLE IDEA.

In my other thread I downgraded my environment from kernel 4.2.8-1-pve to 4.2.2-1-pve to address stability issues.

Now, when I did "apt-get purge pve-kernel-4.2.8-1-pve" it also removed proxmox-ve*, except not entirely.

Then aptitude said a whole bunch of packages were no longer needed, some pve related. Dumb me though this was okay, as I had upgraded from 3.4 to 4.2, and I thought "oh well maybe they changed what packages are really needed now, I'll autoremove these".

NOPE. HORRIBLE IDEA. My webgui for the cluster started dropping off the planet.

I did an apt-get install proxmox-ve again, and it might be okay, but I don't know yet.

EITHER WAY THIS IS SERIOUSLY BAD ADVICE. WHAT THE FUCK.

If you remove a package via apt, the tool tells you exactly what will happen.

root@pve:~# apt-get purge pve-kernel-4.2.8-1-pve
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following packages will be REMOVED:
proxmox-ve* pve-kernel-4.2.8-1-pve*
0 upgraded, 0 newly installed, 2 to remove and 0 not upgraded.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n]


Seems you did NOT read this. If you work on the CLI as root, you should know what you are doing.

If you just want to use another kernel, there is no need to remove the default one.

Please calm down and respect the forum rules.

 

[BloodyIron]

1) I have every intent of following the rules, but I don't think asking me to calm down is reasonable here. Allow me to outline why.

2) The day before I upgraded from 3.4->4.2. I had been putting off the upgrade for ~1yr because I didn't want to rebuild my cluster, or at least, not until I could make the time, and I finally did. The upgrade appeared to go very well!

The next day I observed two VMs were crashing, and I tried to figure out what was going on. Eventually I found out a different kernel fixed the issue.

Now the reason I don't think this behavior (not yours, I mean the package manager behavior) is acceptable is because of the following:

1) In the upgrade I installed 4.2.8-1-pve. I found that 4.2.2-1-pve was stable. However, I was having difficulty getting grub to reliably boot into the kernel I want (due to MY fault of not knowing grub2 well enough, it was much easier in grub1).

2) As such I wanted to remove 4.2.8-1-pve so that 4.2.2-1-pve would boot by default. I HAD actually read the results when it said it was going to remove "proxmox-pve", but I thought that was a new thing since I had the day before just done a major version upgrade. I thought this was just how 4.x behaved.

3) Purging a kernel SHOULD NOT remove the entire hypervisor metapackage. This is completely asinine as all I wanted to do was remove a specific version of a kernel, but I could NOT find a way to remove the kernel WITHOUT removing the meta package.

4) This lead me to the scenario of FALSELY identifying that my hypervisor packages could be safely removed, and thus promptly shot myself in the foot.

This kind of environmental behavior is not okay because it punishes (for no good reason that I can tell) someone who wants to keep their environment tidy, as in, removing kernels I'm not using.

Anyways. I used big letters and such because I wanted people to heed this warning. That's it!

If you remove a package via apt, the tool tells you exactly what will happen.

root@pve:~# apt-get purge pve-kernel-4.2.8-1-pve
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following packages will be REMOVED:
proxmox-ve* pve-kernel-4.2.8-1-pve*
0 upgraded, 0 newly installed, 2 to remove and 0 not upgraded.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n]


Seems you did NOT read this. If you work on the CLI as root, you should know what you are doing.

If you just want to use another kernel, there is no need to remove the default one.

Please calm down and respect the forum rules.

 

[tom]

the proxmox-ve meta package contains always the latest kernel. if you do a manual purge/remove of a package form this meta package, also the meta package is removed.

if you want to boot not the latest kernel, you can edit the grub config file, in my example I got two kernels installed and I want to boot the older one by default:

> nano /etc/default/grub
...
GRUB_DEFAULT="1>1"
...

update the grub config:

> update-grub

root@pve:~# update-grub
Generating grub configuration file ...
Found linux image: /boot/vmlinuz-4.2.8-1-pve
Found initrd image: /boot/initrd.img-4.2.8-1-pve
Found linux image: /boot/vmlinuz-4.2.2-1-pve
Found initrd image: /boot/initrd.img-4.2.2-1-pve
Found memtest86+ image: /boot/memtest86+.bin
Found memtest86+ multiboot image: /boot/memtest86+_multiboot.bin
done
root@pve:~# reboot