How to clean up full /boot safely?

[Liam Proven]

On 12 February 2018 at 14:14, Tom H <tomh0665 at gmail.com> wrote:
> On Mon, Feb 12, 2018 at 7:26 AM, Liam Proven <lproven at gmail.com> wrote:
>
> Your entire reply to Colin is about parsing his language in a rather
> stupidly nit-picking way for the sake of argument rather than for the
> sake of furthering understanding and knowledge.

No, it isn't.

> Other distributions have philosophical rather than technical
> objections to distributing zfs kernel modules.

"Why" does not matter to me. "Will it absolutely definitely just work
in a crisis" matters. If the answer is "no" then "why" is of no
interest to me. The reasons may be perfectly good ones involving
non-GPL code in a GPL kernel but they are irrelevant.

"If I boot this machine off a known good current Linux bootable DVD or
USB medium, will it read all my disks? Yes or no."

"Well it depends on what dist--"

"Yes or no."

"Only if it is--"

"Yes or no."

"But..."

"YES OR NO."

"No."

Then I won't use it.

Why it won't work is not important in this discussion. I don't care.

I understand the reasons, they are valid, they are important, but they
are not relevant to this. If the answer to "will it work?" is not an
unqualified, unhesitating "yes" with zero riders or caveats, then my
answer is "no".

Comparison:

A few years ago, I wanted to move my personal website to a CMS.

All the main Linux CMSs are based on PHP.

I have some free webspace due to a kind friend. He is older and far
techier than me. He will not allow PHP on any machine he administers
or supports.

This meant that, at that time, there was no Free CMS for Linux I could
put on the machine.

It is a simple, absolute rule of his. PHP is a bad language. PHP is
unsafe. You _cannot_ write a known good, completely safe PHP app,
because PHP is not known good. It is known bad. It is unspecified,
badly-written, poorly-implemented and insecure.

So he said no.

I was a bit annoyed. I researched PHP. I learned why. He is right.

I do not run internet-facing webservers for a living.

I have, however, built significant corporate infrastructures that
supported hundreds of millions of US dollars' worth of business, every
day, with very high uptimes.

There was, back then, production stuff I would not allow on my
systems. Even if most of the rest of the industry used it.

My friend has the same attitude. If he knows it is untrustworthy, it
doesn't go on his boxes. Not in a VM, not in a container, not in a
special version, it does not go on, at all, ever.

And he is right to do this.

My attitude to things like filesystems is similar.

"This should work" is a 100% total no.

If someone says "X ought to work" then that means that they can
conceive of scenarios where it won't work. That is enough. No.


-- 
Liam Proven • Profile: https://about.me/liamproven
Email: lproven at cix.co.uk • Google Mail/Hangouts/Plus: lproven at gmail.com
Twitter/Facebook/Flickr: lproven • Skype/LinkedIn: liamproven
UK: +44 7939-087884 • ČR (+ WhatsApp/Telegram/Signal): +420 702 829 053