[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

About keeping a Debian installation happy [was: is it possible to add a secondary disk to an existing debian] systems and install programs to the secondary disk



On Sat, Feb 27, 2021 at 11:34:10AM +0300, Semih Ozlem wrote:
> Also even though I have been using linux for a while, that still does not
> mean that I would know everything about linux or that there would be no
> details that I miss...

Nobody does. The Linux kernel alone receives roughly 6000-8000 patches
*a month* [1] [2]. That amounts to a steady stream of one patch every
five minutes, if I didn't miscalculate.

A tad less than 2000 developers are contributing at any given time.

And a (GNU)/Linux distro is a bit more than that. No way to keep up
on each and every corner of that. Much less as a single person.

That said, I'm around that stuff since... something between 1993
and 1994. My main workstation (and a couple of other installations,
often at customer's) have been purely GNU/Linux things, most of
the time Debian.

I haven't experienced that kind of catastrophic failure since...
well, let's say 1996 or so. Most of the time, I've had fairly clean
upgrades, with few pains (and I /do/ customize my systems. I'm
sometimes picky. I don't like systemd, for example, thus I do
increase my risk by departing from the beaten path here and there).

What may be elements contributing towards keeping an installation
happy? This surely depends on many things. Here are a couple of
advices from an old jeezer:

 - community
   Find people "around" you (geographically is the best, but
   sometimes you don't have that luxury). A mailing list like
   this is an option. People you trust, and you learn to understand.

 - OS usage model
   think about how you are treating you OS: is it a pet, you
   know by name (upgrade frequently, install this-or-that
   utility from source, let it develop a "personality" over
   time, you're kind of sad when it dies) or is it "cattle"
   (you invest a lot into automating install and deployment,
   which happens nearly instantly. When the system coughs,
   you dump it and deploy a new one).

   Those are very different approaches, and have different
   properties.

   It pays off to think about what you want to do. I'm
   (pretty firmly) in the "pet" camp, although I see what the
   "cattle" model is good for. For a perspective from the "cattle"
   camp, see [3].

   I have the feeling that you dither between both positions
   (you deploy as "cattle", i.e. don't dare to install to
   disk) and expect "pet" behaviour (i.e. long-time stability).

Sorry for this philosophical interruption :-)

Cheers

[1] https://lwn.net/Articles/839772/
[2] https://lwn.net/Articles/845831/
[3] https://joachim8675309.medium.com/devops-concepts-pets-vs-cattle-2380b5aab313

 - t

Attachment: signature.asc
Description: Digital signature


Reply to: