[conspire] successful install, at last

[paulz at ieee.org]

 My purpose for using a computer is to accomplish certain tasks.  Although *ux required my learning some basic sysadmin functions sometime around 1985.   That said, including my most recent experience, being sysadmin on *ux is probably easier than dealing with most IT departments to fix some *doz problem.
Yes I was aware that there were various ways to get aps that were not in the "target" release but where in other releases.  However, it appeared to be easier to try use *ubuntu.  

BTW, the normal downloaded *ubuntu.iso file is a live CD that has so you can take for a (limited) test drive before running the installer.  As a plus, while the installer was running, I could, for example, open firefox and read the email and then watch a video with a crowd of people holding signs.

Since that live CD includes partition tools and dd and other things, I could use it as the rescue disk.  I would be interested in what should be the minimum selection of tools on a rescue disk?
A related question is, how difficult is it to create a new live CD?   I mean creating a new ISO file.  Or modifying an existing ISO.  The answer could be as short as, "It ain't easy".  Or maybe it really isn't too difficult.  Then one could create a custom rescue drive.
I note that I said ,"CD" in several places out of habit.  I really meant USB.  It's incredible how much easier and faster it is to use USB 2.0 instead of a spinning disk.


    On Monday, December 17, 2018, 1:55:14 PM PST, Rick Moen <rick at linuxmafia.com> wrote:  
 
 Quoting Paul Zander (paulz at ieee.org):

> Seagate and Western Digital have free disk diagnostic tools. 
> 
> WD is an executablethat runs under Windows. It recognizes and will run
> diagnostics onother brands of disks, including those mounted on USB. I
> only used it toread the SMART results. 
 
I'll just mention again that SMART is (fortunately) vendor-neutral and
openly documented, so smartmontools does an excellent job of parsing and
dealing with that data.  The _bigger_ problem with SMART data is
deliberate lying _to_ the SMART reporting layer by hard drive vendors'
electronics, in an effort to make the vendor look better on a short-term
basis.  Long story.

What *I* would look for in vendor hard disk diagnostic tools is whatever
they offer, under whatever name, zero-filling, the successor to the old 
concept of low-level formatting.  Which I guess I should now discuss.

Back in the 1980s, 'low-level formatting' was a process of setting
sector and interleave information on hard disks.  You did this to ensure
that the HBA (the adapter) interacted optimally with the drive and
establish spare sectors for error-correction, or to re-establish sector
boundaries in case the write patterns had been dodgy.

Later drives starting with PATA ('IDE') were low-level formatted at the
factory with electronics right on the drive that kept the sector
information under control (thus the term 'IDE' = integrated drive
electronics), and low-level formatting in the field thus stopped being
useful.  (In fact, if you try it on a modern drive having forgotten that
it's not 1985 any more, you'll wreck the drive if it permits you to do
the operation.)

On modern hard drives, a zero-filling operation -- which of course blows
away all data contents -- is one way to trigger the drive's built-in
Spare Bad Blocks routine as part of the erasure process.  This can
extend to drive's service life.  Of course, if the bad blocks count is
increasing as indicated by the SMART data, that likely means bearings
are failing in either the actuator or the drive motor, e.g., hurry up
and do those final backups before the Grim Reaper bags it.
(Trustworthiness of drive data reported to the SMART layer is of course a 
different problem solvable only by putting the right people up against
the wall come the revolution.)

> Seagate has asimilar tool. Apparently it requires rebooting Windows
> before it will run. Then it keeps running in the background as a
> monitor. 

Be skeptical about these things.  The drive vendors are in no way averse
to issuing snake oil.

> Seagate also has a bootable USB tool. It is a version of Linux that
> will copy files. Nodiagnostics. IMHO, CABAL people would be better
> served with a live USB of a Linux that includes a partition editor, 

Indeed.  Find a live distro ISO you like for system maintenance
(partitioning, repair, backup, recovery, file/filesystem  undeletion,
etc.).  One old favourite is SystemRescueCD.  There are a bunch of
others.

> Some of the newer versions of gnome and KDE partition manager tools
> will access the SMART diagnostics. Store that USB in a safe place. 

Ugh.  Spare me the GNOME and KDE geegaws.  All you need are the real
tools those wrap around, e.g., GNU parted, etc.

> DISTROS AND APS
> 
> Earlier in the year,I wanted to install a particular ap (openscad). It
> was in debian sid,but not buster. Gee, I was using testing thinking I
> would get newerversions of aps. Obviously not correct assumption.

You _do_ get newer versions of apps, but that doesn't come for free.
There are complications, such as the fact that Debian-testing is the
contents of the Debian-unstable rolling distribution net of the results
of an automated quarantining script applied on an individual-per-package
basis.  Therefore, if a complex application package depends on 15
library packages, and one of those library packages hasn't cleared
quarantine from -unstable into -testing, then the application cannot
currently be installed in -testing because its dependencies cannot be
met.

You can often deal with this situation by directing that in this _one_
case, you want to install the package and its dependencies from the
-unstable track.  I've posted to this mailing list instructions about
how to do this (entailing use of pin priorities) countless times, along
with discussions about the advantage and disadvantages, and methods for
best use, of -testing.

Quoting the first of the _many_ times I've detailed that in this space
(2005).

---<begin>---

> I got into this situation by:
>
> 1.  Trying to install xmms-musepack, but it being similarly not
>    permitted due to conflicts.

If this is Debian, I notice that xmms-musepack is provided only in the
-testing and -unstable branches, _not_ -stable.

On either of those two development branches, you invariably get some
occasional problems with the three primary dependency hairballs:  KDE,
GNOME, and Mozilla-derived programs.  Some of the remedies are noted
above.

> Help! How can I get up off of the floor here?

If your system is tracking Debian-unstable ("sid"), then a bunch of your
options are noted above.

If your system is tracking Debian-testing ("etch"), then consider the
above plus maybe on additional trick:  I mentioned that I make my system
able to fetch -unstable-branch packages optionally.  Here's one way to
do it, which I'm informed is a bass-ackwards way of using apt-get's
"pinning" mechanism, but I played around until it worked, and that's
been good enough for me:

Add the following to /etc/apt/preferences (which you'll probably have to
create):

Package: *
Pin: release a=unstable
Pin-Priority: 50

This says assign a priority of 50 (where higher is greater priority, and
normal is 100) to any package from the -unstable branch.  Thus, no
packages will get pulled from that branch by default, in normal use of
apt-get.

Now, add an -unstable line to /etc/apt/sources.list:

deb ftp://ftp.us.debian.org/debian/ testing main contrib non-free
deb ftp://ftp.us.debian.org/debian/ unstable main non-free contrib
deb http://security.debian.org/ testing/updates main contrib non-free
deb http://security.debian.org/ stable/updates main contrib non-free

Run "apt-get update" to refresh your package listings.  You're
done.  You can now specify _optionally_ that the -unstable branch should
be used to fetch named packages and satisfy any dependencies, by using
the "-t unstable" flag, like this:

$ sudo apt-get  -t unstable  install kde

Good luck.

---<end>---



> So before I did aninstall on my new laptop, I looked at available
> packages. Quite afew, including gnucash (accounting program) and kicad
> (PC layoutprogram sponsored in part by CERN) were also missing from
> testing.The versions in stable were quite old. Meanwhile, ubuntu is
> based onstable, but has much newer versions of the aps I wanted. 

gnucash and kicad would be classic cases of dependency hairballs, GNOME
and KDE basketcases respectively.  And they're thus also classic cases
of the effect I mentioned.  Notice ongoing availability in Sid=Unstable:
https://packages.debian.org/sid/gnucash
https://packages.debian.org/sid/kicad

Maybe what you're looking for is the Siduction installable live distro.
https://distrowatch.com/table.php?distribution=siduction
(Notice that it's available in various flavours, including an LXQt one,
that being the successor to LXDE.  Might be your cuppa.)

> Naturally the lubuntu install is not the same as the debian installer
> I have used anumber of times. The one place that took many tries was
> the disk partitioning. I selected the appropriate drive, but it would
> not do a default install. It was not very helpful. Sometimes, just
> before committing to reformat, it pointed to something I needed to go
> back and fix.

There's no need to put up with that.  As I said, find a live distro you 
like for maintenance work.  Use _that_ to partition and mkfs.  Only
after you've done that prep work, fire up the installer for the distro
you wish to let loose on the filesystem you've already made.  When it
offers to do automatic or guided or whatever bullshit partitioning, say
no, and then say no again when it helpfully offers to mkfs or mkswap
them, and just tell it what mountpoint to use for each data-bearing
partition.  Slap its spastic little hands away from maitenance.  Use the
right tool for that.


> Othertimes, it completed the install, but the install would boot.

Another thing you might consider mainly relying on a good maintenance
distro for is setting up and managing bootloaders.


> BTW, lubuntu 18-10uses LXQt instead of LXDE. The biggest difference I
> found regarded the applications menu. 

Well, the other difference is that LXDE is basically dead.  LXQt is the
successor project.

Basically, the LXDE people finally figured out that the GNOME devs in
charge of the gtk+ toolkit were actively hostile to other desktop
environments, and also that the pain of recoding everything in their app
suite every time there was a gtk+ revision could no longer be justified,
so they made the tough call to cut anchor and migrate to Qt 5.0, joining
forces with the Razor-Qt project in making their successor DE.

Apparently, the XFCE4 people either haven't yet gotten the memo about
gtk+ being a dead-end for everyone outside GNOME/MATE/Cinnamon, _or_
they are so deeply mired in GNOME entanglements that they're screwed and
cannot move sideways.


> Long ago, the full linux kernel and associated functions was much
> bigger than the total RAM. Swap was used to more efficiently save and
> reload pieces that were not needed at the moment. The usual
> recommendation was allocate a swap space twice the size of RAM.

*sigh*

That was always only a very crude rule of thumb from Slowaris (Solaris),
and obviously leads to absurd results, in many cases, if taken
literally.

Here, read this:
https://www.cyberciti.biz/tips/linux-swap-space.html

> Now well into the 21st century, the linux OS has grown to more than a
> GB, but most computers much bigger RAM. With 2GB of RAM, swap is
> possibly not needed.Depending on your usage. For example, if that
> includes handling really big files,a big swap partition could help
> performance. 

I'm curious whether there are any stability problems if one does
entirely without either swap partitions or swap files, on modern Linux
systems.  Swap activity is deeply problematic on an SSD-only system
lacking a spinning-rust drive, so I'd be interested in whether zero-swap
operation is tenable.  I lack data from the field, and would rather
learn from other people's experience if possible.  (Old joke:  The way
you identify the pioneers in computing is from the arrows sticking out
of their backs.)

_______________________________________________
conspire mailing list
conspire at linuxmafia.com
http://linuxmafia.com/mailman/listinfo/conspire
  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://linuxmafia.com/pipermail/conspire/attachments/20181217/9144fd52/attachment.html>