Tuesday July 11th, 2023 NCLUG Meeting

[Bob Proulx]

j dewitt wrote:
> What: Tuesday July 11th, 2023 NCLUG Meeting

Bob started off the meeting with the solution to the pesky disk drive
storage array problem that has been haunting me for a few months.
Periodically I would get a disk kicked out of the array.  I would test
the disk.  It would always test okay.  I would put it back in the
array.  Then on June 17th things just fell over and could not get up
again!  And of course I had created a problem for myself because this
storage array was too big for me to back up.  That's famous last
words!  Because this was the biggest storage array I had and I did not
have anything else large enough to back it up.

I worked to rescue the array.  It would not assemble.  I tried
everything.  I replaced every bit of the hardware.  I shopped and
bought replacement hard drives.  I bought used datacenter pulled SAS
drives.  I cloned the old data onto the new drives.  And the problem
continued!  What?!  I thought the hostname was cursed!

It turns out I had not replaced the power cables.  In order to connect
six drives I needed to use a 4-pin molex Y splitter cable.  There is
nothing active in them.  I didn't suspect them.  But then I found I
could jiggle the cables and watch the console and see the drives
report a drive reset.  Problem found!

The 4-pin Molex power connectors in the system were worn from many
insertions and removals and was loose.  Due to drive seek vibration it
would thump from the drive and then drop the power for a moment!  That
was the problem.  I was able to remove the pins and compress the
sockets in order to make them tight again.  Solving that problem
allowed me to bring all disk drives in the array online.  I was then
able to resilver and scrub and all data restored with no errors.  Yay!
And also, Whew!  All good.  And now I have twice as many disks so can
create a second array now to have space to back up.  Backup won't be a
problem after the second array comes online.

Brian then tagged in talking about one of his home automation
projects.  Using a Raspberry Pi.  And commercially purchased
temperature probes.  And other types of sensors.  Temperature,
pressure, humidity.  Those commercial sensors produce an ASK Amplitude
Shift Keying signal protocol.  He would put on in his garage for
example and know how hot it gets in there in the sun.

Brian made a sweeping statement which caused some guffaws from the
members present.  "Raspberry Pis are terrible."  Ha!  And then he
introduces a Raspberry Pico Zero micro controller using micro python.
A demonstration of micro python on the micro controller was pretty
cool.  The Pico is much more power efficient than the full RPi.  It
does not need to run the full GNU/Linux OS stack of software making
updates much simplified.

Part of the micro python on the micro controller was a very simple and
straight forward radio protocol over the SDR software defined radio to
send what is effectively a UDP sender and a UDP receiver.  And
enabling being able to continuously send temperature and other sensor
data from one place to another place.  Pretty simple.  Pretty cool.

Bill then took the floor and the projector to talk about fzf the fuzzy
finder.  The fuzzy finder fzf can be used to locate quite a few things
using fuzzy search terms.

    https://github.com/junegunn/fzf

For example Bill has a large e-book collection.  Too many books to
remember exactly the list of titles.  Is such and such a book in the
inventory?  Let's look with the fzf fuzzy finder tool to look for the
title names.

Continuing with that Bill says it is a two step process.  Look first
with fzf to find the PDF file.  Then use evince to bring up the PDF
file as a second step.  But he says that can be better by combining
things into ONE step.  Bill added this function.  (Pardon my typos.)

    fzzz() { "$1" $(fzf); }

Call it like this.

    fzzz evince

By doing that it runs the interactive fzf first.  It interacts with
the keyboard interactively.  And PDF file is selected.  Then it is
passed to evince.  Very nice work flow!

Kyle then took the projector to talk about Sfeed.  Kyle shows us a
curl pulling from kernel.org's xml feed into a file.  Sfeed is a
utility that reads RSS feeds, and therefore I assume XML, converting
it to a TAB separated data file format for further processing.  This
utility makes it easier to do simple brute force data extraction from
XML files.  Check it out.

    https://codemadness.org/sfeed-simple-feed-parser.html

Phil took the projector to talk about his recent task and success to
make a backup boot of a UEFI booting system.  [Bob: Put me down as
disliking UEFI because it is so troublesome.]  Phil had an Ubuntu
22.04 UEFI booting system and wanted to make a data clone backup onto
another drive of similar but possibly different size that would boot
the same way.  Being a different size meant not a raw bit copy but a
copy by partitioning the new disk and installing and copying data
from the original source disk to produce a backup booting copy.

Phil walked us through the steps he has found to partition the
incoming drive with a suitable partition table.  Then mounting the
individual partition components and copying the data from the source
to the new drive.  The new drive's partition table will need an ESP
EFI System Partition, a root partition, at the least.  May also need a
separate /boot and swap partitions if desired with the configuration.

Particular points of interest were that if creating a new file system
on the new drives is that the UUID on the new file system will be
different and the /etc/fstab UUIDs will need to be updated to match
the newly created ones.

John asked a question and we had a big group brainstorm session to try
to understand why a previously working AppArmor FireJail broke on
Debian 12 Bookworm.  The details of this are a little involved but if
anyone is using AppArmor FireJail with Firefox on Debian 12 Bookworm
then John would like to hear from you.  We brainstormed up a bunch of
things to try but did not really have a resolution.