SDCard -- install here and deploy there? - Page 2

256 Mb. So 64 MB? <four Yorkshiremen>That's luxury!</four Yorkshiremen>

I can remember when I bought my first computer, in 1981, a CPM/3-based  
"Wren" (*). I decided that I could just about afford the RAM upgrade from 16  
KB to 256 KB, but I couldn't justify or afford the disk upgrade from 2  
floppies to 1 floppy and a 5 MB hard disk.

My first IBM-compatible PC, based on an 8086, came with a 20 MB HDD. It ran  
MS-DOS fine, but it wasn't up to running Windows 1 - and I used up most of  
the HDD even installing it from the multitude of diskettes that I borrowed  
from work to try it out of curiosity.


(*) I remember it was the first time I'd even driven in London, in the car  
that I'd bought a few weeks earlier after passing my test - navigating along  
the A40, Marylebone Road/Euston Road, Gray's Inn Road to Theobalds Road  
where the shop was. I even managed to find my way back home again ;-) The  
Wren still worked until I last tried it a few years ago, when I found that  
the PSU (a very obsolete design) had finally packed up.  

On Sun, 25 Oct 2020 20:23:17 -0000



    <grin>



    A bit later than 1981 I think - CP/M 3 didn't come out until 1983
CPN for the Torch (1982) was based on CP/M 2.2, I'd have just ported CP/M 3
if it had been available.

You could well be right. I thought at first I already had the computer when  
I started at university in 1982, but I may have got it a year later. In  
which case it wasn't just after I'd passed my driving test (in 1981) but a  
while later. Probably still the first time I drove in central London *on my  
own* as opposed to having my dad for moral support and to navigate.

Anyway, CP/M3, BBC Basic (which had been ported from 6502 on BBC Micro to  
Z80), Perfect Office (Writer, Calc, Exec) office suite, Prestel display and  
modem. All in a very heavy case that was "luggable" (though it left you  
walking lopsided!) because the unit with the motherboard, PSU, screen and  
floppies could be slid back to expose the keyboard or slid forward to a lid  
could be clipped over to make a self-contained unit with a (strong) carrying  
handle.

I was quite proud when I managed to write a Z80 routine (embedded in and  
called from a BASIC program) for bubble-sorting an array - which it did  
*many* times quicker than the same algorithm in BASIC. I found a listing of  
it the other day and was quite impressed with what my younger self had  
managed ;-)  I also made little circuit boards which plugged into the  
parallel port and interfaced with a) a digital-to-analogue chip and b) an  
analogue-to-digital chip. Not long ago I came across a sound recording I'd  
made this way from a CD player - Dire Straits' "Why Worry" - and it was a  
perfectly good 8-bit WAV file once I'd added the necessary file header. I  
also made an RGB-to-PAL converter (again, using an IC which did the job,  
with a suitable 4.33 MHz crystal) so I could display colour on a TV from the  
RGB port, since the Wren only had a monochrome amber screen.  

On Sun, 25 Oct 2020 20:23:17 +0000, NY wrote:


Certainly was <<LUXURY>>!

I first learned to program in Algol 60 on an Elliott 503 with 8 Kwords  
(39 bit words) with input on paper tape and output on a lineprinter.

Then learned assembler on an ICL 1901 with 4KWords of 24 bit memory, card  
input, compiler loaded from magnetic tape and output on a lineprinter.

My first home machine was as a 48KB system with an MC6809 running FLEX-09  
from two floppies. Input from keyboard, output to an Epson dot matrix  
printer (remember them?).

At work I was programming an ICL 2966 mainframe running 12 online  
database systems, for a few hundred uses, written in COBOL and on 16MB  
RAM. with a roomful of 400MB, washing machine sized disk drives. (you can  
go see one at TNMOC at Bletchley Park).

Mutter..mutter - tell that the the kidz of today and they won't believe  
you...

Now a Lenovo T440 laptop with 8GB RAM and 500 GB disk, running Linux,  
with Internet connections and a laser printer does everything I need and  
is fast enough to avoid waiting except for system updates.    

On Sun, 25 Oct 2020 22:58:40 -0000 (UTC)



    Lucky sod! Dartmouth BASIC[10] on a 4K word (32 bit) IBM 1130, but
there was a 1442 card reader/punch attached as well as the paper tape
reader and a 1403 (not N1 so I missed out on that joy).

[10] Then 1130 FORTRAN, then assembler. I didn't meet Algol 60 until the
1130 was replaced by an Eclipse and we switched to using terminals and
stopped being allowed into the machine room.

On Mon, 26 Oct 2020 06:02:21 +0000, Ahem A Rivet's Shot wrote:


The Elliott 503 was fast for its time. As well as its 8kword 39-bit 3.6uS  
ferrite core main memory it had another 16Kwords of 39-bit 50uS ferrite  
core memory that could be used as workspace for large matrix operations  
but was normally used as fast external storage: during normal operation  
the Algol compiler was loaded from for each compiler run.  

I was using the Elliott to analyse spectra output by a Mossbauer  
Spectrometer which used a 400 channel Multichannel analyser to capture  
output from a scintillation radiation detector. The Eliott was fast  
enough for me to analyse the results from a 24 hour spectrometer run  
during a standard 3 minute testing slot rather then needing to book time  
on it. Three minutes was more than enough time to read my program from a  
3 - 3.5 inch roll of paper tape, compile it, read in the data tape output  
by the multichannel analyser, analyse it and print the results.  

But the Elliott was a physical monster occupying six 2m x 1m x 0.6m steel  
cabinets, each weighing 465Kg. In addition there was an operators desk  
and full-size 1200 lpm lineprinter. The machine was pre-integrated  
circuit technology, so was built entirely from discrete transistors.  

Its most unusual feature was that its FP arithmetic operations were 5-10%  
faster than integer operations and it stored 2 instructions per word.
  

32, you mean. Also, technically, it's MiB.
https://en.wikipedia.org/wiki/Mebibyte

Yes, Ahem and I made various errors/typos: Ahem wrote Mb (10^6) when he  
should have written Mib (2^2010%24*1024 or *approximately* 10^6). I did the  
same with MB which should have been Mib (not only did I confuse M with Mi  
but I also mutated bits into bytes - the latter was a typo). Then to  
compound my error, I also couldn't divide 256 by 8 ;-)  

On 25/10/2020 20:23, NY wrote:



<four Yorkshiremen>
You were clearly posh. I didn't have a car. I road my bike to pick up my  
first RAM extender and I didn't get to ride on the Marylebone road, I  
had to ride underneath it.
</four Yorkshiremen>

Actually true, a trip to Watford Electronics via Edgeware Rd, a shop  
that specialised in bits for the Acorn BBC B.

On Mon, 26 Oct 2020 10:58:23 +0000



    Oh now that's a familiar trip - going back to before the BBC B when
I used to go there to get transistors and suchlike.

In the 1990s, one of our UNIX servers (hostname "ozz" - yes, someone  
couldn't spell: I know the Wizard was from Oz) was a real pig to get all the  
processes running after a reboot - it required a lot of black-art tinkering  
by one of the UNIX gurus. Consequently it was never restarted Fortunately it  
never need kernel rebuilds for incorporating new device drivers.  
Surprisingly, in view of its kittle and vulnerable state, it wasn't  
connected to a UPS. One day there was a fairly dramatic power cut, when a  
JCB working nearby put its jib through a high-voltage cable (*). Whereas  
everyone else went to investigate the loud explosion, the brilliant flash  
and the cloud of black smoke, ozz's "handler" went into paroxysms of despair  
at the amount of time he was going to waste getting it going again when the  
power eventually came back.


(*) The JCB partially melted. The driver was less lucky.  

On Thu, 22 Oct 2020 17:35:12 +0000, Markus Robert Kessler wrote:



1st rule of Linux :-
Reboots are for hardware changes only

Or a kernel upgrade

On Thu, 22 Oct 2020 18:22:31 +0000 Jim Jackson wrote:



Yes, yes, or for glibc upgrades. Well. But there are special cases where  
the USB (2.0) bus hangs if making snapshots from a webcam with fswebcam.

Then the machine has to be rebooted to reanimate the bus.

Best regards,

Markus

These days, where so much hardware is hot swappable (even on little
computers, where USB is how much hardware is attached), I find kernel /
low level software / underlying hypervisor issues to be the main reasons
to reboot.

I attach and detach sound "cards" and disks with USB regularly on
personal and $WORK laptops. The rack mounted $WORK servers get non-USB
disks, power supplies, and network devices swapped without reboots. But
all of them need reboots for kernel changes.

Elijah
------
the k8s pods get restarted on new hosts regularly as hardware needs work

Actually, that's been a rule for just about any serious
professional shop since time immemorial.  It was Microsoft
who introduced the concept of re-booting anytime for the
hell of it.  Making spontaneous reboots part of computing
culture is one more of their crimes against humanity.

On Thu, 22 Oct 2020 17:35:12 -0000 (UTC), Markus Robert Kessler
wrote:



Can't help think why you start the script every 10 mins instead of
having in a loop with a 10 minute sleep. If you want *exactly* 10
minutes between each sample one would have to code around the
(variable) time taken for the data sampling, processing and up load
but that's not difficult.



So it hasn't been power cycled and/or the SD card reseated? It could
just be a build up of corrosion in the slot connections. Almost all
Rapsberry boot problems I've had have been down to the latter or a
bad micro SD to SD adapater.

Not sure why you think you need to reboot if it's only running a
simple script. A Pi Zero here:

pi@PiZ-StoveB:~ $ w
 19:37:30 up 49 days, 10:24,

That Pi has extra hardware attached: An ENC28J60 ethernet port, ex
nokia LCD phone display (both on SPI buses), a rotary encoder and a
PWM driven solid state relay (4 GPIO's). It's talking on two 1-Wire
buses (2 more GPIO's) with eight or so devices across the two buses.
Every minute it reads all the 1-Wire devices, updates the display,
logs the data (over ethernet and locally), decides if the PWM drive
to the SSR is correct for the data it gathered and adjusts as
required. The display has an animated "heartbeat" symbol that shows
the system is alive when the PWM is off. If the PWM is on it rotates
the symbol at the appropiate speed. This is all under a multi
threaded python script. Oh almost forgot a bi-color LED that flashes
green as the 1-Wire buses are read or pulses red at rate determined
by the PWM.

It's also running a web server (nginx) that can produce plots of the
logged data on demand.

It's also running pi-hole.

As you can see it's doing all that and been up nearly 50 days... I've
never known it crash in use.



Don't follow that either.  B-)  How does the WLAN help with getting
an ethernet connection?

On Thu, 22 Oct 2020 20:11:45 +0100 Dave Liquorice wrote:



WLAN doesn't help at all since at that site Wifi is deactivated. So,  
replacing an RPI-A with a RPI-Zero means one has to attach an eth-to-usb  
module. Better use it just for installing a new SDCard and then put this  
card into the RPI-A.

Well, some months ago I wrote here about lost SDCards during reboot on  
RPI-Zero. Every 10 or 20 reboots the card was no longer found during  
reboot and had to be put into a desktop Linux PC. Then the partitions on  
the card were found instantly and the card worked again even in a RPI-
Zero. Whoever I asked -- no clue why.

So I decided to use RPI-Zero only for testing purposes and for logging  
data I use RPI-A, RPI-B and RPI-3.

So, still interested in getting infos about if or not it is possible to  
install a new SDCard in a different RPI and then just put it into the  
mentioned RPI-A and it works?

Thanks again,
best regards,

Markus

On Thu, 22 Oct 2020 19:54:44 +0000 Markus Robert Kessler wrote:



Sorry, forget to mention:

Yes, I try to install Raspbian OS.

Reason for asking all this, is, that once I saw a warning saying that  
installing newer Raspberry machines like "Rpi-Zero" will only be possible  
by using the most recent Raspbian versions. Older ones cannot do.

So, it seems that there are indeed some differences regarding accessing  
the hardware architecture, but hopefully newer OS-es -- once installed --  
can serve older hardware like "RPI-A" as well.

Let's see.

Best regards,

Markus

On Thu, 22 Oct 2020 22:15:44 -0000 (UTC), Markus Robert Kessler




    The 4B has a significantly different boot process, only supported with
the Raspbian based on Debian 10 (Buster). The foundation actually released
a Raspbian version some two weeks before Buster went official -- just
because the 4B had been released, and no earlier Raspbian would support its
boot.

    The 3B+, I believe, also required some additions to the boot system,
but I think that was still in the late Debian 9 (Stretch) period.

    ALL Raspbian releases contain the boot configuration files needed by
all Raspberry Pi models at the time of its release. The boot loader selects
the correct file once it determines which board it is running on. Just look
at all the bcm####*.dtb files in the FAT partition.