Rick's b.log - entry 2024/05/04

Mowing

Luckily, one of the maintenance guys at work has a sideline in fixing mowers and stuff, and when I gave him the axle, he found a suitable wheel for it. Unfortunately, he didn't remember that I asked for two, but on the other hand he said it was a wheel that was not much good (it was starting to wear so he put it on back to front) so he didn't ask anything for it.
This was enough to get Marte running again. The other wheel isn't good (neither, let's face it, is the entire axle arrangement) but it was enough that I could consider cutting the grass today.

If only the engine would turn over. The solenoid clicked and the engine sort of did a quarter turn. Which was odd, if I took the spark plug out I could turn it by hand and it was stiff but no more than usual.

As a part of the diagnostics, I hooked the new battery directly to the big +ve wire (taking the old battery out of circuit) and the engine spun fast enough to almost start, so I cranked it a bit while fiddling the choke lever until it did. So the old dud battery was sucking power. I hadn't expected that.

By the time I got everything going, it had started to rain. I shrugged and mowed the driveway, the Western Wilderness (including down by Anna), the Northern Passage, the Southern Sun, and just 'cos I also did most of the potager since I had a mower that turned in the direction I pointed it.
It was a fair struggle as the grass was not only wet but several inches tall (in two weeks - eek!) but Marte powered through it.

I left the patch at the end of the driveway (lots of buttercups), and the Picnic Lawn to the east as, well, it was raining quite a bit more forcefully now. I had planned to do the bit by the washing line, but as I was coming down onto the flat the blades caught a clump of grass and the engine stalled. Deciding that I really couldn't be arsed to bring out the battery and jump leads, I just pushed Marte back. I'll need to unclip the deck and clean it, but... when it's drier.

Marte; one good wheel and one less so.

Oh crap!

...the evening of the following day, after work, my little car will go for its MOT test.

Wish me luck!

I'm worried about the emissions test, it's the older Euro2 engine, exactly the same as the one in Felicity which would be exempt from the emissions test due to her age (though, to be fair, Felicity would likely have failed on just about everything else!).

PVRs and recordings

I did try to get myself another receiver, which seemed nice, but the dealbreaker was that it appeared to need some sort of firmware update in order to actually record anything. Which I failed to find online. How utterly dumb. So I sent it back.

I have just ordered another Touchbox HD3 - twenty euros from Amazon Second Hand, I can keep that as a spare just in case.
But, wait, why order a model that doesn't seem to make it more than just shy of two years? Well, looking on Amazon it seems extremely hit and miss whether the recording function actually works. I know this model does work, and I have saved the setup so it ought to be fairly easy to get a replacement set up. For its quirks, it does record.
And, besides, twenty euros... that's like a Happy Meal more than a go-large Big Mac. I'd be hesitant at getting one for sixty euros (the RRP), but a return for twenty? Yup, can do that. Anyway, the return here in France is liable to be because it can't cope with FRANSAT as, unlike Germany and the UK, France's satellite channels are not free to air.

This leads on to the second issue. Recording failures. I figured out pretty quickly that USB flash sticks were useless. I got a couple of 16GB sticks in the winter sales. They lasted about a month before frequently corrupting (and losing data). SD cards were better. They didn't tend to suffer the problem of offering complete gibberish in place of a list of recordings, but they did frequently abort recording. I had to 'nanny' the recordings I made this week of A Kind of Spark. The first one on Monday (a two-parter) on a brand new SD card worked fine. By Thursday, it was recorded in five pieces (with gaps when I had to notice the recording stopped and grab the remote to restart it). I can patch the bits together using ffmpeg on my phone, though for some reason I can't figure out, the concat filter throws away the subtitles and the primary audio track, leaving the narration audio. Still, better than nothing (though I think I still have the original files in case there's some weird magic incantation).

I had a theory that the recording was extremely hard on the media, so in order to prove the theory, I got myself a little USB harddisc from Amazon. It's an external USB 3.0 harddisc from a brand called UnionSine. Which means it's some cheap Chinese thing. But I didn't know if it would actually work with the satellite recorder, so my requirements were that it had to be spinning rust and not any form of flash, and it had to be cheap. This thing ticked both boxes. I don't recall the price, something like €25.

The harddisc that is attached to my satellite receiver.

Upon connection it powered up and sounded like it was spinning correctly (so had good power). I went to the menu and it reported FAT 0MB/0MB. Given that it's a 500GB drive (or around 460 real gigabytes), I would imagine the receiver was failing to understand exfat. So I reformatted it as FAT32 and it took rather a while before saying it had finished.
As it was nearly 6pm UK time, I switched to BBC One and pressed record. Let's see how reliable this thing is.

I made tea, drunk it, fed kitty, remembered I was supposed to take the rubbish up the lane for collection, did that, came back, microwaved a burger, ate it, and as it was nearing 7pm (UK time) and the end of South Today, it was still going.

I set up some programmed recordings through the night and in the morning. All appear to have worked flawlessly.

Some stuff that has been recorded.

My theory? My theory was pretty much confirmed when I rested and ear on the drive as it was recording. Click, click, click, click, click, click, click... this isn't a fault of the drive. What I think is going on there is that the satellite receiver is writing a block of picture data followed by updating the FAT, and just doing doing that endlessly during the recording. It sounds like it is doing it about twice every second. A typical film lasts around two hours (with adverts), which is 240 minutes. Which is 14,400 seconds, which is nearly thirty thousand updates to the FAT in the course of recording one single movie.
I think what was happening was that this behaviour was utterly slaughtering the flash media. USB sticks would die first as they are generally much lamer and not really designed for high intensity use. The SD cards fared better as they are often designed for higher intensity uses - often you see them claiming to be able to handle recording HD video, something you rarely (if ever) see on a USB stick. But the wear levelling was just taking too long to shuffle stuff around for all of those writes to the FAT. In a video camera that might result in dropped frames, but my satellite receiver was just nope and giving up.

Note that there are different types of flash. SLC is the best as each cell represents the actual binary data to be written to the media. Accordingly, it can cater for 50,000 to 1,000,000 writes to the same cell.
Next is the MLC/DLC that stores two bits per cell. This is much cheaper than SLC because you only need half as many cells for the same capacity, but it's at the cost of longevity. MLC NAND supports 1,000 to 10,000 writes to the same cell.
TLC is cheaper yet and stores three bits per cell. But it is good for maybe 300 to 3,000 writes to each cell. Sometimes this is called 3D NAND.
QLC is four bits per cell, and this can be updated a maximum of around 1,000 writes to each cell.
Note that by "write" I'm referring to a full program/erase cycle. Unfortunately it's hard to determine the actual number of writes to any particular cell because wear levelling in the controller will always be shifting things around in order to extend the device lifetime by reducing how many times an individual cell gets overwritten. For my thirty thousand FAT updates, it's entirely possible that the FAT was actually written to thirty thousand different locations on the media.
Note, by the way, that it's never this simple. FAT33 has an FS Information Sector (which may or may not be supported!) that gives information on the free/used clusters, the FAT cluster map that maps where things are on the disc, the directory table, and the VFAT (long filenames) entry. So for every chunk of video data written, there may be four additional writes to keep the media structure up to date with the actual data written (size updated, clusters marked as in use, etc). So that "nearly thirty thousand" could actually be over a hundred thousand, per movie.
You can see why USB sticks got toasted pretty quickly, can't you?

Spinning rust can happily update the FAT a billion times. The head floats over the platter and the magnetic media doesn't wear out like flash does. While the media cannot be written infinitely, it can be considered near-infinite as the parts in a harddisc that are liable to fail are the platter motor, head positioning system, lubrication, bearings, etc. And that's not counting instantly-fatal head crashes due to mechanical shock...the heads are floating micrometres over the platter surface, a surface that could easily be moving at 100kph relative to the head.
If we assume that a drive has a useful service life of, say, three years, then there's no problem with spinning rust at throwing data at it as fast as you can for the entire three years, non-stop.

And it looks like my simple test was valid - my satellite receiver is happier dumping data to a real drive rather than overstressing flash media.

That's good, as it's Eurovision next week. And for a first, the semifinals on BBC One instead of BBC Three or BBC Four. Isn't it earlier than usual this year?
Of course, while the EBU stresses that the contest is not political, it's worth pointing out that Russia got suspended in 2022 due to the invasion of Ukraine, Israel is still in the contest following the atrocities in Gaza. So, yeah, good luck trying to justify punishing one mass slaughter while excusing another.

Let there be light

I looked at plugs, about €3,50 each. Wire? €8,98 for five metres of two-core low current stuff. I could get a two metre extension lead for €7,30.
But you see that one lurking in the middle there? If I don't mind it in black, I can get a five metre two-pin extension lead for €3,99. Like, basically the same price as the plug.

Prices of things.

I only needed two wires for the light unit. As it's all plastic, it doesn't need an earth. So I can get the entension lead (with a 5A plug that looks like a 16A one leading to a 5A socket) and cut off the socket, bare the wires, and hook them into the light.
Like this.

Wiring up the light.

There's no switch. It is supposed to be a fixed unit, however I'm making them mobile for where I might need some extra light.

You can see the light here, and through the magic of buying two of them (sorry Alec!)...I have two.

Two lights.

I have, for now, clipped one up in the dark corner at the back of the living room.

Illuminating the darkness.

Your comments:

Please note that while I check this page every so often, I am not able to control what users write; therefore I disclaim all liability for unpleasant and/or infringing and/or defamatory material. Undesired content will be removed as soon as it is noticed. By leaving a comment, you agree not to post material that is illegal or in bad taste, and you should be aware that the time and your IP address are both recorded, should it be necessary to find out who you are. Oh, and don't bother trying to inline HTML. I'm not that stupid! ☺ ADDING COMMENTS DOES NOT WORK IF READING TRANSLATED VERSIONS.
 
You can now follow comment additions with the comment RSS feed. This is distinct from the b.log RSS feed, so you can subscribe to one or both as you wish.

jgh, 5th May 2024, 03:55

It's amazing how cheap a lot of consumer products have become, and some no-so-consumer stuff. I've seen *new!* hand-held amateur radio transceivers for less than twenty quid.

David Pilling, 5th May 2024, 16:07

Interesting story, one often wonders about how much flash drives are being hammered. I went through a phase of thinking SSD was great, I'd move to that. Then I read how short a time the makers expect it to last - hammering is not the problem, leaving it unpowered is the problem. Seemingly once powered it starts to refresh itself.  Turns out spinning rust is your best bet for long term storage. I bought some cheap hard discs from Amazon - as you do to keep your life's work safe. Mostly these were dated 2019, 500GB for around 10 quid, use with a SATA to USB adapter. But one was dated 2012.  There is crystaldiskinfo that lets you read disk information. Maybe 2012 but it had only been turned on for a few hours.  As ever - anecdote is useless - my Iyonix, 25 years old, original hard disc still going strong. I have USB pen drives 12 years old, no problem.  Why am I keeping all those floppies and tapes - because I believe they'd work.  Goldilocks and storage, usage in moderation.

Rick, 5th May 2024, 17:02

How long storage lasts for depends a lot on how it is used. The uSD in my Pi is years old and works fine, because while FileCore is absolutely not optimised for flash media, it doesn't tend to do much with the media. It doesn't swap and there aren't many applications that hammer the media.  Certainly not like writing HD video as described above. That wears out pendrives in a matter of *days* and can get uSDs wear levelling to kick in really soon.  That being said, my Pi's card might be wear levelling itself constantly, but there's no specific time constraint so it wouldn't be noticed.    I think harddiscs are best if you use them rarely (like once a fortnight to back up stuff) or leave them running constantly. It's the old fashioned "turn on in the morning and turn off in the evening" that stresses them.  Eventually they'll fail, everything dies in the end (even us), so having copies and backups is the way to go.    If your flash is 12 years old, you might want to consider pulling all of the data off and writing it back to refresh it.  Given how much stuff uses flash for firmware, it's a tad concerning that flash chips are often "guaranteed" to be readable for 10~20 years. Lots of potential landfill somewhere down the line? It's hard enough to find firmware for new devices, never mind old things.

Jeff Doggett, 5th May 2024, 17:03

Fat32 is actually worse than you said - there are two copies of the FAT to update plus the info sector.  When I wrote Fat32fs, I tried to ensure that the FATs were only written to once a full sector of 512 bytes was ready. The info sector is only updated later.

VinceH, 6th May 2024, 09:27

It's not just me suffering an unreliable NAS, then - although in my case the issue isn't storage media, but the software & services (e.g. the TV Guide it uses).    Helpfully, though, it's weaning me off of bothering to record things. 🤷‍♂️    I don't watch much *live* TV, so I only look at TV guides with a view to deciding what to record... so, blimey, I hadn't spotted the date for Eurovision - just five days away! Good job I took a look here this morning - which is only because I forgot today is a bank holiday, so my working day alarms went off as normal. I'm avoiding crawling out of bed unnecessarily eary!

Add a comment (v0.11) [help?] . . . try the comment feed!
		Your name
		Your email	(optional)
		Validation	Are you real? Please type 25268 backwards.
		Your comment