You're just trying to avoid comparisons between the surviving candidates for the Tory leader and characters in the Laundry Files, aren't you?

Trust me, when one of the 1Tb drives in your RAID-1 setup claims to have a capacity of 600Pb , comparisons with Lovecraftian horrors are entirely redundant!

Well, I've never described a Laundryverse character as Cockwomble (or even a cockwomble), (Jeremy) Cunt, or Gove, or said that he reminded me of the love child of Mick Jagger and Gollum!

What happens when you try and execute Laundry computational magic on a bent Chinese SD card controller?

Both Paws & Pigeon .... 😎

Have fun rebuilding the RAID. Hope there are backups? ;)

And to spam something unrelated, because it's too good to pass in my opinion:

https://www.theguardian.com/books/2019/jun/20/petition-netflix-cancel-amazon-prime-good-omens-christian-neil-gaiman-terry-pratchett

Brilliant move!

Urg, that could be a controller problem.

Good luck. I hope all goes well.

I feel your pain. RAID 1 - hardware, or software RAID?

I've got a RAID 1 - two 2TB WD Reds, on my workstation at home. I really need to take down the system, and take it apart, and find out why the hot swap drive bay doesn't work, so I can back it all up to a 2TB drive (currently unused....)

comparisons with Lovecraftian horrors are entirely redundant!

As opposed to the array.

Good luck!

Urrgh. Those creeping fails are the worst. I had one on a corporate array once that just quietly corrupted the data without actually failing. Which then left us in the situation where we could go back to the tapes of a few days ago which were known good, and lose a week's work for 1000 people, or carry on. The decision was to do both restore to the previous night then carry on and restore corrupted files from tape as the users found them. Luckily the previous night was OK, though we did lose one small database that we didn't know was open all night and not being backed up.

Having a drive failure is better than having to tear up another novel, I suppose.

Just asking.

Why are you running hardware you seem to own instead of just renting service from a cloud provider?

Seems to be working?

Because you pay for usage from a cloud provider and historically it worked out cheaper to lease an entire box—this blog takes a hammering!

(I may move to cloud hosting when this year's rental is up, as prices are constantly changing, but I'm lazy and the system has worked for me these past 20 years or so, so …)

Inertia is a strong force. Totally understand it.

I've been working on getting rid of a mail server in my house for 3 years. I really really really want to have it done before the end of the summer. We shall see.

On a side note I deal with a blog that shall remain anonymous that gets about 4K unique visitors per day has around 375K comments posted in the last 10 years. We do it on GoDaddy's unlimited bandwidth fairly cheap hosting plan. Response seems about the same as here.

But where you host may be related to where you traffic comes from. And unlimited bandwidth may not be priced under $10/mo for a similar account hosted in the UK.

And I will agree that GoDaddy can make me grind my teeth at times.

I keep getting an error message, something about an AE-35 unit that requires replacement? You might want to look into that.

Charlie Stross @ 15: Because you pay for usage from a cloud provider and historically it worked out cheaper to lease an entire box—this blog takes a hammering!/i>

(I may move to cloud hosting when this year's rental is up, as prices are constantly changing, but I'm lazy and the system has worked for me these past 20 years or so, so …)

Sounds like a good example of "If it ain't broke, don't fix it" to me.

From what I've read, it is significantly cheaper to own your own hardware as long as your usage doesn't change too quickly. Cloud providers trade a lot of performance for flexibility, and it seems to make the most sense at very small scale, where fixed costs dominate, or very large scale, where even small relative changes are large in absolute terms.

Hopefully you can get a fair amount of life out of your other drive(s).

As someone with things on both sides of this line, security updates, hardware failures, OS changes, backups, and other issues are pushing me more and more to cloud services.

I pay about $200/year (CDN) for my unlimited size/usage website, including domain name renewals etc.

I could find cheaper hosting services, but the customer support has been great and they're a local company which I like to support.

There are about three blogs (currently) and five domains (with websites and other stuff) hosted on my box. I'm trying to kill the mail server I've been running there, and eventually centralization of services plus spam will do that for me, but I have more uses for my own box than just running this blog. If I was one-blog-only I'd already be all over a managed hosting option.

I do believe some of my customers would have very much liked to have called me something like that, when I refused to go ahead with a system re-install because the brand new, shiny, just out of its packet hard disk was showing SMART errors to the Linux system I'd just booted up.

In my experience, hard disks tend to fail with only the tiniest of murmurs to the SMART disk monitoring system (unless you go looking for problems, like re-allocated areas), so something moaning a bucketful unasked is likely to be on the verge of dying messily.

The customer then couldn't really fault my logic, but I'll bet there were some mutterings after I'd left, and I bet the supplier didn't half get a rocket!

Several things: first, for the non-computer folks, RAID-1 is two (or more) disks that mirror each other. If one fails, the other is still going, and you have time to replace the failed drive.

It is not a replacement for backups.

Cloud? So, then, who has your data? If there's a bill, or dispute, you ain't got nothing, and they've got the leverage.

OK, since this is a short comments section:-

Wikipedia presently has Accelerando as a prequel to Halting State. Other than in terms of publication date, that doesn't seem right?

You know that Charlie intended Halting State and Rule 34 to be two thirds of a trilogy, but that 'circumstances' were preventing him completing it.

This is just an end run round the problem.

Yes, but HS and R34 were the first 2/3 of the trilogy I thought. Wikipedia lists the 3 books in publication order, so with Accelerando first!

And also s/circumstances/"interesting times" I thought?

Of course Accelerando is first. Once you've had Teh Singularity, anything can follow, including a full resolution simulation of Edinburgh

You can fit everything into the Unified Charlieverse if you try hard enough.

By that argument, I think we can say:- -1) Glasshouse 0) Accelerando 1) The Atrocity Archives...

And prove that Glasshouse is part of the Laundryverse! :-)

I was originally being serious that Charlie might want the Wikipedia page https://en.wikipedia.org/wiki/Charles_Stross#Selected_bibliography fixed.

As it happens I had a go at editing the wikipedia page to show Accelerando as not part of the Halting State series. Whether the edit will stick or not remains to be seen.

Well, it shows Accelerando as standalone for me right now.

I think it should be quite obvious that we, along with all Charlies creations are experiencing the ending of "A Colder War".

"There is life eternal within the eater of souls. Nobody is ever forgotten or allowed to rest in peace. They populate the simulation spaces of its mind, exploring all the possible alternative endings to their life. There is a fate worse than death, you know.'"

Life with CockWomble as PM? ;-)

Of course Accelerando is first. Once you've had Teh Singularity, anything can follow, including a full resolution simulation of Edinburgh.

Are you sure it's not a low resolution simulation of Edinburgh?

Some things use a lot of clock cycles or memory, like science fiction authors and busy pubs, but many things could no doubt be approximated by procedural generation and chatbots, such as blocks of flats you never visit, one-visit tourists, and many blog commentators. grin

The sysadmins would have to clock it up and allocate extra storage for the Fringe, which is a considerable stress test upon reality.

The weather could be modeled at any resolution; how many clock cycles does it take for overcast and rainy?

True, true. But when you've got effectively infinite computing power, why try to optimise?

The Fringe is an interesting phenomenon, in that a lot of comedians want to perform at the Fringe, so they spend the previous months putting together a new show, and the overall effect is an annual cycle starts happening. Right now, you've got a lot of people doing rough cuts and previews, whereas in September ... or perhaps October, September may be recovery time ... it's their award winning Fringe show now going on tour.

And, indeed, you get other people, such as me, who time visits to Embra on the basis "if Festival'time then not Edinburgh'visit".

Cities are different during festivals. My wife and I used to visit Basel, CH for Fasnacht, which is their take on Mardi Gras. The locals who like the festival (which runs from 4:00 Monday to 4:00 Thursday¹) try to stay awake for as much of the 72 hours as possible. Hotel rooms become pretty expensive (we tended to book a room in the next door country), and there are people who get the fuck out of town for that period.

We've been to Basel out of festival, and it's quite a different place.

¹ In the morning. And yes, a lot of people will be up for that opening, though we enjoyed our sleep too much to do it every time. It can be damned cold standing on the middle of a bridge over the Rhine in February or March while it's still hours till dawn.

whitroth @ 24:

Several things: first, for the non-computer folks, RAID-1 is two (or more) disks that mirror each other. If one fails, the other is still going, and you have time to replace the failed drive.

For "non-computer folks", you can buy a pre-built RAID-1 box from Best Buy, Walmart or Target. You'll pay too much and get too little, but if one disk fails, you'll still have your files to transfer to the NEW RAID-1 box you buy from Best Buy, Walmart or Target ... as long as you don't waste any time. And for a fee, Best Buy will send a guy out to transfer those files for you.

OTOH, anyone who can read a book (or watch a YouTube video) and is coordinated enough to walk & chew gum at the same time should be able to build their own RAID-1 box ... even the "non-computer folks".

For a more comprehensive explanation of RAID:-

http://www.myconfinedspace.com/2006/12/19/raid-explanation-with-water-coolers/

Accelerando is a stand-alone and has nothing to do with any other novels. (Although Glasshouse very nearly turned into an explicit sequel before I saw sense.)

"Halting State" and "Rule 34" are the first 2/3 of a trilogy, but book 3 stalled in 2010 … then went on indefinite hold in 2012 because of the Scottish Political Singularity (aka independence referendum #1 and then brexit referendum #1). I hope to be able to put it back on the to-do stack some time after October 31st, but nothing's certain.

https://www.facebook.com/12706231/posts/10104981477660758/

Company led by former generals and intelligence agency heads gets no-bid contract to house migrant children for $775 per night.

The blatant criminality is practiced with a presumption of impunity that so far seems supported by events.

I despair for my country.

Whoops, wrong thread. (But this one does load faster.)

For a more comprehensive explanation of RAID:

The water cooler cartoon is a good one; it's not always easy to illustrate technical arrangements.

I still remember a convention encounter from the early 90s, when I met a guy working on what he described as a "phased array disk drive," multiple disks working as one. It sounded cool at the time and it was a while before any description of RAID schemes got through the advertising fluff and I recognized the idea.

Company led by former generals and intelligence agency heads gets no-bid contract to house migrant children for $775 per night.

Shit, you could buy them a room in a luxury boutique hotel in Manhattan—paying the rack rate, no discounts—for less. (One of the cheaper Kimptons, and admittedly they wouldn't get a suite for that money, but someone is hauling in the dosh hand-over-fist here.)

The Disney corporation has enough capacity in Orlando to take all of them, and their posted public rates for the budget resorts would be about $80 per child per diem.

And it would not be a bad idea all around.

While the usual spoilsports would scream bloody murder if it was actually done, I'd like to see the "hire babysitters and take everyone to Disneyland" plan kept in the public eye. If that is cheaper than what the government is doing, someone is not only cartoonishly greedy but not smart enough to make it look like we're getting value for our money.

"While the usual spoilsports would scream bloody murder if it was actually done,"

Yes. They would. 25% of the country actively wants to see someone suffer. They';re willing to pay top dollar to make sure these children get abused. And I personally want to see that 25% suffer at this point.

I have previously noted that it would be much cheaper to send criminals to full-board Spanish holiday resorts for the duration of their sentences than accommodate them in British prisons and it would be just as effective at keeping them off the (British) streets.

"And I personally want to see that 25% suffer at this point."

The people in charge of the U.S. are really, truly, awful from just about any standpoint you can name. Nixon would be ashamed... Thinks are getting to the point where I'm not sure they can be fixed by any normal democratic process - though I hope very strong that I am wrong.

Sorry, I haven't had my covfefe yet. I meant to write, "...though I hope very strongly that I am wrong about this."

Argle...

People, RAID is not a backup system, it is a high-uptime high-availability data storage system. In some configurations it can provide faster than standard reads and writes by interleaving drive accesses but it is not a backup or a substitute for backups. People who do not understand this concept are destined to lose their data some day because they believed RAID was automagically taking care of everything and they didn't need to bother with making backups.

Pardon me while I recite the data manager's creed:

One backup is no backup. Two backups is a start. Three backups is getting somewhere.

“<>someone is hauling in the dosh hand-over-fist here” http://www.dccapitalpartners.com/overview/boardOfAdvisors.html https://www.bloomberg.com/research/stocks/private/board.asp?privcapId=587630454 Recent addition General John Kelly, former White House chief of staff.

America's two party system is an emergent property of its variant of first-past-the-post, and the most glaring flaw in the system is that to get that desired 51% of the vote, there's a shortcut that obviates a lot of hard work you'd have to do to build up support from the voting constituency: pander to racists. It's an easy 15 to 20 percent bonus, and both parties historically used it.

So unfortunately, democratic means are not useful for putting an end to the sadistic abuse of children along the border. Too many voters WANT IT. And at this point I as a New Englander am increasingly interested in walling those people off. We really are split between Canada and Jesusland, and I find I have much more in common with Maritimers than Jesuslanders.

Three backups is getting somewhere. And if you've never verified that restore works, assume that there is a strong possibility that those backups are useless.

I said "Getting somewhere". Even running a CRC check on all of the backed up files helps as does reloading a few files as a basic sanity check.

I just happened to be making a backup of assorted folders on my desktop machine when I read the RAID comments earlier. I've since run a scan on the backup media (an external 6TB USB 3.0 drive) to verify it's not dying and that I can read back random files followed by a comparison pass of the backed-up files with the originals and everything checks out.

That backup drive goes somewhere safe soon to be replaced by Backup 2 (an external 5TB USB 3.0 drive). Stuff I really can't replace like photos or do without also gets copied to a few different online storage facilities.

A backup that is never verified is a backup that will fail.

A timely anecdote: Last Thursday I moved files from my tablet's internal memory to the new micro SD card I'd been wanting a while. Friday morning I wanted to listen to music and discovered that the contents of music directory had not been copied over to the card before being deleted from the tablet. (Restoring from a thumb drive was trivial, but it left me without music a few hours until I got home.) I'm sure I started that copy; what brain fart got into the machine I don't know. But it's a good lesson in having multiple backups in different places.

I'd go even further and say that a backup that is never verified is not a backup.

Pretty much by definition, yes. It's the foetus to the checked-restore baby — it's a necessary step, but it's not there yet.

One of my first tasks when I joined a certain Cambridge software company (one that sort-of lives on in your phone) was to do the daily server backups. The server ran a pair of 400K floppy drives, and 'backup' involved copying each disk to a fresh floppy.

When both disks had been copied, it was the copies that were left in the drives.

We just got a new box of hard disks delivered. So now we wait a week for it to initialise ... you'd think RAID6 would write everything in parallel but that's not what happens. So it's roughly the same time as writing 60TB as one big sequential write (~120 hours IIRC). But once that happens we can dispose of the legacy 10krpm SAS system (because we can't buy drives acceptable to it any more) and go with something using commodity 3.5" SATA disks. We don't really need the speed, it's only got a GB network port, but now we will have hot swap drives!

All we need to do is make sure the backup process works. Without, sadly, also having access to a 60TB second NAS to test on. So we're partitioning the one system and restoring to the other half. Not ideal, but it beats "the light went green I assume it worked" as your verification step.

Various - Thanks for the refresher course on back-up strategies guys.

you might be able to restart "Liz Cavanagh #3" late this year

Nope. I might be able to restart it after this year.

However, the to-do list looks like this, currently:

• Finish rewrite of "Invisible Sun" (in progress)
• Edits/rewrite of "Dead Lies Dreaming"
• Second draft of "Ghost Engine" (halted in mid-2017 due to parental death)
• Write "The Valkyrie Confessional" (Laundry Files book 10/11—the Senior Auditor's story)
• Write "Prime Cuts" (sequel to "Dead Lies Dreaming", aka Laundryverse Series 2 Book 2)
• Write the not-yet-titled Last Laundry Novel (Bob's finale)
• ??? Possible sequel to "Ghost Engine" (if it's commercially desirable)

The third Scottish crime novel goes on the end of this stack, although it could conceivably be slotted in before "The Valkyrie Confessional", but that doesn't make it hit publication any time before 2021 and it'd mean a two year gap between Laundryverse novels right after I start spinning up a new series in that setting, which might make readers who don't religiously follow this blog go WTF?!?

Yeah, include me in for copies of all of those as well.

Oggie Ben Doggie @ 47: 25% of the country actively wants to see someone suffer.

Sounds good to me as long as I get to choose who does the suffering! How does the saying go?

"I came here to comfort the afflicted and afflict the comfortable."

Troutwaxer @ 49: Nixon would be ashamed...

And well he should be, since he started most of this shit!