reminds me that time when *everybody* was buying silicon wafer cutting saws from one guy that made them in his home garage...
Wait, what? The Linux Kernel Mailing List archives lived on ONE PC? One BROKEN PC?
Spare a thought for Jasper Spaans, who hosts the Linux Kernel Mailing List archive from a single PC that lives in his home. And since things always happen this way the home machine died while he was on holiday. The lkml.org archive was therefore unavailable for much of the weekend, although Linux developers could still use …
COMMENTS
-
-
Monday 15th January 2018 00:13 GMT Yet Another Anonymous coward
Or all the chip encapsulation epoxy came from a single chemical company in Kyoto - then there was an earthquake.
It used to be that the single source of the material for the crucibles needed to melt the 99.999999999% pure silicon for chips wafers - all came from a single seam in a single mine.
-
-
-
-
-
Monday 15th January 2018 16:08 GMT phuzz
Re: £120 for 4MB
A whole megabyte? Kids today etc.
I remember buying (or rather, my dad buying for me), a 512KB upgrade for my Amiga 500, which was about £60 if memory serves.
We bought it from Evesham Micros, who were basically operating out of a double garage at that point. Little did I know that I'd end up working for them some fifteen years later.
-
Tuesday 16th January 2018 16:44 GMT mrvvg
Re: £120 for 4MB
You were lucky, first Z80 single board system we built, wire wrapped, had 16k, [might have been less], ran at 1MHz, but not at 2, spent a LOT of time persuading the manager to let us cut a circuit board to increase the clock speed, 1982 I think..
My first PC had 256k with a 384k add-on board. Still have it, might not work tho :-)
-
-
-
-
-
-
Monday 15th January 2018 01:05 GMT Youngone
Oh, come on now!
Let's be fair to Jasper.
He probably set it up because he has an interest then it got big and:
what the hell, it works OK I will get a cheap RAID card at some point,
the backups are running and restore OK,
I really must get around to virtualising that lkml.org box, hell, why bother it runs fine and besides, it's not like it's making a fortune in ad click money or anything.
Shit, has it stopped? I'll fix it when I get home.
-
-
-
Monday 15th January 2018 12:22 GMT JMcL
Re: Oh, come on now!
@Tigra 07
"IMDB was hosted by a teenager in his basement before it became big and was sold on for big bucks."
It was actually hosted in Cardiff University and called the "Cardiff Movie Database" At one point in 1994 myself and a friend had discussed with the guys behind it setting up a mirror in Ireland (we were actively updating it at the time and were having increasing trouble with congestion). We'd gotten to the point where we had been sent the source code, but unfortunately never got round to actually doing anything about it.
-
-
-
Monday 15th January 2018 16:31 GMT BinkyTheMagicPaperclip
Re: Oh, come on now!
mdraid? Not my experience. I've seen more weirdness with one mdraid raid 10 implementation than I have in practically all the hardware raid configurations I've come across, and hardware raid is generally easier to use too..
When I have a moment I'm moving towards ZFS, and BSD as soon as FreeBSD adds the features I need..
-
-
Monday 15th January 2018 01:21 GMT Anonymous Coward
On the flip side
Would it have been as stable on a cloud provider?
I guess you could put on more than one. Then throw in fail over routing Then make sure your cloud infrastructure is secure.
Finally make sure its managed by more than one person on a different continent, to protect against, death, disablement, super volcanoes, sharknados, limited nuclear war, and squirrels
http://cybersquirrel1.com/
-
Monday 15th January 2018 02:00 GMT Concrete Gannet
Don't stress, LKML is mirrored in many places
Like any well-known mail list, LKML is mirrored in other places, e.g.
http://linux-kernel.2935.n7.nabble.com/
and anyone who cares can use tools like PonyMail (https://ponymail.incubator.apache.org) to create their own archive.
So while the home location might be fragile, it can be re-established in days or weeks, and the history won't be lost.
-
Monday 15th January 2018 02:56 GMT Mark 85
Reminds me of the old BBS days....
Someone has an interest in something and runs a BBS (or server, or mailing list). Lots of happy users.
Owner doesn't make any money but is doing it because they want to and it's a good service.
Service goes down because hardware. Suddenly mad panic, even if it's mirrored. User base gets upset. Owner has to find parts/money to fix problem.
Once fixed, everyone is happy and using his/her service (usually for free) again.
Ah the joys of the past come to the present. I would hope, in this case as it sometimes did in the past, that someone (or a lot of someones) helped out with some cash or parts.
-
Monday 15th January 2018 06:20 GMT Kevin McMurtrie
Old school
This is refreshing in a world where I see people burning insane amounts of money on their cloud hosted Kubernetes + Node.js + React + Redis + CPython + PyBrain + Memcached + CloudSQL + Debezium + BigQuery + Cassandra + Kafka + Lucene + Elasticsearch + Splunk + nginx cluster, which they boast can serve one million pages a day and be maintained by just 10 full time developers.
-
-
Monday 15th January 2018 14:40 GMT Androgynous Cupboard
Re: Old school
Ah, you mock but those technologies have replaced thousands of lines of brittle, home grown code - now it's a simple task of maintaining a few hundred configuration files in XML, YAML, JSON and M4. Much easier! Now, if you don't mind I'll continue searching NPM for a library that can tell me the length of a String.
-
-
-
-
Monday 15th January 2018 10:50 GMT Anonymous South African Coward
Hats off to an old-school admin who runs a physical server under his stairs in his spare time as a hobby project, and as a public service to boot.
That's how you learn how these things *really work* and keep your own hardware skills up to date.
Agreed, plus it is fun shutting it down and listening to the cries of outrage
-
-
Monday 15th January 2018 15:57 GMT Lord_Beavis
Like have a prod server under a desk in a cubicle...
The companies internal IM system was running on a workstation with a server OS under the app managers desk for 2 years before some figured it out. Other than no one being able to use IM if something were to happen to it, it wasn't "mission critical". Luckily enough, the guy's cube was on a circuit that was tied in to the server room's UPS/Generator.
Oh, and the system was listed as Proof Of Concept. In Prod.
They do almost everything that way.