It was there before 9.00AM this morning then it suddenly disappeared in the middle of my move! I guess its been hacked again.

Bloody hell! If anyone's got Dan's e-mail, give him a shout, quick!

It has rather shrunk.. I hope the games haven't been lost!

I've sent Dan a message about it already. 9:30 am in London is 2:30 am in Portland though, so I don't suppose he's up and about.

Yes, the middle of the night problem. We'll just have to wait. Hope it's ok...

[Tuj] Did you guess the ruleset?

[Kim] No, do tell ;)

Ah. Very odd. I'll take a look.

Well done, Dan - no idea what you did, but you fixed it! :-)

It may be a serious problem. Stay tuned.

It's not behaving very well. I've asked the crew at the facility to cycle the power. I have last night's nightly backup here -- it's automatically fetched on a schedule, so nothing will be lost. But as I say, there may be a hardware problem so things might be iffy for a while.

Okay, it's going to be out of service for a while, as in multiple days, I think. I'm going to need to reconfigure my mail to point somewhere else, too. Very annoying.

(thank you, rab, for allowing me to use your server as a bulletin board.)

[Dan] Nae worries. As you know, I've hardware problems of my own.

BTW if anyone wants to email me -- though I'll be pointing my usual email at another server within a day or so -- there's always my last name at gmail dot com. Normally I don't check it very often but I'm using it now.

Get Well Soon MCiOS!

MCiOS thanks you. It would only take a moment to restart it at my home server and point the domain name at it (though it'd be a day or so before that change propagated through the intarweb), but the site gets a LOT of hits, including various kinds of attacks, and it would be a bad idea to inflict that on my home router for several reasons. Anyway, I'm going to have the machine professionally serviced, which will probably bring us into next week. This feels weird. It hasn't been down for more than a few hours since I launched it. I may give up on this colocation lark and go back to a virtual server of some kind; there are better and easier options for rapid failover when you do it that way.

[Dan] Indeed, there can't have been many days since last century when I didn't at least check in and look around at MCiOS. Still, I'm viewing it as a test of character!

Yeah, I still find myself typing "p" into the address bar as a shortcut to MCiOS. ("d" goes to Orange, "r" goes to here. Obvious reasons. But I'm so used to those shortcuts, if I use another computer it knocks me sideways for a minute.)

[Darren] Same with me! Except I use "pa", "du" and "ra".

How do you do that then? I just got 'em in me favourites.

Sounds like do-re-mi.

Nono, do is domaintools.com, re is resource-zone.com and mi is usually nothing, but right now it's mikael.jansson.be/rydis.html.

That's got to be an application for Web 3.0. Opening a webpage by whistling a merry tune.

Another reboot required this morning. I shall try and move to a different host this week.

I thought I couldn't get in last night.

For those who are able to access it - tonight [Sunday] sees a series of programmes on BBC TV 4 beginning at 2100, which are all related to the London Underground. Includes the 'Design Classics' prog about the LU map.

[rab] With server farms you tend to get what you pay for -- decent reliable dedicated servers run at least a hundred a month in dollars and not very much less in pounds. If you want to pay a lot less than that and want top quality hardware and good service, you might consider a Xen host instead. The decent providers use very good hardware and they maintain it with a lot of care because if it goes down, more customers are affected than just you. And you can't tell the difference in terms of availability and performance; if anything it'll be much better than a cheap colo because most of the time the (often very muscular) CPU is way underutilized. You never get less than the share you're paying for and more often than not you get much more when you need it. And it's spectacularly easy to fail over to another machine -- a good provider will be able to switch you to new hardware in a matter of minutes if smoke starts pouring out of the box you're on. Frankly I'm seriously considering going back to that my self, mainly for that last reason.

xensource's list and hostingfu's. Both list UK providers, though you have to scroll down to find them in the latter.

Oh, for three years I ran mcios and several other sites, both http and https, an SMTP server, a jabber server with several transports, mysql, pop3 and imap, the homebrew chat server of course and a variety of other services that came and went over the years ... all in 64MB instance of User Mode Linux; a poor man's Xen. It was a little swappy but completely responsive and reliable. It was on good hardware though. A 256M Xen host would be an Aston Martin by comparison.

[Dan] Thanks for that - I was considering a 128MB Xen Vps but was slightly concerned it might not be up to the job of serving this site, which gets something like 5k hits a day.

[rab] How about hosting in Edinburgh? Though you'd probably want to traceroute them to find out if it's really close to you in terms of network hops; I'm three miles but 12 hops and 85ms from mine because of a lack of local peering between carriers, but your situation is likely to be different. Very low latency could be handy for maintenance; you could just mount drives remotely.

(Though it's not clear from their site that they offer Xen in their Embra facility, you'd have to phone to find out. And I'm not vouching for them as a provider -- I know nothing about them.)

[Dan] That company was on my shortlist of two. Haven't got round to emailing them yet, but I'll ask if they host Xen in (or, if my hunch as to the actual location is correct, near) Edinburgh.

128 should be fine. I run an identical setup at home on 128MB Xen server, and by far the biggest memory hog is mysql. Here's the memory footprint, 12 days since the last reboot:

dan@flint:~$ free total used free shared buffers cached
Mem: 131252 110836 20416 0 4632 38480
-/+ buffers/cache: 67724 63528
Swap: 131064 2228 128836

Those are pretty healthy numbers -- just enough swapping to indicate that it doesn't have more memory than it needs, without being significant.

Yikes, looks like you're discarding newlines. Okay, look in the chat game on mcios.

Whoops, had forgotten there would be instances where white-space would be significant. Hmm, looks like I need a PRE-tag handler.

But thanks for that info - as you (probably) know this thing hangs off a mysql database and although I've tried to minimise the number of queries needed to display any particular page, when it comes down to it you've still got to through a few Mb of tables around.

s/through/throw/. Another thing I want to add is a "whoops" feature that lets you correct typos in your previous move.

The Banter Page (this page) is wider than the others at the moment, so I have to scroll right to get to the 'up' and 'next' options.

Although it might revert when Dan's post has disappeared offscreen.

Ah, it doesn't do that on Firefox.

I wouldn't object if you modified or replaced it.

BTW, the web host for xtrahost.net appears to be located in the Scotland facility, if node names are anything to go by. I'd be tempted to relocate there myself and damn the exchange rate, but unfortunately I get 160ms ping times to it from my colo and 200ms from my home. That's fine for web stuff but I'd be unhappy with shell access and any other activity involving lots of small round trips.

Not to beat it to death but according to their FAQ the Xen hosting in Scolocate, which according to Google maps is within a reasonable definition of Edinburgh; perhaps 5 miles from the big rock with the funny house on top.

I'll get onto it - work's been a bit crazy today, but this machine just dropped off the internet completely for a bit - was a bit worried cos I'd be poking around with some of the security settings earlier and thought it had been a hack. Ho hum, we all still seem to be here, and the mysql tables are all still up and running.

The Gyle, eh? Who'da thought it. A bit closer than Bush, which was guess #1. Definitely worth looking into!

A VPS has been ordered, although I managed to lock myself out of my own online banking service in the process... Anyway, it should be with us over the next few days, so I hope to be able to migrate in the next week or so. Which is just as well, cos this machine needed another reboot last night. And my internet at home has stopped working, which is the next thing to try and sort out.

It's arrived... installing stuff as we speak, but since my internet at home seems to be fubar it'll probably take longer to get sorted than it would normally.

VPS? I'm trying to follow this talk of hosting and hardware because I would like to understand how it all goes together, but it's mostly greek to me unfortunately. I read it that the computer that this site is kept on is playing up and so a new one is needed. If you've a mind to explain, rab, I'd love to know whether you own the faulty computer or lease space on someone else's? Please excuse any howlers caused by my ignorance of the technology behind the web.

Also, please excuse howlers caused by my inability to proof read my own punctuation. Sorry.

Tis ok. This should probably all be in 'Let me check my notes' but it seems appropriate here as this will be the last post before shutting down to move elsewhere.

The hosting of this site has a long and chequered history, mostly due to my trying to do it as cheaply as possible. It started off in a user account on a shared linux machine, rented out at 30quid a year. It got hacked into about three times, and the whole thing demolished, which led to downtime and me having to spend a lot of time reconstructing from various "back-ups", including (on one memorable occasion) the Google cache. After the third of these incidents, Nik kindly stepped in and offered - free of charge - disk space and bandwidth on a linux box sitting in his front room and running on his DSL connection, and there things ran happily for a while. But it was not to be forever, and (for various reasons) a new home was needed.

At this time, my good friend Andy informed me he had bought a dedicated server - cheap because it was on old hardware. At the time we thought that old would just mean slow, but the fact that it needs rebooting twice a week indicates that old means crap. We've been trying to get it sorted out, but Andy's been migrating to VPSes, and I was looking into it when Dan gave me the filip I needed to sort it out once and for all.

And with that, I shall begin the process of moving this to there. It should be working in an hour or so - but you may not see it for a couple of days because the DNS will have to propagate. I also won't be on hand over the weekend to sort out any problems because there's a fault on my broadband line and BT are coming round on Monday to look at (and hopefully fix) it.

Who knows if this works or not.

good luck - and thanks with all efforts, rab. I shall be migrating to south Holland this weekend, to clean the windows of a windmill and to spend some time with my lovely Dutchman, so I\\\'ll be back on Monday too :o)

Bani'shed?

Yay! I always forget about that little gotcha. And penelope - congratulations on being the first to see the new place - some people might not get in for a couple of days.

[rab] Thanks, as ever.
[pen] This Dutchman is starting to sound serious...

[Cdm] Yes, he is. Do you know any rib-cracking windmill-based jokes that might make him smile a bit more?

Why to windmills make the best shops? Because they've always got a sail on.

[pen] I hope you get a bunch of flours.

(pen) Don't know any windmill jokes but you could try call him clever clogs. Is he called Wim? One scours the place for names like that. ATB, BTW.

This place is still up and running.

A most wonderful migration, one can hardly see the join. Well done, rab!

[Darren, Projoy] Good efforts.
[Rosie] Nope, he's called Jan - equally typical. I think 'Clever Clogs' is a mainly British expression, innit? He actually brought me klompen the first time he visited. They're unfortunately too big, and might go on ebay next week!;o)

[rab] Thanks for a quick flit. I can see the new place from this part of .ac.uk now.

[rab] Hadn't even noticed the change! Guess that's good praise.

Wow. It's fewer network hops from my house to here than it is to my server. A lot fewer. Farther in terms of latency though, geography won't be completely ignored. And the fact that the route from my house to my server has to go through Los Angeles and San Francisco for what I'm sure are not very good reasons has a lot to do with it. [rab] Out of curiosity, what sort of ping times are you getting?

(pen) I suspect "clever clogs" is simply light-hearted British alliteration but the Dutch are so good at English he might even geddit. (rab) I'm with Tuj, q.v.

I'm finding the site a lot more responsive now.

[Projoy] There was a slow period this morning but that doesn't seem to have happened again since - suspect someone else on the machine was using the CPU time.

[Dan] Ping time from JANET is about 24ms, not sure if that's good or bad. It's a direct hop from JANET to the xtrahost network, but this seems to take place in London rather than Edinburgh... so there you go.

[rab] pssht. 24ms is less than the distance to my DSLAM from my router. (Which I really, really don't understand.) Good enough for pretty much anything. Though I imagine in the UK low latency is probably fairly normal.

[Dan] I am quoting the right figure: "64 bytes from rab.org.uk: icmp_seq=0 ttl=54 time=24.1ms"? www.google.com comes in at about 21.6ms, and www.bbc.co.uk at 11.0ms by the same measure.

[rab] Yep. Pity you don't have local peering, but that looks more than good enough. You'll have to check from home as well, if that's where you do your maintenance. I'm really jealous. Xtrahost looks great. They even have an SLA, which a lot of consumer-level VPS providers don't.

Anyway, I'm more than persuaded that virtualization is the coming thing. The idea of a server being physically implemented on a corresponding hunk of hardware already seems a little quaint. The amount of versatility and control you get from decoupling those concepts and essentially commoditizing CPU and memory along with bandwidth and storage is almost inarguably compelling. MS is putting up a fight as they always do when a shift occurs that challenges their business model. It's amazing to me that they stay in business when every good thing that happens is a threat to them rather than an opportunity.

(Though it must be said that a 99.9% uptime service level agreement means they're allowed to be down for 8 hours a year, which some categories of business would consider unacceptable. Still, it's much better than you or I have managed lately.) (And god, how I love Google sometimes. I just typed "0.1 % in hours per year" into the search field and up it comes.)

[Dan] I like the fact you can do things like "Current local time in New Zealand" too.

Well, the last place was down for 8 hours a week so 8 hours a year should be at least a 12-fold improvement. I'm not getting too excited about these people yet - all my previous providers (bar Nik) had certain shortcomings. But I was very impressed with the quick set-up (within an hour of my funds hitting their account) and am comforted by the fact that it's not infeasible to pop round and speak to them in person if things go horribly wrong.

rab] Congratulations!

I was disappointed to read that SABRE (the global airline res. system) is to be taken off mainframes running TPF. They used to quote ridiculous donwtime costs (something like USD1m per minute), due to the fact that they handled up to 9000 transactions per second. They're porting the system to Compaq NonStop, of which I (currently) know nothing, but I'll be looking into it. A few thousand TPF apps and ops programmers could be re-training in a few year, methinks.

Hmm, PHP doesn't seem to have realised.