14 ms·
Server Retired After 18 Years Running on FreeBSD 2.2.1 and a Pentium 200MHZ
- icebraining 11y agohttp://lists.ausnog.net/pipermail/ausnog/2016-January/034052.html http://lists.ausnog.net/pipermail/ausnog/2016-January/034052...
- codemusings 11y agoNews worthy! Jokes aside I just can't get over the writing style of The Register: "The post, made by a chap named Ross..." "The drive's a Seagate, for those of you..."
- ronbeltran 11y agoA picture of the $uptime command would have been great!
- oneeyedpigeon 11y agoa) It's British, which might explain some of the language and humour (with a "u"). b) I keep forgetting about 'El Reg', but whenever I return there, I enjoy my visit. It has a strong, unique tone which I find a breath of fresh air amongst all the dull, neutral news reporting out there.
- throwaway7767 11y ago> b) I keep forgetting about 'El Reg', but whenever I return there, I enjoy my visit. It has a strong, unique tone which I find a breath of fresh air amongst all the dull, neutral news reporting out there. They're basically the tabloid press of the tech news industry. But their writing style is fun, I'll grant them that.
- orf 11y agoWe have various news feeds displayed on TV here at work, The Register is one of them. It's always fun to see some of their headlines/copy pop up, the style of writing is really great (especially when they bash the Government's IT projects with colourful language).
- laumars 11y ago> It's British, which might explain some of the language and humour It's better explained by simply describing the site as a tabloid. Tabloids aside, British news reporting isn't typically that informal. > I keep forgetting about 'El Reg', but whenever I return there, I enjoy my visit. It has a strong, unique tone which I find a breath of fresh air amongst all the dull, neutral news reporting out there. Personally I find the site appalling. It's up there with the The Daily Mail for rewriting other people's articles. Granted this is basically how 99% of reporting happens these days, but these kinds of tabloids often miss the entire crux of the original piece they're plagiarizing due to their editorialisation. For example, their articles on OpenSSL parallels the Daily Mail's absurd anti-immigration propaganda. And often, again like the Daily Mail, El Reg will completely miss-report a story just for the sake of having a clickbait headline. Frankly, I'd rather have dull neutral news reporting over biased misinformation.
- collyw 11y agoIts a parody of the British tabloids and their ridiculous headlines. As such I find it's headlines hilarious sometimes.
- laumars 11y agoI'm not convinced it's a parody. That would suggest their writing style is done as a joke; referencing British tabloids while distancing themselves as a comedy version. More likely El Reg's informal writing style is a deliberate and serious decision driven by the requirement to gain more readers (read: eyeballs on their ads). ie the same reasons any tabloid is written they way they are.
- teh_klev 11y agoEl Reg since day one has always had a somewhat less than deferential style of writing. The original founder, Mike Magee[0], intended the reporting style to be tabloid-like right from the get-go. You either love or hate The Register, personally I've enjoyed reading it since 1997 and get a good chuckle each morning from some of their amusing bylines. [0]: https://en.wikipedia.org/wiki/Mike_Magee_(journalist) https://en.wikipedia.org/wiki/Mike_Magee_(journalist)
- pja 11y agoFairly informal, occasionally jokey writing is a hallmark of the British press. For us, it’s a bit of a culture shock to read the mainstream US press which takes everything very seriously.
- phillc73 11y agoI've just spent five minutes thinking, "what's wrong with this writing style?" I still don't know. As a later poster pointed out, perhaps it is a peculiarly British style, which just doesn't register as abnormal to me. There are also certain things which grate rather badly with me when reading US "style" too. For example, "Recall our previous conversation...." Use of recall in this way feels so impolite and demanding. Normal use for me would be, "You may recall our previous conversation...." Anyway, this is not a bashing session, I just genuinely didn't find anything odd about El Reg's writing style.
- TillE 11y agoIt's a very tabloid-y tone. Other commenters have used the word "informal", but that's not quite right. There's a certain over-the-topness that clearly distinguishes the tabloid press (British, American, German, etc) from a casual blog covering the same subject, for example.
- phillc73 11y agoAnd to think I am just a Guardian reader.....
- unethical_ban 11y agoWhen I hear "You may recall...", I don't think of it as asking for permission. I think of it as saying "You might be able to, or maybe not". Just saying to recall the previous conversation is an indicator that previous context is required.
- timthorn 11y agoIndeed - the loss of precision over may, can, might & could is a pet peeve of mine. Often language evolves, but with these words meaning is being lost - and sometimes inverted. I recently disagreed with someone who said that "you may not do something" (because you were allowed to), but it turned out that she meant might. She was a teacher, too. :(
- keithpeter 11y agoHaving read and enjoyed this thread and the later follow up thread on The Register, I was struck by the number of commenters who could not clearly remember the dates/machine types or who posted anachronistic descriptions. People here forging ahead with innovative hardware, why not just record brief details of dates and setups in the back of a diary or something. In 30 years time, you'll be able to start threads like this!
- mnw21cam 11y agoThat's nothing. https://news.ycombinator.com/item?id=10952016 https://news.ycombinator.com/item?id=10952016
- static_noise 11y agoWhy degrade something when you found something else that you think is better? The game of one-upping doesn't raise the quality of the discussion. - Chuck Norris
- chris_wot 11y agoHey buddy, I can degrade any criticisms of one-upmanship FAR better than you can. There's no need to involve Chuck.
- static_noise 11y agoYou could have shown it right here. On the other hand it's pointless to argue since Chuck made his point.
- oneeyedpigeon 11y ago"Some have had disks replaced, the odd battery renewed, ..." Reminds me of the old gag: "I've had that broom for 14 years. It's only had 2 new heads and 3 new handles." At least the server in the original story stayed original.
- _Codemonkeyism 11y agoThis was thought of by the Greeks, it's called 'Theseus paradox' https://en.wikipedia.org/wiki/Ship_of_Theseus https://en.wikipedia.org/wiki/Ship_of_Theseus
- saalweachter 11y agoEven better (especially when talking about things like teleportation), the Roosevelt Birthplace: https://en.wikipedia.org/wiki/Theodore_Roosevelt_Birthplace_National_Historic_Site https://en.wikipedia.org/wiki/Theodore_Roosevelt_Birthplace_... It was torn down in 1916 and then a replica was built in the same place in 1919.
- krylon 11y agoI remember a discussion on a FreeBSD mailing list, around 2003-2004, where people bragged about the impressive (though in comparison to this headline, puny) uptimes of a few years. One of the developers remarked that while he was proud the system he worked on could deliver such uptimes, having an uptime of, say, three years, on a server, also meant that a) its hardware was kind of dated and b) it had not received kernel updates (and probably no other updates, either) for as long. (Which might be okay, if your system is well tucked away behind a good firewall, but is kind of insane if it is directly reachable from the Internet.) Still, that is really impressive.
- sagischwarz 11y agoI always wondered how people got to these uptimes. I have to reboot my linux box at least once ever other week to install updates.
- bcook 11y agoKernel updates probably need a restart (or use kexec), but most other updates only require a service to be restarted, right?
- sagischwarz 11y agoThat's right. But updating everything except the kernel also seems strange.
- krylon 11y agoWell, on Linux distros it is not uncommon for individual packages to be updates as updates become available. So if there is an update to, say, the web server, it is sufficient to restart the web server. On BSD systems, kernel and userland are developed in lock step, so it's usually a good idea to reboot to be sure they are in sync.
- laumars 11y ago
- grabcocque 11y agoAnd now its watch is ended.
- bechampion 11y agoit would've been nice a photo of the uptime or something like that... i believe him tho.
- ljosa 11y agoThe authoritative DNS server for pvv.ntnu.no is still a MicroVAX II from the late 1980s. It runs an (up-to-date, I think) NetBSD. Logging in by SSH takes several minutes, even with SSH v1.
- olemartinorg 11y agoSounds just like the kind of half-crazy stuff NTNU students do when you give them some time to tinker. I heard a story from Samfundet where they were displeased with the speed of their key/value database for some payment processing stuff, so they replaced it with BIND. Even though the thing is old, I'm guessing the uptime is not that impressive?
- CaptSpify 11y agoDo you have a source for that? I'd be interested in seeing how they did it
- hucker 11y agoHe's probably refering to itkacl, which we use for authorization/acls (not payments). The README (and the code) can be found here: https://git.sesse.net/?p=itkacl;a=blob;f=README;h=3dcb059f306140838f22dd6fd8541a5a9e5a26b8;hb=HEAD https://git.sesse.net/?p=itkacl;a=blob;f=README;h=3dcb059f30...
- Tepix 11y agoRemote protocol version 1.99, remote software version OpenSSH_4.4 NetBSD_Secure_Shell-20061114 Not really up-to-date I'd say...
- digi_owl 11y agoLet me guess, they could have it if they removed it from its original location themselves?
- Christan 11y agoUndertaker had kept the name “Cain the Undertaker” during his career, read this and more on WWE http://goo.gl/ecmfBn http://goo.gl/ecmfBn
- MikeNomad 11y agoGreat run for all-original equipment. I worked at Shell's Westhollow Research Center in the mid-90s. We handled the nightmare of standardizing the desktop space (for the first time ever). A lab was decommisioning an instrument controller that had been running non-stop since they had first spun it up, fresh out of tge paking box, a decade previous. And they had never backed up any of the data. Sure, the solution was the pretty straight forward use of a stack of floppies. It was still pretty nerve-wracking having a bunch of high-powered research scientists watching over my shoulder, "making sure" I got all their research data off the machine they were too smart to ever back up themselves. Good Times.
- MikeNomad 11y ago"fresh out of the packing box" - apologies.
- geggam 11y agoI tossed out a similar system not too long ago Pentium Pro 180Mhz running OpenBSD 64MB RAM with a perl BBS averaging around 10k hits / day. Wasn't worth the electricity to run that thing. It still worked when I put it out on the corner.
- unethical_ban 11y agoIt belongs in a museum!
- rogeryu 11y agoAlmost as impressive is the fact that in 18 years, the electricity had no downtime.
- monatron 11y agoTrue, but I wouldn't be surprised if there was a UPS that this thing was plugged into
- rogeryu 11y agoYes, I thought about that too, but I guess that UPS will not survive 18 years - the battery for sure, the device probably neither. Can you replace one without shutting down the server?
- kuschku 11y agoIt's not too uncommon to have such uptimes for electricity. The last blackout I personally saw was in 2006, the famous one which killed half of Europe's grid. With average yearly downtime of 17min, it's not hard to find places that had no downtime for decades.
- FilterSweep 11y agoThats the benefit of lines below the ground :) Situation is a bit different in America
- frik 11y ago
- crishoj 11y agoIn fairness, from the article it's not actually clear whether the server literally had an uptime (as reported by the OS) of 18 years, or whether it had simply been in constant service (modulus power cuts) for 18 years.
- orblivion 11y agoThe article said "24/7".
- crishoj 11y agoYour average electrical utility also provides electricity "24/7".
- digi_owl 11y agoAny proper server should be on some kind of UPS, no?
- Thrymr 11y agoMost UPS batteries won't last anywhere close to 18 years, either. Can you hot-swap your UPS battery as well?
- Symbiote 11y agoMaybe a dual PSU, with each connected to a different UPS. Or some kind of bypass switch. Or a really fancy UPS, such as might be used in a hospital or laboratory.
- iolothebard 11y agoAnyone running old machinery that had DOS drivers would likely have older computers. I remember working on base seeing 386/486s in an aircraft hanger area that were so covered in grime I was astounded they were still used.
- deutronium 11y agoImpressive! I made a silly kernel module that 'fakes' your uptime, by patching the kernel. https://www.anfractuosity.com/projects/uptime/ https://www.anfractuosity.com/projects/uptime/
- theandrewbailey 11y agoArs Technica had an article a few years back about a machine that was up for 16 years. Had pics, too! http://arstechnica.com/information-technology/2013/03/epic-uptime-achievement-can-you-beat-16-years/ http://arstechnica.com/information-technology/2013/03/epic-u...
- ommunist 11y agoWell, I have TI-92 graphing calculator still working OK, since 1995.
- NDizzle 11y ago18 years is a really good run. I had some white-box Cisco networking equipment that had 10 year uptime. I shut it down when we closed the office they were in.
- SEJeff 11y agoThis was an old RHEL4 external dns server I ran at $last_job: http://www.digitalprognosis.com/pics/bye-bye-uptime.png http://www.digitalprognosis.com/pics/bye-bye-uptime.png I was sad that we had to shut it down, but we had to shut it down due to migrating our primary colo to another city and were going to retire all of the hardware. I'd been manually backporting bind fixes, building my own version, and had to do some config tweaks when Dan Kaminski released his DNS vulns to the world. It is always a sad day to retire an old server like that, but 18 years... What a winner! Edit: But 1158 days for an old dell 1750 running RHEL4 isn't too bad considering it serviced all kinds of external dns requests for the firm. Its secondary didn't have the uptime due to constant power issues in the backup datacenter and incompetent people managing the UPS.
- dozzie 11y agoAt my previous job I had a server for big, commercial version control system (an SCM, as people in selling those call them) that was running RHEL4 and had similar uptime. I remember my team celebrating round 1024 days of uptime.
- emcrazyone 11y agooh man, I have them so beat! I have a Slackware Linux box with similar specs. 200 MHz Pentium, 32MB of RAM, and I think I have an old 10GB barracuda 80pin SCSI drive in it connected to an Adeptec PCI SCSI card. Every so often the hard disk starts making a high pitch noise but throws no errors and the noise goes away after a few minutes. It sits on a UPS and I probably have an uptime of a few years on it now. It has been running nearly 24/7 since 1996! Only powered off when I needed to move the box from a home-office and a few rented offices over the years. When it was in the basement of my home/office, I would sometimes hear it's disks wine as I was working out (lifting weights and such). It was even in my basement through parties in my early bootstrap years. I originally bought it to run WinNT 4.0 for a new company a friend of mine and I bootstrapped. I would guess a couple years later is when I put Slackware on it. It's running a 2.0 linux kernel. It's not exposed to the public Internet. It use to be a local Samba, DHCP, and DNS server for the company. I eventually upgraded to new hardware and left this server around for redundant backups. I develop software so copies of my git repositories find their way onto this box each night. It is in no way relied upon other than to call upon it out of convenience if another server is down or being upgraded, etc... At one point the box was in the basement of my home when a small amount of water got to the basement floor and because the box sat just high enough on rubber feet, no damage. Occasionally I go back there and pull the cob webs off it. There is no SSL on it. We still telnet into it or access the SMB shares for nostalgia. It's sort of a joke in the office these days to see how long it will last or if it will simply out last us.
- autoreleasepool 11y agoWhere do you all live where the power doesn't go out a couple times a year? I'm from South Florida and I'm jealous. Tropical climates are no fun for always-on computers. I guess I don't need to worry about static discharge when I build my boxes, so that's a plus.
- emcrazyone 11y agoMichigan - and yes, power is awful in certain parts. Especially in subdivisions where much of the power cables are above ground and strung through trees... So bad that I was considering a whole house generator at one point but eventually moved to an industrial park for expanding business.... I also run equally old Trippline UPSes. They run off 220v. Every couple years we have to replace the batteries in them and other than the batteries, they still work. The UPSes pretty much keep things running for a couple hours - most of my mission critical hosting servers are are in data centers these days and for the office, our servers sit on the UPSes where at most we lose power maybe an hour every couple years in the office park. Much more often in a residential area. When that large blackout of 2003 occurred we were down for couple days. https://en.wikipedia.org/wiki/Northeast_blackout_of_2003 https://en.wikipedia.org/wiki/Northeast_blackout_of_2003
- jedberg 11y agoI used to run the FreeBSD box for sendmail.org. When I left that job in 2001 it had already been running for 2+ years. Considering that the datacenter it was in is now the Dropbox office, I'm guessing it had to be shut down and moved at some point, but 2+ years seemed like a really long time even then! FreeBSD is just really good at lasting forever.
- rconti 11y agoI don't get the "even then" part of your statement. If anything, big uptimes are a thing of an older, more monolithic era when patching was less common. It was extraordinarily common to have multi-year uptimes on important servers, whereas these days they seem far less common In my experience, the past decade has been a time of trying to be more rigorous than ever about regular patching, routine reboots, and respinning VMs to ensure that your provisioning systems work as intended, so you never again end up with these monolithic irreplaceable systems. These days, the only time I find servers with big uptimes is when they've been neglected -- they're some old bastard child of some former employee or the ancient rickety crap some department is too afraid to touch... And even then, it doesn't raise an eyebrow until it's 1000+ days.
- jedberg 11y agoBecause even 15 years ago it was rare for a server to have such a high uptime, since you usually had to unplug and move it every once in a while, and hardware still didn't last years and needed replacement.
- gtrubetskoy 11y agoFunny, in today's world, the uptime on my Linux (virtual) box is several times greater than that of the macbook within which it's hosted.
- meesterdude 11y agoThis is beautiful to me; it's ROI is off the charts from any kind of reasonable expectation. Keeping it cool certainly helped, and having it serve a role that could even exist for 18 years is another important factor.
- webkike 11y agoIt will be given the highest honor a sys admin can give a piece of hardware: casual reference to it as "what a box" in the future.
- vondur 11y agoReminds me of the old Netware servers we used to have running file services and print queues for a few computer labs at a University I worked at. Netware was really stable and we only restarted them when some of the hard disks in the raid array were dying.
- koolba 11y agoIs there a way to track uptime across kexec[1] restarts? That way you could differentiate between a hard reboot and "soft" one (ex: automated kernel upgrade). Having a system like that working for a 18 years would be insane! [1]: https://en.wikipedia.org/wiki/Kexec https://en.wikipedia.org/wiki/Kexec
- wazoox 11y agoI always had many Unix machines with high uptimes around. My home PC (Linux) typically reboots 2 or 3 times a year. My office DNS server has currently 411 days of uptime and is the best of my bunch ATM. In 2002 I had installed on the machines under my guard some program that reported uptime to some website. One of my machines, an SGI Indy workstation, had a high uptime, about 2 years. Then a new intern came, and we installed him next to the Indy. Unfortunately, his feet under the desk pulled some cables and unplugged the Indy and broke my hopes of records :)
- metaguri 11y agoI set up a FreeBSD box at a computer shop I worked at in high school. We had a T1 and static IP, so I set up some routes to make it internet accessible (my boss wanted to use it to host pictures for his eBay transactions). I set it up, threw it under a table in the corner with nothing but a power and ethernet cable, and moved on. I was surprised when 5 years later he called to ask why it had stopped working. I told him where it was, he rebooted it, and it came back. (My memory was a little fuzzy but I probably set it up in 2001-2002 and it ran until at least 2007-2008)