Sign in / up

back to article Server retired after 18 years and ten months – beat that, readers!

The Register has learned, thanks to a post to a semi-private mailing list, of a server that has just been decommissioned after running without replacement parts since 1997. The post, made by a chap named Ross, says he “Just switched off our longest running server.” Ross says the box was “Built and brought into service in …

COMMENTS

Post your comment
House rules Send corrections
Add to 'My topics' unstar *

Page:

  1. Voland's right hand Silver badge

    Not surprising

    If you are not constrained on space a custom job is guaranteed to have a higher MTBF than an off-the-shelf server. Nearly all off-the-shelf systems are too dense and too hot to deliver anything like these numbers.

    While I am not surprised about the electronics and the disk I am still surprised about the fan MTBF. So the 64000$ question is who made the fans - I cannot think of a single fan vendor from ~ 1997 which would deliver a fan with MTBF of > 5 years (even with a "speed reduction" resistor).

    19 1 Reply
    1. Jess
      Reply Icon

      Re: Fan lasting 5 years +

      In about 2002 (+/- a year) I set up a large fan from an old olivetti 386 to act as an extractor in the entrance to my house.) It was running 24/7 (powered by an old gamegear psu).

      It packed up about 9 months ago. And that was in a harsh environment, dust, damp and temperature extremes.

      8 0 Reply
      1. David Pollard
        Reply Icon

        Re: Fan lasting 5 years +

        I set up a large fan

        That's the thing. A decade or so ago desktop fans used to be 5", now the trend has been to 3.5" or smaller, less efficient, faster running and generally more noisy.

        12 1 Reply
        1. Joe User
          Reply Icon
          Holmes

          Re: Fan lasting 5 years +

          David Pollard: A decade or so ago desktop fans used to be 5", now the trend has been to 3.5" or smaller, less efficient, faster running and generally more noisy.

          You have that backwards. Years ago, desktop machines were equipped with 80mm chassis fans. Nowadays, they usually have 120mm, 140mm, or even 200mm fans.

          5 0 Reply
          1. Daniel 19
            Reply Icon

            Re: Fan lasting 5 years +

            For server fans it is the opposite. Servers have been taking up less rack space and therefore the fans are all really small weird sizes. The are also small. The general thought is the servers will be in separate rooms anyway so who caresaid if they are loud.

            0 0 Reply
    2. Anonymous Coward
      Reply Icon
      Anonymous Coward

      Re: Not surprising

      We have a couple of NT4 servers still running on hardware that dates from 1996. We have had at least one fan and one PSU replacement along the way.

      3 2 Reply
      1. You have not yet created a handle
        Reply Icon
        Coat

        Re: Not surprising

        Same broom for 20 years eh...

        30 1 Reply
        1. Aqua Marina
          Reply Icon

          Re: Same broom for 20 years eh..

          How could anyone downvote that?

          Can't give you a medal, so I'll give you an upvote!

          As a side note, in 11 months it will be 20 years since this particular episode, and probably everyone's all time favourite.

          As another side note, the original joke was in another David Jason sit-com.

          https://www.youtube.com/watch?v=HisD_pqlRHQ

          6 1 Reply
          1. Myvekk
            Reply Icon

            Re: Same broom for 20 years eh..

            Ah yes, a variation of Grandfathers axe. And even older than that. It was a philosphical conundrum long considered, but still funny as it is always relevant in some way.

            From wiki: The ship of Theseus, also known as Theseus' paradox, is a thought experiment that raises the question of whether an object that has had all of its components replaced remains fundamentally the same object. The paradox is most notably recorded by Plutarch in Life of Theseus from the late first century. Plutarch asked whether a ship that had been restored by replacing every single wooden part remained the same ship.

            The paradox had been discussed by more ancient philosophers such as Heraclitus, Socrates, and Plato prior to Plutarch's writings; and more recently by Thomas Hobbes and John Locke. Several variants are known, including the grandfather's axe, which has had both head and handle replaced.

            4 0 Reply
            1. Anonymous Coward
              Reply Icon
              Anonymous Coward

              Re: Same broom for 20 years eh..

              "Several variants are known, including the grandfather's axe, which has had both head and handle replaced."

              A Dixon of Dock Green episode had that too. He talked about the pipe that his late wife had given him many years earlier - which had had several replacement bowls and stems. My memory suggests it was at the end of an episode and he was talking to his future? son-in-law Andy.

              Isn't memory wonderful - that has probably been dormant for at nigh on 50 years from one viewing.

              Evenin' all.

              5 0 Reply
      2. Unhandle
        Reply Icon

        Re: Not surprising

        NT3.51 would run for eternity and never need a reboot. NT4 was a little buggier due to more stuff allowed into ring0, but those were the days when a windows os could just go and go and go. Anybody still running os/2 applications? That was also extremely stable.

        2 0 Reply
        1. Manaz
          Reply Icon

          Re: Not surprising

          Yep - NT 3.51 kept printer and video drivers out of kernel mode. While chasing better printing (really?) and video performance, Microsoft allowed printer and video drivers to execute in kernel mode - meaning that bugs could cause a kernel panic or impact on memory operations of the kernel, reducing reliability in a trade-off for performance.

          0 0 Reply
        2. Anonymous Coward
          Reply Icon
          Anonymous Coward

          Re: Not surprising

          "but those were the days when a windows os could just go and go and go"

          They still do at least on the server side. We have found Server 2000, 2003, and 2008 boxes with years of uptime. Which is actually not that great as they were not patched...

          1 0 Reply
    3. Anonymous Coward
      Reply Icon
      Anonymous Coward

      Re: Not surprising

      I presume you're talking about desktop kit ... we recently permanently retired a couple of ancient dust-covered Sun Netras, which had never suffered any sort of hardware failure apart from a couple of SCSI drives which had given up. We had originally sourced all of that old kit second hand, including the SCSI drives, so no suprise about the disc failures.

      The retirement was only necessary because we're no longer running the service they support. The boxes themselves are still perfectly happy, and in a way I'm sad to see them go.

      Our disks were also Seagate, the SCA-80 / 10,000rpm variety - which do seem to last a very long time before failure as long as you let them keep spinning for years on end. Our longest continuous uptime was 3.5 years for one of the Netras, running Sparc FreeBSD ... and the downtime was only because someone managed to unplug our box by mistake.

      2 0 Reply
      1. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: Not surprising

        I second that.

        I had a customer call me about a Sun Netra that had been running happily in a corner for 14years.

        The only reason they called was to now if they could replace it if anything went wrong!

        0 0 Reply
    4. Manaz
      Reply Icon

      Re: Not surprising

      Papst and Panaflo would be the brands I'd expect to give this length of service if any would.

      Neither exist as an existing product (in terms of 80-120mm 12v DC axial fans) any more.

      0 0 Reply
    5. one-of-many
      Reply Icon

      Re: Not surprising

      At the time of construction, we used large, 24V case style fans with proper bearings, but running on the 12V rail. These ran slowly and quietly, yet moved plenty of air. The clean conditions probably helped

      them survive. All the fans were still running at the time it was switched off.

      0 0 Reply
      1. jimmyj
        Reply Icon

        Re: Not surprising

        ah yes... longer fan life : I've got two versions in my elderly XP (ha) tower -

        1) two fans connected in series at the rear - so 2 at ~ half speed '=' one

        with ~same air flow, ~ twice life & ~ half the noise! & 2) a large squirrel-

        cage axial fan kludged onto the P4 (ha) 'sink with 5 diodes in series which

        subtract ~ 2.5v from the 12 for, again, enough air (under 40*C P4) and

        much less whine even with 20% o'clocking. repurpose & rest !

        0 0 Reply
    6. John Doe 6
      Reply Icon

      Re: Not surprising

      PAPST

      1 0 Reply
  2. werdsmith Silver badge

    There's an 80386 netware box at a place that I worked at in the 1980s, that I am reliably informed by my former colleagues is still running since 1988 and doing useful work, it had managed a 10+ year uptime score before its PSU had to be replaced because the fan stopped.

    9 0 Reply
    1. Voland's right hand Silver badge
      Reply Icon

      I find this one a bit difficult to believe

      1988 means an original AT PSU with those lovely two plugs instead of the ATX 20 pin block connector which superseded it. Your chances of finding a working one in the last 10 years are hovering just about zero. While not completely impossible, whatever you find is not likely to be in a good working order.

      0 37 Reply
      1. werdsmith Silver badge
        Reply Icon

        Resourceful not your middle name then.

        I don't think I said that it happened in the last 10 years did I?

        Anyway, only a small amount of resourceful would be needed to deal with such a minor issue.

        I find it difficult to believe that you know what you are talking about.

        52 1 Reply
      2. Lee D Silver badge
        Reply Icon

        Re: I find this one a bit difficult to believe

        There are ATX -> AT power convertor cables readily available.

        14 0 Reply
      3. Allan George Dyer
        Reply Icon
        Boffin

        Re: I find this one a bit difficult to believe

        "AT PSU with those lovely two plugs" - I'm still using one of those as a bench power supply, though it is necessary to attach a 5V load (incandescent bulb) for the voltage regulation to work.

        The lovely two plugs are not my favourite design - I once plugged them into a motherboard swapped. Fortunately, the motherboard survived. Non-reversible plugs are much preferred, and non-reversible plugs where it is bloody obvious which way round is correct (unlike USB plugs) are the best.

        15 0 Reply
        1. Lockwood
          Reply Icon

          Re: I find this one a bit difficult to believe

          Back to back, black to black.

          9 0 Reply
          1. J. R. Hartley
            Reply Icon

            Re: I find this one a bit difficult to believe

            Red and red and you are dead.

            0 1 Reply
            1. Tom 38
              Reply Icon

              Re: I find this one a bit difficult to believe

              Red next to black, jump the fuck back. Red next to yella, cuddly fella

              2 0 Reply
        2. Peter Simpson 1
          Reply Icon
          Thumb Up

          Re: I find this one a bit difficult to believe

          After checking the prices of 300W 13.8V DC supplies (to power a land mobile FM transceiver used indoors), and recovering from the considerable sticker shock, I once reverse engineered and heavily modified one of these AT supplies (acquisition cost = $0) to output 13.8V at 20+A. As far as I know, it's still working fine, complete with overcurrent and overvoltage protection.

          They are versatile, cheap and highly modifiable supplies, and very underappreciated, in my opinion. It's a robust design (they're all cut from the same pattern) with many possibilities.

          4 0 Reply
        3. RIBrsiq
          Reply Icon

          Re: I find this one a bit difficult to believe

          "Non-reversible plugs are much preferred, and non-reversible plugs where it is bloody obvious which way round is correct (unlike USB plugs) are the best".

          My vote is for reversible plugs such that it doesn't matter how something is plugged in.

          It's hard to manage, but I have seen what happens when the standard Molex power connectors used for old IDE drives, which should not be reversible, are reversed nonetheless. It's not pretty.

          7 0 Reply
          1. theOtherJT Silver badge
            Reply Icon

            Re: I find this one a bit difficult to believe

            I have seen what happens when the standard Molex power connectors used for old IDE drives, which should not be reversible, are reversed nonetheless.

            Wow. How much brute force and ignorance was required to get one of those in backwards? They're hard enough to connect the right way around!

            11 0 Reply
            1. viila
              Reply Icon

              Re: I find this one a bit difficult to believe

              "Wow. How much brute force and ignorance was required to get one of those in backwards? They're hard enough to connect the right way around!"

              ...which ensures that brute force will be expected and used. Although I've managed once without much force because the socket had only the smallest hint of keying that didn't actually key anything.

              1 0 Reply
              1. Anonymous Coward
                Reply Icon
                Anonymous Coward

                Re: I find this one a bit difficult to believe

                "Although I've managed once without much force because the socket had only the smallest hint of keying that didn't actually key anything."

                Killed a 3.5" floppy drive like that - it was too easy to get the power connector in upside down when you couldn't see the connector on the drive. The only key on some was the springy tongue of plastic that was very flexible. Every drive seemed to have the power connector in a different place and possibly different orientations too.

                1 0 Reply
        4. Anonymous Coward
          Reply Icon
          Anonymous Coward

          Re: I find this one a bit difficult to believe

          "I once plugged them into a motherboard swapped"

          Did that once when getting punch-drunk swapping components to diagnose why a new motherboard's video card wouldn't work properly. I was very surprised, and gratified, that it hadn't killed the board. The question is why not? Good intrinsic design somewhere - or luck?

          0 0 Reply
      4. phuzz Silver badge
        Reply Icon

        Re: I find this one a bit difficult to believe

        No need to find an AT power supply, a brand new ATX PSU can supply all the voltages you need, so you just need to solder on the old plugs.

        (there seems to be plenty of AT supplies on ebay though)

        3 0 Reply
        1. viila
          Reply Icon
          Boffin

          Re: I find this one a bit difficult to believe

          Not necessarily. The old AT supplied and used some oddball voltages, of which -5V hasn't been in ATX standard for quite a while. Whether you can live without it depends on the peripheral cards you have.

          3 0 Reply
          1. Mr. Byte
            Reply Icon

            Re: I find this one a bit difficult to believe

            I might be wrong, but I think the -5 was used only for the RS-232 signaling on serial ports.

            0 0 Reply
            1. BinkyTheMagicPaperclip Silver badge
              Reply Icon

              Re: I find this one a bit difficult to believe

              Nope, it's used for a limited number of old ISA cards. If the ISA card is used in later motherboards that don't supply -5V, some/all of their functionality may not work. Serial ports still exist on modern kit, and there's no -5V on an ATX 2.x PSU.

              It's still possible to buy PSUs with -5V on them if necessary.

              0 0 Reply
      5. Tom 38
        Reply Icon

        Re: I find this one a bit difficult to believe

        Not only are adapters available, but its trivial to make your own using a molex crimp tool and the appropriate 3x2 and 2x10 molex socket/plugs.

        9 0 Reply
      6. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: I find this one a bit difficult to believe

        "Your chances of finding a working one in the last 10 years are hovering just about zero"

        Since the server was commisioned in 88 and he said the PSU needed replacing 10 years later, see if you can do the maths and figure out the date. Hint: it wasnt in the last 10 years.

        10 0 Reply
      7. Frank Leonhardt
        Reply Icon

        Re: I find this one a bit difficult to believe

        If you're running old hardware (to support old software) you'll have plenty of spare parts. I've got at least two spare AT PSUs within reach right now. As others have commented, it's often the fans that go but when you've got an incentive it's easy enough to adapt a standard fan to blow enough air in the general direction. And old hardware didn't get nearly as hot.

        1 0 Reply
      8. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: I find this one a bit difficult to believe

        > Your chances of finding a working one in the last 10 years are hovering just about zero.

        They are about thirty quid all up off ebay.

        1 0 Reply
      9. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: I find this one a bit difficult to believe

        How many new AT power supplies do you want...?

        http://www.nipron.com/product_detail/index.cgi?p=00504905&gclid=Cj0KEQiAq920BRC8-efn57XrotYBEiQAlVlMQ0UO0P3LjqvNCzUq9N_GuIWnxjgTV9oPYg31TJdW12MaAqiH8P8HAQ

        0 0 Reply
      10. Vic
        Reply Icon

        Re: I find this one a bit difficult to believe

        Your chances of finding a working one in the last 10 years are hovering just about zero.

        I've got one. I retired it about 5 years ago because I wanted an upgade...

        Vic.

        0 0 Reply
  3. Anonymous Coward
    Anonymous Coward

    I have a firewall running on a P3, will have to keep it going for just a few more years...

    processor : 0

    vendor_id : GenuineIntel

    cpu family : 6

    model : 7

    model name : Pentium III (Katmai)

    stepping : 3

    cpu MHz : 447.645

    cache size : 512 KB

    fdiv_bug : no

    hlt_bug : no

    f00f_bug : no

    coma_bug : no

    fpu : yes

    fpu_exception : yes

    cpuid level : 2

    wp : yes

    flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pse36 mmx fxsr sse up

    bogomips : 895.29

    clflush size : 32

    cache_alignment : 32

    address sizes : 36 bits physical, 32 bits virtual

    power management:

    2 0 Reply
    1. Voland's right hand Silver badge
      Reply Icon

      I recently retired a couple of Vias

      The power cost of running this is more than the cost of replacing it.

      A Katmai at 450MHz will consume ~ 90W of power. That at UK retail prices is 90£ per year.

      This is the exact reason why I retired my old Via based firewalls despite them being considerably more economical (~ 24W) and moved them onto my main server to run under virtualization. They were about 12 years old (across several reincarnations from case to case) at that point. I could have left them to run and they would have clocked 15 years in a couple of years time, but it was clearly not worth it financially.

      6 5 Reply
      1. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: I recently retired a couple of Vias

        I get your point but if it was just the cost of power involved I'd have replaced it years ago!

        4 0 Reply
      2. StripeyMiata
        Reply Icon

        Re: I recently retired a couple of Vias

        Where they VIA C3s? I recently retired one of mine because I had problems with the latest Linux distros and the processor's lack of CMOV support. Shame as it was a reliable (if slow) old girl.

        2 1 Reply
        1. Voland's right hand Silver badge
          Reply Icon

          Re: I recently retired a couple of Vias

          Where they VIA C3s

          One C3 (fanless M600) and one C7 (fanless S1000). 15W-22W and 20-27W measured at the wall respectively. Nowhere near what you get from an early vintage P3 (100W or thereabouts). If you are still running one of those, that is like burning money and enjoying the glow for its geekiness.

          I still miss the integrated crypto in the C7. The new crypto instructions in AMD64 set are yet to make it into stable openssl so openvpn, encrypted backupts, etc continue to be run via "brute force". The C7 could swallow hundreds of MBit of AES in its stride and not even notice. Not something I would say about any of the Intel or AMD CPUs till this day.

          However for what I used them, the case of moving to a virtualized environment was fairly clear cut (even without accounting for the fact that I worked on network virtualization at the time and it was "eat yer own dogfood").

          1 1 Reply
      3. Anonymous Coward
        Reply Icon
        Anonymous Coward

        Re: I recently retired a couple of Vias

        "This is the exact reason why I retired my old Via based firewalls despite them being considerably more economical (~ 24W) and moved them onto my main server to run under virtualization"

        A software firewall running on the same hypervisor as the main server OS is not really a firewall.

        18 2 Reply
        1. Calleb III
          Reply Icon

          Re: I recently retired a couple of Vias

          "A software firewall running on the same hypervisor as the main server OS is not really a firewall."

          Care to explain? Never heard of a real vulnerability that can seep between a VM and host running one of the major Hypervisors, or between VMs running on the same host through the hypervisor layer.

          Virtual appliances Firewalls are offered by all major vendors for a reason

          2 1 Reply

Page:

POST COMMENT House rules

Not a member of The Register? Create a new account here.

Forgotten password ?

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like