If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Ars Technica)   If your server remains up for more than 144725 hours, please consult the Internet   (arstechnica.com) divider line 61
    More: Cool, internet, Windows NT, GUI  
•       •       •

6117 clicks; posted to Geek » on 30 Mar 2013 at 3:10 PM (1 year ago)   |  Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



61 Comments   (+0 »)
   
View Voting Results: Smartest and Funniest

Archived thread

First | « | 1 | 2 | » | Last | Show all
 
2013-03-30 03:13:16 PM
imgs.xkcd.com
 
2013-03-30 03:13:27 PM
Man those Quantum drives sure could take a beating.
 
2013-03-30 03:18:08 PM
I wonder if it actually being used for anything or was it just sitting there wasting electricity for 16 years.
 
2013-03-30 03:21:07 PM
That'll do, server. That'll do.
 
2013-03-30 03:25:58 PM
16+ years... no, I can't compete with that.
 
2013-03-30 03:26:40 PM
Netware?

Ayup. Netware will do that.
 
2013-03-30 03:32:12 PM
I will say Netware had it's charms. B-Trive too. I got my start in consulting writing Netware utilities... for a company that used an IBM Token ring network.

But there's a few reasons we don't use them anymore.
 
2013-03-30 03:42:01 PM
That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?
 
2013-03-30 03:44:33 PM

wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?


Aye.  If you're not doing quarterly updates, you might as well put a sign that says "HACK ME PLEASE" on the front.
 
2013-03-30 03:49:04 PM

blue_2501: wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?

Aye.  If you're not doing quarterly updates, you might as well put a sign that says "HACK ME PLEASE" on the front.


In the ars thread the guy said it was already orphaned in 2004 (when he started), not connected to the internet, and (if i understood correctly) only had IPX running on it.

So these guys have a ton of space and power they waste to keep an old netware server running, doing abosultely nothing, for over a decade.  Makes you wonder what else just have running pulling down juice and increasing the heat of the server room because they don't want to decomm anything.  Hell, I'd give them a pass if they were just using old boxes to runs SETI, Folding or some such thing.  At least those are doing something other than being very inefficient heaters in a room you are trying to cool.
 
2013-03-30 03:56:00 PM

Skyfrog: I wonder if it actually being used for anything or was it just sitting there wasting electricity for 16 years.


In the discussion thread the OP mentioned that it was decommissioned a few years ago after being a print server and file share at the start of its life.

blue_2501: Aye.  If you're not doing quarterly updates, you might as well put a sign that says "HACK ME PLEASE" on the front.


It wasn't internet connected.
 
2013-03-30 03:58:07 PM
Netware was a yucky C API, and not many people knew how to work with it, it had a steep learning curve. But then who writes in Assembler and C anymore?

It was stable as could be, but we don't have to worry about that in this new world of cheap and reliable hardware and safe failover. Microsoft has rock-stable database products that extend out oh... I dunno, one or two or 32 processors worth at least. Google and Beowulf take that a step further. Smart RAID controllers do stuff unimaginable in days of yore. You don't have to *care* if a particular piece of hardware lasts a million years any more, though it's impressive in the same sense of a light bulb lasting 100 years.

So yeah... Impressive achievement, certainly, but.
 
2013-03-30 03:58:49 PM
Aaaah, I see this guy finally found his server
 
2013-03-30 04:19:35 PM

maxheck: I will say Netware had it's charms. B-Trive too. I got my start in consulting writing Netware utilities... for a company that used an IBM Token ring network.

But there's a few reasons we don't use them anymore.


Token ring. <shudder>
 
2013-03-30 04:26:44 PM

wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?


You know, there was a time, before Microsoft came along and convinced us that the proper solution to any computing error was to "reboot it", where *any* down time would have resulted in the vendor putting four guys on an air plane to your site, and a complete post-mortem and RCA done before they would even consider leaving.
 
2013-03-30 04:28:43 PM
Sixteen and a half years later, INTEL's hard disks-a pair of full height 5.25 inch 800 MB Quantum SCSI devices

Worst. Video. Server. EVAR.
 
2013-03-30 04:48:24 PM
Ahhh Netware. We used to use it to share warez during lan parties on IPX. Good times.
 
2013-03-30 04:58:50 PM
Back when I worked at a place with a server farm, the servers got a reboot every Sunday at Midnight.  Every one of them, Microsoft, Netware, Linux.... If they needed it or not.

Kept the issues on the server side to a minimum.  Any issues were handled by the on call person who was standing there doing the reboots.

My last time on call there, I remember having to reboot 49 units.
 
2013-03-30 05:07:36 PM

davidphogan: blue_2501: Aye.  If you're not doing quarterly updates, you might as well put a sign that says "HACK ME PLEASE" on the front.

It wasn't internet connected.


It had IPX.  That's enough for Skynet to hack into.
 
2013-03-30 05:13:22 PM

Rent Party: wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?

You know, there was a time, before Microsoft came along and convinced us that the proper solution to any computing error was to "reboot it", where *any* down time would have resulted in the vendor putting four guys on an air plane to your site, and a complete post-mortem and RCA done before they would even consider leaving.


That still happens on systems that are worth the cost. Of course, such apps generally have not been run on simple non-clustered Intel-Microsoft systems. Routine reboots are for systems that are allowed maintenance windows.
 
2013-03-30 05:16:30 PM
Wouldn't the real achievement there be actually having clients while using servers that were 15 years old?

//dedicated mainframes/etc are one thing. But these were actively used servers..
 
2013-03-30 05:24:14 PM

Rent Party: wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?

You know, there was a time, before Microsoft came along and convinced us that the proper solution to any computing error was to "reboot it", where *any* down time would have resulted in the vendor putting four guys on an air plane to your site, and a complete post-mortem and RCA done before they would even consider leaving.


I don't disagree with that.  The thing though is that having those four guys on standby, system engineers that designed and build both software and hardware to perform to that is not cheap.   MS is 'good enough' for most of the businesses out there and has a huge third party software base to choose from.  For truly mission critical apps, this is still around.

But you can't say that this Netware server was so awesome because it didn't need a reboot in 16 years.  From what I gather, it sat farking idle for most of that time.  I'd can't say I'd be surprised if I saw a windows 95 workstation sit idle just fine for 16 years, totally disconnected from the network.  When we had netware servers, before we moved to NT and later, it was not uncommon that a super busy file server would need a reboot on occasion.  We also had to deploy regular patches to address issues and vulnerabilities in the OS and NDS.
 
2013-03-30 05:30:15 PM
Novell 4.11 -- 6.7 yrs is the longest uptime without ever rebooting I have ever seen. It was in a credit union in downtown oklahoma city.
 
2013-03-30 05:34:01 PM

diaphoresis: Novell 4.11 -- 6.7 yrs is the longest uptime without ever rebooting I have ever seen. It was in a credit union in downtown oklahoma city.


I had a solaris box that I took over admin of that had been up for almost 8 years. I patched and rebooted that farker the first week.
 
2013-03-30 05:48:11 PM

GWSuperfan: diaphoresis: Novell 4.11 -- 6.7 yrs is the longest uptime without ever rebooting I have ever seen. It was in a credit union in downtown oklahoma city.

I had a solaris box that I took over admin of that had been up for almost 8 years. I patched and rebooted that farker the first week.


How much dust and debris was inside it? When I opened the case, there wasn't not 1 dust bunny.. there were several dust elephants, many dead spiders, spiderwebs, and a cockroach that got caught in the CPU fan, which had quit spinning long long ago. 2x speed CD-ROM made me giggle, along with an old tape backup and two 3.5" floppy drives.
 
2013-03-30 05:49:17 PM
Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.
 
2013-03-30 05:52:44 PM

poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.


http://en.wikipedia.org/wiki/Uninterruptible_power_supply

Just a guess
 
2013-03-30 05:55:22 PM

poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.


with a redundant UPS device, thoeretically it could be done. The one I took down had a 75-80 lb UPS device.

/I'm also skeptical, but since I had and experience with a long lasting one, I'm willing to give it a little faith
 
2013-03-30 05:56:41 PM
Hm. I just did a Google search and it looks like IPX/SPX can play nice with TCP/IP on the same Ethernet.  They can't talk to each other, of course, but they can share the same physical layer.

So it's possible this old box was doing something useful and connected to the company the network while just about everything else got upgraded to TCP/IP.

It was probably just sitting idle all that time, though.

===========================

And while this machine may be an uptime champ, there is still FAR older computer hardware in heavy use today-- Critical stuff.
 
2013-03-30 05:58:04 PM

Hand Banana: poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.

http://en.wikipedia.org/wiki/Uninterruptible_power_supply

Just a guess


While I Am well aware that these exist, I am thinking along the lines of http://en.wikipedia.org/wiki/Northeast_blackout_of_2003

Where absolutely NOTHING had power. so even if it has a backup there would be zero purpose to keep it online as nothing else would have had power to connect or use it. Plus the length of the blackout was sufficient that depending on the location it may not have had enough backup power anyways.

several other situations like hurricanes/Tornados and such which knocked out power in some areas for weeks. I personally was effected once when one of my vent servers was offline for 3 weeks while they repaired the installation.
 
2013-03-30 06:00:22 PM

diaphoresis: How much dust and debris was inside it?


Not nearly as much as you might think. It was in a proper data center, so the air was clean.
 
2013-03-30 06:01:43 PM

diaphoresis: poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.

with a redundant UPS device, thoeretically it could be done. The one I took down had a 75-80 lb UPS device.

/I'm also skeptical, but since I had and experience with a long lasting one, I'm willing to give it a little faith


If they had of said 8 or maybe even 9 years I might have believed it no matter where its located. Although I would be asking why it was never rebooted as a process and not for necessity. Like a few others have said, most places would have policy to reboot at certain times on certain intervals. So unless this was shielded from that as an experiment, I would seriously doubt the validity of it. then add in geographical local and there are several reasons why this could be 100% false and just someones bad memory forgetting THAT ONE TIME ....
 
2013-03-30 06:06:53 PM
not to mention the screenshot in the article shows it was online for only 6000 hours. Now I don't know what kind of variable size they gave to this in the programming or what happens when it reaches its max. maybe it loops around or maybe it causes an error. But to me this is another point that would need to be investigated for proof. 6000 hours is basically 250 days. not even a full year.
 
2013-03-30 06:09:05 PM

poisonedpawn78: not to mention the screenshot in the article shows it was online for only 6000 hours. Now I don't know what kind of variable size they gave to this in the programming or what happens when it reaches its max. maybe it loops around or maybe it causes an error. But to me this is another point that would need to be investigated for proof. 6000 hours is basically 250 days. not even a full year.


whoops I miss read the screenshot .. it was 6000 days not hours .. my bad.
 
2013-03-30 06:09:39 PM

GWSuperfan: diaphoresis: How much dust and debris was inside it?

Not nearly as much as you might think. It was in a proper data center, so the air was clean.


Nice.. Theirs was under a worktable in the copy/mail room.  /facepalm

poisonedpawn78: If they had of said 8 or maybe even 9 years I might have believed it no matter where its located. Although I would be asking why it was never rebooted as a process and not for necessity. Like a few others have said, most places would have policy to reboot at certain times on certain intervals. So unless this was shielded from that as an experiment, I would seriously doubt the validity of it. then add in geographical local and there are several reasons why this could be 100% false and just someones bad memory forgetting THAT ONE TIME ....

When I got to the one I was speaking of, they had told the staff it never needed to be rebooted since it was Novell. They just never touched it... since all the 'computers' in the bldg were the old black and orange screened dumb terminals, its existence was essentially forgotten until they hired me to put in all new equipment and get them off IPX/SPX..
 
2013-03-30 06:10:43 PM
upload.wikimedia.org
This computer is at 33.5 years and counting. Though not connected to the Internet.
 
2013-03-30 06:57:58 PM
If your server has a high uptime, it means it hasn't been rebooted.  Which means you haven't updated its kernel in a long time.  Which means you aren't doing your farking job.

My work to-do list currently includes installing the latest stable kernel on a whole bunch of systems.  I don't need years of uptime.  I need years of not getting pwned.
 
2013-03-30 08:35:26 PM
Polar:

This computer is at 33.5 years and counting. Though not connected to the Internet.

That computer has rather a specialized purpose.
 
2013-03-30 08:48:15 PM

poisonedpawn78: Hand Banana: poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.

http://en.wikipedia.org/wiki/Uninterruptible_power_supply

Just a guess

While I Am well aware that these exist, I am thinking along the lines of http://en.wikipedia.org/wiki/Northeast_blackout_of_2003

Where absolutely NOTHING had power. so even if it has a backup there would be zero purpose to keep it online as nothing else would have had power to connect or use it. Plus the length of the blackout was sufficient that depending on the location it may not have had enough backup power anyways.

several other situations like hurricanes/Tornados and such which knocked out power in some areas for weeks. I personally was effected once when one of my vent servers was offline for 3 weeks while they repaired the installation.


My facility has building wide UPS, keeps the whole building up for about 5-10 mins while the diesel generators spin up. We went through that whole blackout without the lights flickering. Same with the hurricane that took out power in our industrial park for 3 days and caused gas shortages.

If their facility is set up for it, nothing short of bomb dropping on them can interrupt power.
 
2013-03-30 09:27:31 PM

GWSuperfan: Netware?

Ayup. Netware will do that.


Indeed. Used to support a couple at a law firm that had been up for four and five years respectively.

A pity Novel couldn't sell anything, ever.
 
2013-03-30 10:48:09 PM

DammitIForgotMyLogin: Aaaah, I see this guy finally found his server


Huh. I got one just like that.
 
2013-03-30 10:49:19 PM

GWSuperfan: diaphoresis: How much dust and debris was inside it?

Not nearly as much as you might think. It was in a proper data center, so the air was clean.


"There's so much dust in the case you can't see *any* single component. It's like a bird's nest."
 
2013-03-31 12:43:29 AM

The universe is laughing behind your back: Rent Party: wingnut396: That's great and I'm happy for them... but whats the point?   To highlight sysadmins that keep needless systems running for 16 years?

You know, there was a time, before Microsoft came along and convinced us that the proper solution to any computing error was to "reboot it", where *any* down time would have resulted in the vendor putting four guys on an air plane to your site, and a complete post-mortem and RCA done before they would even consider leaving.

That still happens on systems that are worth the cost. Of course, such apps generally have not been run on simple non-clustered Intel-Microsoft systems. Routine reboots are for systems that are allowed maintenance windows.


This.

In fact when things get to the point that they need to send the 4 guys out to figure out what the fark happened the companies that still do it often send a 5th guy just in case the answer to the question 'how can we keep this from happening again?' is 'fire Jim.'

In the unfortunate event that the answer is 'fire Jim' and your name is Jim this 5th guy will often do such distasteful things as say 'You're fired, Jim.' and 'your boss is fired for keeping a dumb fark like you around, Jim.' and 'Security, please escort Jim from the office, with his shiat, oh and his ex boss too.'

/Seen it happen
//Fortunately my name isn't Jim.
///You don't shut down the primary backbone connection to the datacenter for 4 hours during peak load.
////Especially not at a financial firm during trading hours.
//Yay slashes!
 
2013-03-31 12:59:46 AM
I knew some guys who ran an IBM 3081 for the Air Force and that thing had been turned on before the building was finished and is was still running as a few years ago.  It would have had about a 3 decade uptime.   One of its sysadmins used to report its uptime in the weekly status reports but he would unit every week so sometimes he would report it in centuries and other times milliseconds.  For some reason, his uptime graph didn't look like any other.
 
2013-03-31 01:09:31 AM
Fun thought for this thread...the launch control computers for the Minuteman III ICBM force underwent their last major upgrade in 1996, when they changed to REACT.
I'm not a missileer, but I don't imagine those systems undergo a routine reboot either.
 
2013-03-31 02:41:38 AM

Warmachine999: Back when I worked at a place with a server farm, the servers got a reboot every Sunday at Midnight.  Every one of them, Microsoft, Netware, Linux.... If they needed it or not.

Kept the issues on the server side to a minimum.  Any issues were handled by the on call person who was standing there doing the reboots.

My last time on call there, I remember having to reboot 49 units.


I worked at a call center that auto rebooted the workstations every morning at 3am. The traffic flood of 900 workstations all rebooting over shared 10bt hubs using IPX was....interesting.
 
2013-03-31 03:12:43 AM

Malacon: poisonedpawn78: Hand Banana: poisonedpawn78: Anyone know where this server is located? because there has been quite a few power "events" which would have shut it off over the last 16 years.

I'm also calling shenanigans on this. likely its just some ones wish that it could last that long.

http://en.wikipedia.org/wiki/Uninterruptible_power_supply

Just a guess

While I Am well aware that these exist, I am thinking along the lines of http://en.wikipedia.org/wiki/Northeast_blackout_of_2003

Where absolutely NOTHING had power. so even if it has a backup there would be zero purpose to keep it online as nothing else would have had power to connect or use it. Plus the length of the blackout was sufficient that depending on the location it may not have had enough backup power anyways.

several other situations like hurricanes/Tornados and such which knocked out power in some areas for weeks. I personally was effected once when one of my vent servers was offline for 3 weeks while they repaired the installation.

My facility has building wide UPS, keeps the whole building up for about 5-10 mins while the diesel generators spin up. We went through that whole blackout without the lights flickering. Same with the hurricane that took out power in our industrial park for 3 days and caused gas shortages.

If their facility is set up for it, nothing short of bomb dropping on them can interrupt power.


We just moved out of our old data center.  It had three massive UPS systems so it had plenty of backup power.  Unfortunately, the city wouldn't authorize a permit for a backup generator and the AC was not on the UPS.  So when we lost power our servers kept running.  For about 20 minutes when the thermal protection kicked in and shut everything down because of excessive heat.

Our new data center has all the proper power equipment so that problem is now gone.
 
2013-03-31 03:14:25 AM

poisonedpawn78: so even if it has a backup there would be zero purpose to keep it online as nothing else would have had power to connect or use it.


You underestimate the desire to have longest uptime bragging rights.
 
2013-03-31 07:47:38 AM

OgreMagi: If your server has a high uptime, it means it hasn't been rebooted.  Which means you haven't updated its kernel in a long time.  Which means you aren't doing your farking job.

My work to-do list currently includes installing the latest stable kernel on a whole bunch of systems.  I don't need years of uptime.  I need years of not getting pwned.


There are many things out there that don't need their kernels updated frequently.  Or ever.
 
2013-03-31 08:35:58 AM
Linux will do that too.
 
Displayed 50 of 61 comments

First | « | 1 | 2 | » | Last | Show all

View Voting Results: Smartest and Funniest


This thread is archived, and closed to new comments.

Continue Farking
Submit a Link »






Report