Now that's what I'm talkin' about, 2 of 2

[drl]

Hi.

I also like stability.

I've often been amazed at Sun hardware and software. The way they interact with customers often has been frustrating, but it's hard to beat their up-time numbers. One of my ISP's uses Suns. I'm sure this is not a record, but I know that this would have been even longer if they had not needed to power the server down to physically move it to another building.

Solid hardware, solid software ... cheers, drl

Tue Sep  5 11:17:15 CDT 2006
 11:17am  up 293 day(s),  9:11,  2 users,  load average: 0.05, 0.03, 0.03

[gtmtnbiker98]

In my opinion, that is rather average uptime for a well maintained server. I've had Linux machines up for over a year without a restart and gasping when I say this, I have come rather close to similar up time values with Windows Servers on a private network (not patched of course, no Internet connectivity).

Impressive uptimes in my opnion, depend upon server work load and particular tasks. Heavily used SQL Servers posting large uptimes are something to brag about; whereas, 400 day uptimes on Name/DNS Servers is rather child's play to achieve.

[drl]

Hi, gtmtnbiker98.

I was going to respond that you obviously had trained your children better than I had trained mine -- but, of course, I agree that the purpose can affect the up-time.

My background betrays my fascination here: I am more used to large boxes (hence the other talkin' post) which are taken down quite often for PM. Therefore to see up-times longer than a week is impressive (although getting less so). At one center for which I worked, the staff invented a super-disk controller that was really a smallish IBM mainframe. IBM customer engineers came around every few months to check on things, but that mainframe was not there for speed, just to manage a large volume of data.

I did see that one report at Netcraft that listed a top 50, the longest being around 5 years (1500+ days), but it dropped off to 800 days at the bottom. We don't know if there is a big bulge after that, what else they might be monitoring, or other factors, but there are obviously some very long-running servers, as you suggest. There is another report there that tracks servers at Apple, showing how over time the Solaris & BSD servers were replaced with OS X ... cheers, drl

http://uptime.netcraft.net/up/today/top.avg.html

[gtmtnbiker98]

I've also read stories where AS/400's run for years and years without a restart, but I haven't seen one of these in a couple of years. As for the underlying purpose of a server, I do agree that this makes all the difference in the world.

[easuter]

a linux server will reset its uptime couter every 450 days or so, meaning that netcraft's list is somewhat tainted in that respect

[drl]

Hi, easuter.

Yes, that might a contributing reason for the preponderance of BSD boxes in that list.

Now that you mention that issue, it does sound familiar, but I had not thought about it for this topic. One hit on the net suggested a better way to test for up-time. Apparently the reason for the reset is that 497+ days is about the limit of a counter, at least in 2.4. I tried the last command noted, and, while the box I'm on is not near 500 days, it does seem to track uptime ... cheers, drl

> ... You can use:
> last -xf /var/run/utmp runlevel
>
> to get true uptime in this instance.
>
> Nick

I would add that if you need to get valid outputs after such an uptime,
you can apply the vhz-j64 patch available at Robert Love's (RML) on
kernel.org.

http://lkml.org/lkml/2006/1/19/285