[TriLUG] server up for 262 days

[Jon Carnes]

I remember reading about that bug (a long time ago...), I'm sure that bug
was fixed and is no longer in the kernel.

Jon
----- Original Message -----
> We had two boxes make it to 497 days, 2 hours, and 17 or 26 minutes
> depending on the box.  i.e. They both died within 9 minutes of each other.
> (Within the +- 5 minute range of our custom heartbeat monitoring)  Do
> some math and you'll see that is about 2^32/100 seconds, the point at
which
> that particular kernel's jiffies counter rolls over.  The load on the
boxes
> spiked dramatically and they went down... down... down...  We don't know
> which particular program or kernel piece couldn't handle it, but it would
> require a lot of patience to debug. (Yes, we considered building a kernel
> with an uptime counter that doesn't start at 0. ;) )  Instead, we'll just
> reboot before that from now on.  We've got 7 other boxes +300 days.  I
> think the boxes that died were RH 5.2 boxes running 2.0.36.
>