Old kernels and new compilers [LWN.net]

Old kernels and new compilers

Posted Aug 24, 2006 9:44 UTC (Thu) by nix (subscriber, #2304)
Parent article: Old kernels and new compilers

Very long uptimes, yep: I just had a forced powerdown of one machine due to a power failure (running, I think, 2.4.29) with an uptime of (497 + 140) = 637 days. It'd been running without trouble so long that when a two-year-old bug surfaced and knocked it off the net I had trouble remembering the root password to fix it...

Linux 2.4: for systems that Just Work Dammit.

(Log in to post comments)

Old kernels and new compilers

Posted Aug 24, 2006 15:33 UTC (Thu) by jbailey (subscriber, #16890) [Link]

$ uname -a
Linux ia64 2.6.8-3-mckinley-smp #1 SMP Mon Mar 14 18:23:12 MST 2005 ia64 GNU/Linux
$ uptime
11:23:49 up 459 days, 21:52, 1 user, load average: 0.00, 0.00, 0.00

While it's admittedly not being *heavily* used, I do use it for compiler and glibc builds and testing.

I don't know where people have complaints about 2.6 stability. I've had almost no problems, including on heavily used public machines. The public machines don't have this sort of uptime only because of the once-every-week-security-upgrade-reboot.

Tks,
Jeff Bailey

Old kernels and new compilers

Posted Aug 24, 2006 22:29 UTC (Thu) by cventers (subscriber, #31465) [Link]

Agreed. It's not that I haven't had 2.6 bugs, but I've never seen 2.6
just decide to panic out of the blue. If something's wrong, it's going to
bite almost immediately.

In fact, you can abuse the piss out of 2.6 (repeatedly insert buggy
kernel modules you're working on and then rmmod -f them, oopsing the
kernel over and over again) and it will still pretend like there's
nothing wrong afterwards. I've seen a 2.6 system that oopsed due to some
ATI proprietary driver nonsense that continued to work for weeks
afterwards (albeit not with the console).