Two of our machines (one running 2.4.18-xfs-ipsec, one running
2.4.19-pre10-ac2) started spitting this message before some commands
(might be in Debian procps 2.0.7-8) after an uptime of about 248 days
(certainly in the latter case, probably the same in the former).
This smells a bit as about 248 is about half of about 497, which was a
trigger (which I managed to hit :) in the mid 2.1.x series for the uptime
counter to wrap.
Apart from the message the machines were still running fine. I rebooted
the 2.4.18 one recently after about 330-340 days' uptime due to a power
outage confusing its routing. The 2.4.19-pre10-ac2 one is our mail server
and will need rebooting sooner as this bizarre message gets mailed back to
anyone who mails any of our majordomo lists :-/
Anyway, can't grumble. Thanks for all the uptime :)
Matt
Matt Bernstein writes:
> Two of our machines (one running 2.4.18-xfs-ipsec, one
> running 2.4.19-pre10-ac2) started spitting this message
> before some commands (might be in Debian procps 2.0.7-8)
> after an uptime of about 248 days (certainly in the latter
> case, probably the same in the former).
>
> This smells a bit as about 248 is about half of about 497,
> which was a trigger (which I managed to hit :) in the
> mid 2.1.x series for the uptime counter to wrap.
Yep. Debian-unstable has procps-3.1.5 now, which won't emit
this message with 2.4.xx and 2.5.xx kernels. Your system is
still inconsistant though, so rebooting would be a good idea.
(processes may appear to have been started in the future and
run for negative time)
The recent 2.5.xx kernels should be able to handle huge
uptimes better; you'll need the procps-3.1.6 release for
the most recent 2.5.xx kernels though.
http://procps.sf.net/