[Wylug-discuss] Rebooting Unix/Linux boxen...

[Jim Jackson]

On Thu, 29 Jul 2004, Mike Williams wrote:

> > I personally think its safer to have a degree of regular reboot
> > scheduling - however I would ideally do it in the few month category.
>
> I concur, for another very good reason too.
> Do you *know* your box and services will start correctly if it accidentally
> gets rebooted at 2am?

I'm glad someone mentioned this, it is THE NUMBER ONE REASON for doing a
reboot. Several times after we've had a power outage, certain services
on long uptime servers didn't start properly because of a lack of a real
test.

On Thu, 29 Jul 2004, Nigel Metheringham wrote:
......
> This is mainly to get round memory leak and similar problems:-
......
>       * After some rather longer uptime the jiffies count wraps - this
>         tends to have bad juju, and rebooting is the fastest fix.

You've said this before, and in my perversity (:-) I kept my little server
at home running well best jiffy wrap time and notice absolutely no
problems. Uptime lied of course, you had to add 400days or so. NFS, bind,
sendmail, apache, ntp, ssh (and rsh/rcp/rlogin, wash my mouth out etc etc)
and a 2.2 kernel all worked just fine. I'm sure there may be issues, just
my mix of use didn't seem to hit them.

Interestingly we used to reboot out windows servers pretty regularly here
to stave off problems, but over the last 1/2years we've given up and they
do well (windows2000 and 2003) with fairly long uptimes. The reboots
needed as part of regular security updates and maintenance seem to be
enough :-)

 Jim