Vincent Lefevre wrote: > Bob Proulx wrote: > > I don't think uptime challenges are useful. It makes people want to > > do something that they shouldn't want to do. When kernel security > > upgrades come along just install them and reboot. > > That's theory. In practice, old machines get no longer supported... > I submitted a bug report (and a patch), but AFAIK the bug has never > been fixed. I upgraded everything except the kernel, without being > sure I could boot it again (udev incompatibilities...). That's why > the machine was no longer rebooted.
And if you get into a situation where the machine reboots whether you desire it or not? Power, cosmic ray hit, dead cooling fan, other? It happens. Even with UPS mains and redundant power supplies. Hardware doesn't last forever. Will it boot? If so then great. If not then you have a nasty problem to sort out and the machine is down until you do. I would rather know about it on my schedule rather than its schedule. Whenever I come across a machine that has been running continuously for a very long time one of my big worries is that someone has installed something perhaps hackishly and that the boot for it is not correct. This could mean that the machine won't boot. Or it could mean that the daemon won't be started. Or other variations. Therefore one thing that I always try to do before *I* work on a machine like that is to reboot it first. Then if there is a problem I know it is a pre-existing problem and not one that I created by the new work upon it. And I schedule it for a time convenient to me when it isn't going to be a panic. If you have a machine that will not come up from a clean boot then I think that is a scary situation to be in. Bob
signature.asc
Description: Digital signature
