long term support release

Les Mikesell lesmikesell at gmail.com
Tue Jan 29 03:12:06 UTC 2008


seth vidal wrote:

>>>> theoretically one could update a kernel without technically
>>>> rebooting... but at what point are you just being silly to just say
>>>> you have the longest uptime (and is it uptime if you have dropped all
>>>> your services to do your update?)
>>> Think remote access, reboot is a dangerous operation. Anyway, if a
>>> reboot buys you nothing you don't reboot, do you? :-)
>>>
>> I reboot religiously. What does 5110 days of uptime buy me anyway? Not
>> even a  cup of coffee.
>>
> 
> At my last job we put a policy in place where no system would have a
> greater than 150 day update, unless it had extenuating circumstances.
> What I discovered in general was this:
> 
> - systems that haven't been rebooted in a while sometimes gather cruft
> that has not been properly laced into the startup. so it doesn't come up
> on its own. Rebooting frequently ensures that people remember to do that
> 
> - any system that "MUST BE UP ALL THE TIME" should be redundant. If it
> is not redundant and it is that important then that service is a pretty
> precarious position.
> 
> - reboots ferret out problems in hardware that you don't always see
> until a powercycle. Like a disk that will just keep on spinning provided
> it is never stopped.
> 
> 
> I agree with religiously rebooting boxes.

Having had a machine up for 4+ years until I had to move it and was too 
lazy to drag along a UPS on one of its dual power supplies, I have to 
disagree.  It is nice to have services like dhcp, dns, email, etc. 
running all the time and rebooting just covers up problems like memory 
leaks that should be fixed instead of worked around.  If something isn't 
reliable, why do you want to use it at all?

-- 
   Les Mikesell
    lesmikesell at gmail.com




More information about the devel mailing list