Re(2): [SunRescue] a bit OT but...

[Paul Khoury]

On Wed, 15 Mar 2000 18:30:02 -0500, Tim Hauber wrote:

>rescue at sunhelp.org writes:
>>
>>Didn't I hear somewhere that you don't 'need' to reboot to upgrade to
>>Solaris 8?  Or am I wrong?
>>
>>-Jonathan
>
>Oh goody!, someone invented a kernel that can hand off running processes. 
>Somehow I think that if this is true then someone has re-defined the term
>"reboot"  As far as I know any changes that require replacing or changing
>the kernel always require a reboot, because the kernel lives in memory. 
> I suppose someone could invent a method of changing the kernel while it
>ran, but it would be an academic exercise on an extremely lightly loaded
>machine, and I imagine the code to monitor exactly where the kernel was
>executing at a given moment in time would require breakpoints built into
>the kernel itself, so you might as well just shut it down and upgrade it. 
>The trend in Linux toward dynamically loaded modules may at some point be
>utilized to make a very upgradeable while running system, but uptimes
>really aren't that big a deal anyway.  uptime before the machine has to be
>brought down, or crashes, is important, and this is the Windows weakness.  
>I don't feel a reboot for upgrade in any way besmirches the quality of the
>machine or the admin.  I mean, a machine with a year+ uptime, of course
>you are sentimentally attached to it, and proud of it, but if the machine
>needs to do something more than what the installed OS can do, record your
>uptime for posterity, and shut it down. 

Heh heh, I thought I was the only one who does "w > uptime_01".
I am proud of my system, one reason I don't like to shut it down.
But that's slightly a problem, because I'd like to add 2 more SCSI CD-ROMs to it.


 In a growing system things like
>memory very likely need to be upgraded, and in all but the very largest
>configurations that requires a full powerdown.  In smaller systems,
>running IDE disks, even disk upgrades need a powerdown.  Besides, isn't it
>some of the Solaris versions that the uptime buffer rolls over and
>restarts at 0?


I've noticed this on my OS/2 boxes (it's probably had as much an uptime as the SS2,
until my bad kitty caused a small outage that outlasted the UPS's).
Seems it rolls over at 45 days, and it's probably been on otherwise for 3+ months.