From: Mike '/m' 

For the most part, I don't view uptimes as critically as others.  I do view
unexpected and unwanted downtime critically.  There's a difference.

For example, to plan and take down a server occasionally for whatever is
fine. It is when "occasionally" becomes "frequently" 
and "plan" becomes "unplanned" that I become concerned.
 I consider all the patching needed on MS servers to be unreasonable, and
is one of the reasons why my server is running FreeBSD.

Having said that, my server in canada is running about 145+ days of uptime
(though Netcraft shows it as 120 daysor so), but if you query Apache on
that server, you'll see that Apache was bounced a week ago when I changed a
config parm.

But to your point, yes, these are kernel uptimes, not BIND or Apache uptimes.

 /m



On Sun, 17 Aug 2003 11:38:24 -0400, "Frank Haber"
 wrote:

>After noting that *all* the top 50 uptimes in web servers are BSD or FreeBSD,
>I'm moved to ask some really mindless questions.
>
>o 1300-1500da uptimes are a stunt, right?  Meaningless in any production
>application?  250 days is fine, right?
>
>o There's a bunch of BSD users out there with no load on their servers who got
>tired of eating goldfish, stuffing themselves in Volkswagens and web-mobbing,
>right?  They're doing uptime stunts?
>
>o These uptimes are kernel uptimes, right?  Nowhere is there a metric for
>Apache uptime, or BIND uptime?  Heck, in 1999 you'd have needed a
>high-resolution timer for BIND.
>
>(I'm being playful, but I really don't know.)

--- BBBS/NT v4.01 Flag-5