[geeks] Re: Metaframe [WAS] Bad practice in the Win/X86 world

[R. Lonstein]

On Mon, Apr 21, 2003 at 10:24:42PM -0500, Dan Sikorski wrote:
	[snip]
> the old terminal sever that was replaced by citrix/nfuse was scheduled
> to reboot itself EVERY NIGHT.  and it is STILL screwed up, and has to be
> rebooted right before the beginning of the day at LEAST once a day.  (It
	[snip]

My former employer was an early adopter of Winframe and then Metaframe
(how can you tell who the pioneers are? they have the arrows in their
chests...) and I led the deployment and I'll say that a well-planned
and well-managed TS+Metaframe farm should not need daily reboots. Once
we got things set up, with good hardware and realized that per-client
memory estimates were off by at least a factor of three, there were
few problems. We only had to do reboots because poorly behaved apps
(some in house, some not) would die and refuse to give up their
resources and would linger in the process list preventing their
restart. The hardest part of managing the system was integrating new
apps and determining if they would play nicely in the multi-user
environemnt. We did have was a brief period of time with Metaframe 1.0
when a necessary patch from MS had a weird interaction with one from
Citrix and the boxes would spontaneously reboot. Just black screen and
then reboot. That time was no fun and I'm glad it didn't last
long. Also glad I handed that off and went fully to Unix SA.

- Ross