lkml.org 
[lkml]   [1997]   [Oct]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Process Migration on Linux - Impossible?
On Wed, 1 Oct 1997, Andreas Kostyrka wrote:

> On 1 Oct 1997, Ketil Z Malde wrote:
> > > As has been repeatedly proven, moving an already started process is a lose
> > > almost 100% of the time.
> >
> > Right. Unless somebody just typed shutdown -hnow at the #-rompt.
> What does process migration help you in this case? *wonder*
> What if the admin is doing a shutdown because of a fork bomb? (Or some
> stuck processes in D state?)
> Any network client will be still accessing the TCP/IP stack of the node
> being brought down, right? What do you do with the CD-R being burnt? What
> do you do with some broken hardware that is really difficult to
> checkpoint. What do you do with printjob? Anything using external
> resources is bound to break anyway ;(

Real reasources would be accessed through one or more 'head' computers.
Great care would have to be taken that they dont get messed up. Also,
routers (linux boxes, modified for routing only) can be made cluster aware
and auto direct packets to the right place.. In earley implimentitons all
hardware access would be through a 'head' alone.. Later, with smart
routers the network could go directly to the target node..

Most of this stuff would not be of intrest to the linux community in
general. This all should be merged with Beowolf and become "Cluster Linux"
and be kept seperate from the main tree, hopefully there would be enough
intrest to keep the cluster linux upto date.



\
 
 \ /
  Last update: 2005-03-22 13:40    [W:0.181 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site