xen_changeset : Mon Oct 3 05:26:29 2005 +0100 f22fba673196 updates 4k and 8 k kernel stacks

xen_changeset : Mon Oct 3 05:26:29 2005 +0100 f22fba673196 updates 4k and 8 k kernel stacks

[Ted Kaczmarek]

changeset:   7157:f22fba673196

#########################################
FC4 dom0, Tyan 2462 SMP, 4k kernel stack
#########################################

xm shutdown does not  work at all, always get the d flag , that d flag
also seems to be the stopper for xm destroy as well.
I have yet to be able to destroy a domain with the d flag.

Unable to create SMP guests

ethtool for e1000 as eth0 breaks after xend starts, don't know if this
ever has worked.



#########################################
FC4 dom0, Tyan 2462 SMP, 8k kernel stack
#########################################
smp guests get created with no issues

ethtool for e1000 as eth0 breaks after xend starts

xm shutdown works



Now at this point I don't think it offers ant more value testing 4k
stacks, if someone make a change that deems such testing worthwhile
please do let me know. I have redone my test box to be able to flip
quickly between the two, lots of stuff has to be recompiled for the
specific kernel stack.


I am also planning on working out some bridge testing scenarios.
I have 3 nics available on my 2.X and unstable test boxes.
Be more than happy to spin up some specific topologies if anyone is
interested.

Right now I am in process of trying to get a EBGP/OSPF quagga setup in
place for some network testing as well. Anything along these lines I
would be willing to toy with as well. 

Regards,
Ted

Re: xen_changeset : Mon Oct 3 05:26:29 2005 +0100 f22fba673196 updates 4k and 8 k kernel stacks

[Ewan Mellor]

On Mon, Oct 03, 2005 at 12:54:00PM -0400, Ted Kaczmarek wrote:

> changeset:   7157:f22fba673196
> 
> #########################################
> FC4 dom0, Tyan 2462 SMP, 4k kernel stack
> #########################################
> 
> xm shutdown does not  work at all, always get the d flag , that d flag
> also seems to be the stopper for xm destroy as well.
> I have yet to be able to destroy a domain with the d flag.

The d flag is most likely a symptom of your problem, not a cause, as it
indicates that something is holding the domain open.  A broken driver might do
this, for example.  I don't expect xm destroy to be able to destroy a domain
with the d flag.  Domains with that flag have _already_ been destroyed, and it
is the broken whatever-it-is that is preventing final cleanup.

Since you are also having problems with the network drivers, I'd hazard a
guess that the brokenness there is causing the domain to hang around.

Cheers,

Ewan.