monolithic versus modules

monolithic versus modules

[Ted Kaczmarek]

I spun a new xen kernel from pristine sources using modules, had the
APCI errors and scsi timeouts.

I took pristine source and spun a monolithic kernel and everything
appears to be working.


gcc version 3.4.4 20050721 (Red Hat 3.4.4-2)
if it matter.

xen_changeset  : Wed Oct 26 11:59:13 2005 +0100 7500:20d1a79ebe31

I have had this APCI issue on this SMP Athlon with modules for a while
now.

Guess I will stick with monolithic kernels for xen for now.


Regards,
Ted

RE: monolithic versus modules

[Ian Pratt]

 
> I spun a new xen kernel from pristine sources using modules, 
> had the APCI errors and scsi timeouts.
> 
> I took pristine source and spun a monolithic kernel and 
> everything appears to be working.
> 
> 
> gcc version 3.4.4 20050721 (Red Hat 3.4.4-2) if it matter.
> 
> xen_changeset  : Wed Oct 26 11:59:13 2005 +0100 7500:20d1a79ebe31
> 
> I have had this APCI issue on this SMP Athlon with modules 
> for a while now.
> 
> Guess I will stick with monolithic kernels for xen for now.

OK, this is interesting. Are you 100% sure about this? Are you sure you
were remebering to install the new module set and make a new initrd?
This is 32b non PAE, right?

Its hard to see why we'd have a bug like this, but it might explain why
we haven't had any luck reproing your woes.

Ian

RE: monolithic versus modules

[Ted Kaczmarek]

On Thu, 2005-10-27 at 21:45 +0100, Ian Pratt wrote:
>  > I spun a new xen kernel from pristine sources using modules, 
> > had the APCI errors and scsi timeouts.
> > 
> > I took pristine source and spun a monolithic kernel and 
> > everything appears to be working.
> > 
> > 
> > gcc version 3.4.4 20050721 (Red Hat 3.4.4-2) if it matter.
> > 
> > xen_changeset  : Wed Oct 26 11:59:13 2005 +0100 7500:20d1a79ebe31
> > 
> > I have had this APCI issue on this SMP Athlon with modules 
> > for a while now.
> > 
> > Guess I will stick with monolithic kernels for xen for now.
> 
> OK, this is interesting. Are you 100% sure about this? Are you sure you
> were remebering to install the new module set and make a new initrd?
> This is 32b non PAE, right?
> 
> Its hard to see why we'd have a bug like this, but it might explain why
> we haven't had any luck reproing your woes.
> 
I am 99.99% sure, just to be 100% I am going to do another pristine
module spin. This is indeed odd as one would assume everyone is using
modules for server type iron. I also got bit by the xen-br0 to xenbr0
change, xm create should have slapped me for thhis.

I was holding this email up until I spun a module one again but..

[tazman@pluto ~]$ telnet xenbits.xensource.com 80
Trying 217.147.82.57...
telnet: connect to address 217.147.82.57: Network is unreachable
telnet: Unable to connect to remote host: Network is unreachable

My hg co is failing :-(

Regards,
Ted

RE: monolithic versus modules

[Ian Pratt]

 > I was holding this email up until I spun a module one again but..
> 
> [tazman@pluto ~]$ telnet xenbits.xensource.com 80 Trying 
> 217.147.82.57...
> telnet: connect to address 217.147.82.57: Network is unreachable
> telnet: Unable to connect to remote host: Network is unreachable
> 
> My hg co is failing :-(

It works for me -- maybe a local network failure? 

Ian

RE: monolithic versus modules

[Ted Kaczmarek]

On Fri, 2005-10-28 at 02:03 +0100, Ian Pratt wrote:
>  > I was holding this email up until I spun a module one again but..
> > 
> > [tazman@pluto ~]$ telnet xenbits.xensource.com 80 Trying 
> > 217.147.82.57...
> > telnet: connect to address 217.147.82.57: Network is unreachable
> > telnet: Unable to connect to remote host: Network is unreachable
> > 
> > My hg co is failing :-(
> 
> It works for me -- maybe a local network failure? 
> 
> Ian

It came right back, a user in IRC also was having the same problem,
likely a link flap somewhere in the cloud. 

I think I am may be having power supply issues on this machine now, this
morning I hear the power supply fan sounding like a jet turbine every 10
seconds or so :-( 

Yesterday I did do some tests, and with a fresh spin of kernel using
modules it booted clean as a whistle.

 At this point I am ready to jump out the window. Start looking over
stuff and accidently issued a reboot, much to my surprise and recovery
of my temporarily lost sanity the APCI problem came back. Another reboot
did not change the condition. 

Until I get a new power supply I think this is a moot point though, the
PITA here is I really want something bigger than the stock 460w and am
having a hard time finding one. I ordered a 550w Enermax from
XIcomputer where I bought it, and the it blew up on power on. Nothing
like the sound of a switching power supply squealing like a pig, hehe.
The replacement they sent me wouldn't boot the toaster at all :-(

I had my eyes on a 4 Way Opteron as that would be next logical step up,
but that much extra coin is hard to come by right now.

Regards,
Ted