Re: XEN3_DOM0 5.99.55 on grinds to a halt shortly after boot (repeatedly)

[Johan Ihrén <johani%johani.org@localhost>]

Hi,

On Jul 30, 2011, at 12:55 , Jean-Yves Migeon wrote:

> On 30.07.2011 01:42, Johan Ihrén wrote:
>>> At ddb(4) prompt, try "show event". You may also try "show event" first,
>>> then "continue", then break again in ddb and show event again.
>> 
>> All counters are in the tens or hundreds except for two:
>> 
>> event type 1: vcpu0 ioapic0 pin 20 = 708325251
>> event type 1: vcpu0 clock = 814442
>> 
>> I suspect that at least the first one qualifies as "ridiculously high" ;-)
> 
> Yep, the pin 20 is way too high.
> 
>> I'm not entirely sure what to look for here, but looking for ioapic
>> "pins" I notice that a few pins (16-19) attach to various PCI devices.
>> "Pin 20" is not one of them.
>> 
>> However, pin 20 is mentioned later on:
>> pciide0: using ioapic0 pin 20, event channel 7 for native-PCI interrupt
>> ...
>> pciide1: using ioapic0 pin 20, event channel 7 for native-PCI interrupt
> 
> I suppose that you have a harddrive connected to pciide0 (somewhere in
> dmesg you probably have an "atabus0 at pciide0").
> 
> H67 is likely to have an AHCI mode -- could you try that one out,
> please? It's generally configured in BIOS.

AHCI seems to work like a charm (only nuisance was that in AHCI-mode the disks 
were numbered differently). Subjective speed up about a gazillion times and now 
this box performs I expected it to (right now quite happily running 28 DOMUs 
striped across 3 SSDs...).

But pciide should of course also work so I guess there's still a bug in there 
somewhere? I'd be more than happy to help with further testing if anyone wants 
to look at that (except next week when I'm offline for vacation). This box is 
intended for testing.

Jean-Yves: Many, many thanks for your quick help.

Regards,

Johan