Hi,
We've been trying to debug a problem when bring up VMs on a 48 core
AMD machine (4 x Opteron 6128). After some investigation and some
helpful comments from #kvm, it appears that we hit a serious lock
contention issue at a certain point. We have enabled lockdep debugging
(had to increase MAX_LO
On Fri, Mar 18, 2011 at 7:30 PM, Joerg Roedel wrote:
> Hi Ben,
>
> Can you try to run
>
> # perf record -a -g
>
> for a while when your VMs are up and unresponsive?
Hi Joerg,
Thanks for the response. Here's that output:
http://paste.ubuntu.com/582345/
Looks like it's spending all its time in k
On Mon, Mar 21, 2011 at 3:35 PM, Avi Kivity wrote:
> It's actually not kvm code, but timer code. Please build qemu with
> ./configure --disable-strip and redo.
Bit of a status update:
Removing the usb tablet device helped a LOT - lowered the number of
machines we could bring up until the lock %
On Mon, Mar 21, 2011 at 5:28 PM, Ben Nagy wrote:
> Bit of a status update:
[...]
> Since the underlying configs have changed so much, I will set up, run
> all the tests again and then followup if things are still locky.
Which they are.
I will need to get the git sources in order to b
>> On Mon, Mar 21, 2011 at 7:38 PM, Avi Kivity wrote:
>> > In the future, please post the binary perf.dat.
>>
>> Hi Avi,
>>
>> How do I do that?
>
> 'make nconfig' and go to the kernel hacking section.
Imprecise question sorry, I meant how do I get the perf.dat not how do
I disable the debugging
On Mon, Mar 21, 2011 at 10:57 PM, Eric Dumazet wrote:
> Le lundi 21 mars 2011 à 19:02 +0200, Avi Kivity a écrit :
>
>> Any ideas on how to fix it? We could pre-allocate IDs and batch them in
>> per-cpu caches, but it seems like a lot of work.
>>
>
> Hmm, I dont know what syscalls kvm do, but even
On Tue, Mar 22, 2011 at 12:54 PM, Eric Dumazet wrote:
> Ben Nagy reported a scalability problem with KVM/QEMU that hit very hard
> a single spinlock (idr_lock) in posix-timers code, on its 48 core
> machine.
Hi all,
Thanks a lot for all the help so far. We've tested with Eric'
On Tue, Mar 22, 2011 at 4:20 PM, Avi Kivity wrote:
[...]
> Looks like the posix-timer issue is completely gone, to be replaced by the
> load balancer.
>
> Copying peterz.
Hi all,
I feel bad about such a big cc list, but I don't know who can be left out :/
Still got the performance issue with th
On Tue, Apr 5, 2011 at 2:48 PM, Peter Zijlstra wrote:
> On Tue, 2011-04-05 at 11:56 +0300, Avi Kivity wrote:
>>
>> Could be waking up due to guest wakeups, or qemu internal wakeups
>> (display refresh) or due to guest timer sources which are masked away in
>> the guest (if that's the case we shoul
