Re: serious performance regression in .41

["J. Hannken-Illjes" <hannken%eis.cs.tu-bs.de@localhost>]

On 23 May 2014, at 13:42, Christos Zoulas <christos%astron.com@localhost> wrote:

> In article <09108BB8-536E-48E2-906E-CDF6E0962F3E%eis.cs.tu-bs.de@localhost>,
> J. Hannken-Illjes <hannken%eis.cs.tu-bs.de@localhost> wrote:
>> 
>> Does the attached diff help?
> 
> No, I don't think it does.

Christos, please try it.  Of the locks involved locking the vnode is the
one that is much more expensive and intrusive as the others.

What raid level are you using?
Are the file systems mounted with "-o log"?
Null mounts?

> As I said all operation suffer, not just sync.
> Doing 150K locks and unlocks and vget/vrele without needing it makes the
> system unusable. This needs to be fixed ASAP (or the changes that caused
> it backed out) because as more people move to running HEAD they will be
> affected.

--
J. Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig 
(Germany)