[389-users] Serious write-performance problems on RHEL6 - CoS cache repeatedly rebuilding?

Steve Holden S.J.Holden at brighton.ac.uk
Tue Apr 1 21:19:01 UTC 2014


(Swapped to inline posting for this mail)

> -----Original Message-----
> From: 389-users-bounces at lists.fedoraproject.org [mailto:389-users-
> bounces at lists.fedoraproject.org] On Behalf Of thierry bordaz
> Sent: 01 April 2014 18:37
> 
> On 04/01/2014 07:04 PM, Steve Holden wrote:
> > The CoS rules are working correctly, but they massively impact directory
> > update performance.
> > If I delete all of the 15 rules, my simple LDIF update takes milliseconds.
> > If I re-add them (and wait for re-indexing), it takes 2m30s!

> > -----Original Message-----
> > From: Steve Holden
> > Sent: 01 April 2014 14:49
> >
> > After this completed, I imported the LDIF again, which should not have
> > affected any CoS attributes (just the 'title' one), but which nonetheless
> > triggered a CoS change and apparently a refresh of the entire CoS cache!
> >
> > The cache appears to being continually rebuilt...

> Hello,
> 
> The message "cos_cache_change_notify:updating due to indirect template
> change(uid=st8,ou=people,dc=brighton,dc=ac,dc=uk)" means that the cos
> cache will be rebuilt.
> 
> A possible cause is an update of
> uid=st8,ou=people,dc=brighton,dc=ac,dc=uk is related to an indirect cos
> specifier. That trigger a rebuild of the cos cache.
> 
> Would it be possible to get the cos definitions.
> Also, if you enable audit log it can help to identify the modified
> attribute in the entry.
> 
> regards
> thierry

Hi, Thierry

Thanks for responding. I'm feeling stuck!

I understand that the CoS cache will be rebuilt, but not why - or why
it should impact performance so badly, or why the CoS cache seems to
be being constantly rebuilt!

Here's the LDIF that's being imported: http://pastebin.com/NyNY650L
Here are the CoS definitions: http://pastebin.com/GeWb0VW3
No other (user-generated) activity is taking place.

As you can see, the title attribute isn't included in any of the definitions,
so I can't understand why should it cause a CoS cache rebuild :-/

Best wishes,
Steve


___________________________________________________________
This email has been scanned by MessageLabs' Email Security
System on behalf of the University of Brighton.
For more information see http://www.brighton.ac.uk/is/spam/
___________________________________________________________


More information about the 389-users mailing list