Re: Also, Re: What is err=52?

[Howard Chu <hyc@symas.com>]

Dieter Kluenter wrote:

> > > But I think I realise the source of your problem now. Have you ever
> > > checked the properties of the socket? If not, set the appropiate
> > > properties (rwxrwxrwx).
> > >      
> > Hrm.  Mine are indeed set to:
> > srwxr-xr-x   1 root     other          0 Aug 17 08:51 ldapi
> >    
>
> The clients need to write to the socket.
>  
Socket permissions are dependent on the host operating system. For 
example, Linux enforces them but Solaris doesn't. On systems like 
Solaris you can only control access to the socket by controlling the 
permissions on the socket's parent directory.

> > Any idea why it works for about 5 days or so and -then- dies?  It
> > seems like if it were a path or permissions issue, it would be broken
> > immediately and never work.  In my case, it works like a charm for
> > approximately 5 days and then stops working, returning the error in
> > question.  It really "feels" like a buffer issue of some sort simply
> > because, we used to have a heavily used product that was using
> > o=NCSU,c=US and, during that time, it only took about 1 day before the
> > err=52 started showing up.  Once we make that product use
> > dc=ncsu,dc=edu, it takes significantly longer for the error to show
> > up.
> >    
If you can't see a resource leak (memory leak) in slapd itself then most 
likely your operating system kernel has a resource leak. Still, until 
you can reliably reproduce it, there's no way to tell for certain.

--
 -- Howard Chu
 Chief Architect, Symas Corp.       Director, Highland Sun
 http://www.symas.com               http://highlandsun.com/hyc
 Symas: Premier OpenSource Development and Support