Re: (ITS#4216) slapd terminates unexpectedly

[hyc@symas.com]

The bug reported in #3567 was clearly due to faulty (perl) application 
code. As such, the bug you're reporting here seems to have a different 
cause. Please try reproducing it in 2.3.12. I'll note that we at Symas 
have subjected 2.3 slapd to tremendous loads using slamd, and have not 
seen any such failure, but we were running under Linux.

james.macnider@bridgewatersystems.com wrote:
> Full_Name: James MacNider
> Version: 2.1.6 and 2.2.29
> OS: Solaris 8
> URL: 
> Submission from: (NULL) (216.113.7.2)
>
>
> I think this may be the same issue as was previously reported in ITS# 3567 which
> was previously closed.
>
> I have been seen the slapd server exit (gracefully, no core) under high load
> when using the NULL backend that is included as part of the OpenLDAP
> distribution.  I've been able to reproduce this almost at will with versions
> 2.1.6 and 2.2.29 of the OpenLDAP slapd process.  I'm using SLAMD (www.slamd.org)
> to generate the load. 
>
> Tracing through the code it again seems to be that the problem is getting an
> ENOSYS code back from the call to select(), as in ITS# 3567.  Because I'm using
> the null backend I can't see how this defect could be in the backend code as
> suggested in ITS# 3567.  It is of course possible that this is a defect in
> Solaris and not OpenLDAP, but that has yet to be proven one way or the other.
>
>
>   


-- 
  -- Howard Chu
  Chief Architect, Symas Corp.  http://www.symas.com
  Director, Highland Sun        http://highlandsun.com/hyc
  OpenLDAP Core Team            http://www.openldap.org/project/