Hi,

2013/1/23 Bálint Réczey <balint@balintreczey.hu>
Hi Quanah,

Thank you for the answer.


2013/1/22 Quanah Gibson-Mount <quanah@zimbra.com>
--On Sunday, January 20, 2013 11:58 AM +0100 Balint Reczey <balint@balintreczey.hu> wrote:

Dear Members of OpenLDAP Technical List,

Could you please help in finding and explanation and potentially a fix
for the bug reported at [1]?

Per Howard Chu:

I think it's because slapcat's attempt to get a read lock failed, probably because slapd already had a conflicting write lock on it and the deadlock detector chose slapcat to lose and slapcat doesn't retry on lock errors, unlike slapd.

The solution, of course, is to use back-mdb.
Forgive me for asking this with so little experience with OpenLDAP,
but would fixing slapcat to retry on lock errors be an other option?

It seems it could help people managing LDAP servers with high write-rate
and others who are not be able to migrate to mdb in short term.
 
I have a attached two patches hoping that they may be useful.

Cheers,
Balint