Re: index corruption (1164) still present in 2.0.15 with db 3.1.17 (ITS#1359)

[leifj@it.su.se]

On Tue, Oct 02, 2001 at 12:03:48PM -0700, Kurt D. Zeilenga wrote:
> At 09:04 AM 2001-10-02, leifj@it.su.se wrote:
> >[ATTACHMENT TRIMMED - Available via http://www.openldap.org/its/]
> 
> Note that openldap-its@openldap.org normally forwards to the
> openldap-bugs@openldap.org mailing list.  I manually approved
> your message after trimming the attachment.  Best to provide
> a URL to such things.  You can upload to
> ftp://ftp.openldap.org/incoming/ as needed.

So should I cc to -its or -bugs??

> 
> 
> >On Tue, Oct 02, 2001 at 07:14:42AM -0700, Kurt D. Zeilenga wrote:
> >> It was done yesterday...
> >
> >I will try that! I have some additional information. Attached you
> >will find my slapd.log for one of the session which creates the index
> >corruption. I am not sure it makes a difference but I saw messages
> >from idl in it.
> 
> IIRC, those messages can occur during the normal course of
> proper operation.  But I'll dig into them a bit.
> 
> >The modifications I make are of the form

I have more info. I just tried the REL_ENG_2 with the extra debug
turned on in idl.c. I assumed that was meant to dump core (assert)
when something turned up inconsistent? It did not. Another thing
that strikes me is how _extremely_ repeatable this is. If i do a
dump+load+operations I get exactly the same corrupted entries that
I get by doing the same set of operations on the database w/o 
dump+load. Should I be surprised at this?

I may try to change my replaces to adds wherever appropriate and
see if that affects things -- since it is presumably the objectclass
index that breaks.

	MVH leifj