Buchan Milne wrote:
On Friday 08 February 2008 03:35:54 quanah(a)zimbra.com wrote:
> --On Thursday, February 07, 2008 8:13 PM -0500 Oren Laadan
>
> <orenl(a)cs.columbia.edu> wrote:
>> will fix the symbols now. thanks.
>>
>> as for the patch, it seems unrelated as it fixes a problem during the
>> start-up of the daemon; also, it's for 2.4.7, and I just downgraded
>> back to 2.4.49 (which I was using originally). the decision to move
>> up was because I hoped that the problem would disappear when using a
>> more recent version of openldap. as it turns out, it didn't :(
> There's no such thing as 2.4.49. I assume you mean 2.3.40? Or 2.3.39? I
> would use 2.3.40 over 2.3.39
Quoting Howard:
"While ITS#5342 is still being investigated, I would recommend that everyone
use 2.3.39 and not 2.3.40. Sorry for the trouble."
ITS#5342 is still open, and while the last two followups seem to indicate that
the corruption was not the fault of 2.3.40 ... I'm planning an upgrade for
this weekend (from 2.3.34 for one set of servers, and from 2.3.11 for
another), and thus far I'm hoping to run with 2.3.39 with most fixes in
2.3.40 (plus #5358's fix), but not the dn2id lock fix ...
It looks like 5342 was caused by an earlier corruption, and not 2.3.40. And,
running without the dn2id lock fix seemed to cause more problems. Since we
haven't been able to reproduce the corruptions reported there using just
2.3.39 or 2.3.40, I'm pretty sure that issue can be closed.
Maybe I should make provision for an export/import during my upgrade?
Might as well. You still need to regenerate any presence indices and integer
indices when upgrading to 2.3.40.
--
-- Howard Chu
Chief Architect, Symas Corp.
http://www.symas.com
Director, Highland Sun
http://highlandsun.com/hyc/
Chief Architect, OpenLDAP
http://www.openldap.org/project/