Hello,
I am using OpenLDAP 2.4.25. After reboot I see logs (Logsetting -d 256) like : => bdb_idl_delete_key: c_del id failed: DB_LOCK_DEADLOCK: Locker killed to resolve a deadlock (-30994) I checked this and found that I can ignore this. But one minute after this message OpenLDAP stopped working. I automatically detected it and restarted OpenLDAP.
After restarting I see a logmessage: hdb_db_open: database "o=mydomain": unclean shutdown detected; attempting recovery.
Then I see for 1 to 3 minutes that OpenLDAP seems to work, before stopping the work again. I repeated restarting. After the sixths restarts LDAP now seems to work correctly.
So now my question: When I see the message: "unclean shutdown detected; attempting recovery" How long does the recovery take? Is it possbile that the recovery was still in progress as I restarted (after 5 mins) OpenLDAP? Is there a Logsetting I could activate to see when the recovery is finished? Would it be better to manually do the recovery after stopping slapd?
Regards, Frank
--On Wednesday, November 30, 2011 9:50 AM +0100 frank.offermanns@caseris.de wrote:
Hello,
I am using OpenLDAP 2.4.25. After reboot I see logs (Logsetting -d 256) like :
You need to cleanly shut down slapd before rebooting. And/or fix your init scripts to correctly shut down slapd when rebooting. Otherwise, you should be performing a db_recover *prior* to starting slapd after such a reboot, and you may have lost data, and certainly the database is likely not in a good state.
--Quanah
--
Quanah Gibson-Mount Sr. Member of Technical Staff Zimbra, Inc A Division of VMware, Inc. -------------------- Zimbra :: the leader in open source messaging and collaboration
Hello,
thanks a lot for your info.
You need to cleanly shut down slapd before rebooting. And/or fix your
init
scripts to correctly shut down slapd when rebooting.
Due to windows OS i cannot make sure that slapd.exe is not killed, after 40 secs at shutdown every process gets killed.
Otherwise, you should be performing a db_recover *prior* to starting slapd after such a
reboot,
and you may have lost data, and certainly the database is likely not in
a
good state.
I searched the list and found several different informations to db_recover. In older versions it was not advised to do a manual db_recover.
Can db_recover still harm a consistent database (I use db-5.1.25.NC / hdb)?
To be more precise, should I try to find out, if a unclean shutdown occured and then db_recover manually or can I db_recover with the 5.1.25 DB and Openldap 2.4.26 at every startup?
Regards, Frank
openldap-technical@openldap.org
-
frank.offermanns＠caseris.de -
Quanah Gibson-Mount