On Thursday 15 September 2005 09:24, Damir Koscic wrote: > I'm not sure that title reflects my problem entirely. The problem I'm > having is that when computer crashes my ldap database is ether ruined or > outdated. Data in databases can be even few days old, e.g. changes made in > past period (of few days) are lost. It seams that databases are not synced > to disk. So, set up checkpointing, and run database recovery at startup (until you run 2.3, which does recovery at start as necessary). You should be able to find enough information on checkpointing in the available docs. > > This presents a big problem to me, because LDAP server is used as a backend > for SAMBA, and if server is inproperly shut down, Windows DOMAIN is > unusable. E.g. no one can work. > > If slapd is shut down properly everything is fine. Database is put on > separate ext3 partition on IDE disk. There is little data in database. > It is used as SAMBA backend with cca 60 entries (machines and users). > > I'm running open ldap server on Debian Linux with 2.6.11 kernel. Slapd is > version 2.1.30 and using bdb 4.2.52 as backend. No configuration options > for bdb backend are given in "slapd.conf" file. AFAIK the init script in Debian can be configured to run database recovery at startup (why this isn't the default for < 2.3 is beyond me ... instead they should make it possible to disable recovery). Regards, Buchan -- Buchan Milne ISP Systems Specialist B.Eng,RHCE(803004789010797),LPIC-2(LPI000074592)
Attachment:
pgpRVKySvJhjQ.pgp
Description: PGP signature