Hola Torsten Landschoff! El 11/06/2005 a las 11:31 escribiste: > Sorry, but that's not a good idea. slapindex can take quite long and if > the index got corrupted chances are that the database itself it borked.
This has not been the case for me, in more than three ocassions, after the database was recovered by db4.2_recover the indexes were left in an inconsistent state, which meant we experienced weird behaviors, and the cause of them was really a pain to find. So, please, if don't think that the init script should call slapindex after calling db4.2_recover at least document this possible problem. > Nothing will help you in case of power failure (short of mounting the > relevant filesystems read-only) so I suggest to fix the real problem and > not the symptoms. I don't need to mount my filesytems in readonly to fix the problem. What do you suggest? Should I stop using berkeley db? If that's the solution I suppose the default for the slapd package should be changed as well. In the last few years great efforts have been made to have more solid systems, that is, we moved from a non journal filesystem to a journal one, we also have data journaling as an option, even in user space [1]. So, we need to find a solution to this slapd problem as well. [1] http://users.auriga.wearlab.de/~alb/libjio/ -- "I'd rather listen to [Isaac] Newton than to [MicroSoft's] Mundie. He may have been dead for almost three hundred years, but despite that he stinks up the room less." --- (Linus Torvalds) Saludos /\/\ /\ >< `/
signature.asc
Description: Digital signature
