http://www.openldap.org/lists/openldap-software/200210/msg00111.html You can compile back-ldbm with -DLDBM_DEBUG_IDL to enable index validity checks. This will slow things down quite a bit; it may be worthwhile to build slapindex this way and leave the checks turned off when building slapd itself. However, I don't think you should need to do this any more. Any indexing bugs we knew about in 1999 (re: the message you referenced below) are fixed in the current releases. The last known indexing bug in back-ldbm was fixed in 2.0.26 (and 2.1.5; back-ldbm in 2.1 is mostly the same as in 2.0). Since back-bdb uses a completely different indexing scheme from back-ldbm, no back-ldbm indexing bugs exist in back-bdb. Of course back-bdb may have bugs of its own. But again, the last known back-bdb indexing bug was fixed in 2.1.5. -- Howard Chu Chief Architect, Symas Corp. Director, Highland Sun http://www.symas.com http://highlandsun.com/hyc Symas: Premier OpenSource Development and Support > -----Original Message----- > From: owner-openldap-software@OpenLDAP.org > [mailto:owner-openldap-software@OpenLDAP.org]On Behalf Of Kevin > McEachern > Sent: Wednesday, October 30, 2002 9:13 AM > To: openldap-software@OpenLDAP.org > Subject: RE: Follow up to Corrupt Index files > > > > > -----Original Message----- > From: Kevin McEachern > Sent: October 30, 2002 11:49 AM > To: 'Tony Earnshaw' > Subject: RE: Follow up to Corrupt Index files > > > Thank you for your reply, however some questions still remain > unanswered: > > > Why in the name of all that is wonderful, if integrity > means so much to > > you (which naturally it ought to), don't you run a test DSA with > > whatever it is you want to test? > > Data integrity is not the only issue at hand here. A major > issue is that we need to be able to DETECT data corruption > (which could occur at any time, in production or testing). > The only reason I mention this here is because in one of the > posts listed in my previous email, it was mentioned that > index corruption was the result of an application level error > (see: http://www.openldap.org/lists/openldap-software/199911/msg00024.html), and not a database error. So my (restated) questions are: Is this error corrected by using back-bdb instead of back-ldbm, and if not, are there tools (available or in development) that could be used to detect these application level failures. > Oh - and do try to make backups as the BDB 4 documents > describe? Not by doing slapcat, but by taking advantage of the > fine-grained, roll-back logging facilities offered by the latest > Berkeley technology. slapcat seems more reasonable to me and (I believe) its how we're planning on doing backups, since this eliminates dependence on a Berkeley database backend, or even OpenLDAP as a directory server. Thanks again, Kevin McEachern.
<<attachment: winmail.dat>>