home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.kerberos
- Path: sparky!uunet!cs.utexas.edu!sun-barr!ames!agate!stanford.edu!LEMUR.CIT.CORNELL.EDU!mqh
- From: mqh@LEMUR.CIT.CORNELL.EDU (Mike Hojnowski)
- Subject: Re: The incredible missing Principle trick
- Message-ID: <9211131408.AA13130@mqh.cit.cornell.edu>
- Sender: news@shelby.stanford.edu (USENET News System)
- Organization: Internet-USENET Gateway at Stanford University
- References: <9211130336.AA02678@tsx-11.MIT.EDU>
- Date: Fri, 13 Nov 1992 14:08:30 GMT
- Lines: 23
-
-
- >One possible reason is that you got unlucky with a hash collision which
- >caused the DBM file to grow bigger than the max. allowable filesize on
- >your system, and you have a badly implemented DBM library that doesn't
- >deal well with that situation (i.e., fails silently) --- but that's only
- >a guess. At MIT, we haven't had that problem happen with our Kerberos
- >database, which has almost 20,000 entries in it. However, we have had
- >this sort of problem with our /usr/lib/aliases file on our mailhub,
- >when the length of the /usr/lib/aliases.pag gets bigger than 2
- >gigabytes.
-
- >I'm not sure I believe this is what's really happening, though, since
- >6,000 principals isn't all that many. We haven't seen your problem at
- >MIT, although we're not running our server on a RS/6000.
-
- Ted, Thanks for the idea. I don't think that's it either. Our principal.pag
- is only about 98M. I've got a call into IBM to see if anyone's reported
- DBM bugs on AIX 3. I'm also going to check my error logs on the off chance
- there was actually a disk error reported somewhere.
-
- Thanks again,
-
- Mike
-