[37164] in Kerberos
Re: Strange full resync on a quiet day
daemon@ATHENA.MIT.EDU (Greg Hudson)
Sun Jul 19 13:10:02 2015
Message-ID: <55ABD9D9.5020703@mit.edu>
Date: Sun, 19 Jul 2015 13:09:45 -0400
From: Greg Hudson <ghudson@mit.edu>
MIME-Version: 1.0
To: John Devitofranceschi <jdvf@optonline.net>, kerberos@mit.edu
In-Reply-To: <9C16FBC1-3138-422A-A425-EF25DA0D18D8@optonline.net>
Content-Type: text/plain; charset="utf-8"
Errors-To: kerberos-bounces@mit.edu
Content-Transfer-Encoding: 8bit
On 07/19/2015 08:01 AM, John Devitofranceschi wrote:
> For the next 8 hours and a bit, "kpropog -h’ on the master reported this:
[...]
> Log version # : 1
> Entry block size : 2048
[...]
> A change came in at 21:30 and this seems to have reset the ulog in some way:
[...]
> Log version # : 1
> Entry block size : 4096
In order to provide random access, the ulog uses the same size for every
update record. If an update record exceeds the block size, the ulog is
reset and the block size is increased.
To the best of my understanding, large update records are most likely to
occur from overuse of the -keepold flag. Long password histories might
also cause it.
________________________________________________
Kerberos mailing list Kerberos@mit.edu
https://mailman.mit.edu/mailman/listinfo/kerberos