[1356] in Kerberos_V5_Development
Re: kdc performance and rcache
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Thu Jun 27 21:06:28 1996
To: Jonathan Kamens <jik@annex-1-slip-jik.cam.ov.com>
Cc: bjaspan@MIT.EDU, krbdev@MIT.EDU
From: Ken Raeburn <raeburn@cygnus.com>
Date: 27 Jun 1996 21:01:25 -0400
In-Reply-To: Jonathan Kamens's message of Thu, 27 Jun 1996 18:32:47 GMT
"Barry Jaspan" <bjaspan@MIT.EDU> writes:
> can respond, and I've been doing some profiling of the KDC under these
> conditions.
>
> Fabulous! What profiling utility are you using?
I'm just using gprof. So I'm losing about 25% of the cpu time to
"mcount" (which suggests to me that we may have lots of tiny functions
that should be made macros or inline functions), and who knows how
much more to the kernel profil handling and anything that mcount needs
to call for bookkeeping.
Jonathan Kamens <jik@annex-1-slip-jik.cam.ov.com> writes:
> Our short-term fix for this was to disable the lookaside cache
> completely. However, we did eventually fix it properly, by storing
> the database modification time in the replay cache entries and not
> using replay cache entries which differed from the current database
> modification time.
There is a db_age field in there now which does get checked.
