OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
From: Andreas Hödle (Hoedlekwsoft.de)
Date: Sun Dec 02 2001 - 11:17:54 CST

  • Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

    Zitiere Matthias Andree <madt.e-technik.uni-dortmund.de>:

    > Andreas Hödle <Hoedlekwsoft.de> writes:
    >
    > > 1.) ext2 (CPU 70% Idle)
    > > Client Ready : 0:58 min
    > > 19 Msg./sec
    > > Server Ready : 3:13 min (first smtpd -> last mail delivered according
    > to
    > > logfiles)
    >
    > Did you have /var/spool/postfix set to +S (check with lsattr) or mount
    > it sync?

    No sync mount and i haven´t explicit tested if set to +S (will have a look at
    monday) but i thought "postfix check" set it up this way?

    >
    > > 2.) jfs (Ver. 1.0.5)
    > >
    > > Many errors.
    > > Dead-Lock on <mailbox>.lck file
    > > Corrupt files in postfix queue
    >
    > jfs 1.0.5 is not exactly current. Neither is Linux 2.4.10.

    I know. This is what SuSE 7.3 is shipped. As i have to recompile the kernel for
    xfs support anyway i could have a look with current versions.

    > > 4.) ext3
    > > Client Ready : 4:16 min
    > > 4,3 Msg./sec
    > > Server Ready : 6:12 min (first smtpd -> last mail delivered according
    > to
    > > logfiles)
    >
    > Again, Linux 2.4.10 is not exactly current. If it was +S as per lsattr
    > or mounted sync, updating to 2.4.16 may increase throughput. What
    > journal mode did you use?
    >
    > > No tests about sync behavior or other advanced stuff.
    >
    > So the test does not reflect production setups?

    No. As you can see by the used hardware this is only a short compare about speed
    of the different fs for this special setup. In the first step no tunning at all.

    --------------------------
    Andreas Hödle
    Kühn & Weyh Software GmbH

    -
    To unsubscribe, send mail to majordomopostfix.org with content
    (not subject): unsubscribe postfix-users