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: Tue Mar 12 2002 - 09:40:01 CST

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

    Wietse Venema schrieb:
    >
    > Andreas H?dle:
    > > Hello
    > >
    > > today i have found the following warnings in the maillog
    > >
    > > > Mar 12 01:29:09 mailer postfix/master[1668]: warning: master_status_event: partial status (1
    > > > bytes)
    > > > Mar 12 01:29:39 mailer postfix/master[1668]: warning: master_status_event: partial status (4
    > > > bytes)
    > > > Mar 12 01:29:42 mailer postfix/master[1668]: warning: master_status_event: partial status (3
    > > > bytes)
    > > > Mar 12 01:29:48 mailer postfix/master[1668]: warning: master_status_event: partial status (3
    > > > bytes)
    >
    > Hmm. Postfix daemon processes send little fixed-length status
    > messages to the master across a shared pipe.
    >
    > The implementation assumes that the writes are atomic. This assumption
    > is however not necessarily correct.
    >
    > The implementation uses sockets as obtained with the socketpair(2)
    > system call. Except on FreeBSD, which has full-duplex pipes. It
    > uses pipe(2) there.
    >
    > Apparently, what happens on your machine is that the socket fills
    > up with status messages. When a socket fills up, a write may succeed
    > partially, and leave some data unwritten. Postfix daemon processes
    > detect this (as an error result from master_notify()), and terminate
    > voluntarily.

    Do i get it right : The socket(-buffer) used by postfix wasn´t
    read/cleand fast enough an therefor the postfix daemon could only write
    a part of its status?
    If this is correct i wonder why this occurs in the middle of the night
    with 2-3 msg./h load.
    Can any other process block the socket operation?
    I´m not a programmer so it isn´t obvious for me where to search.
     
    > So, to summarize, it's a minor bug, but it has no significant impact
    > because the code is self-repairing.
    >
    > Wietse

    This means i can ignore the warning without getting sleepless :-)

    Thanxs for reply

    -- 
    Andreas Hödle
    

    Kühn & Weyh Software GmbH Linnestr. 1-3 79110 Freiburg

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