Johan Andersson:
> The problem was solved with a full server reboot (rebooting the
> zone where Postfix is running did not help).
> Strange....but I'm happy for now.
I found a similar problem with Linux on some third-party website,
where a Postfix daemon has been reported to fail in fstat() with
the same error message.
The probable cause there is that the program was built in 32-bit
mode, causing the inode number for a pipe file handle to become
larger than 4 billion. Recompiling in 32-bit "large file"
mode or in 64-bit mode should fix the problem.
Wietse
>
> -----Ursprungligt meddelande-----
> Fr?n: owner-postfix-us...@postfix.org
> [mailto:owner-postfix-us...@postfix.org] F?r Johan Andersson
> Skickat: den 20 januari 2012 16:05
> Till: Postfix users
> ?mne: SV: SV: Strange Postfix error
>
> Hi.
>
> The binaries used are compiled on a Solaris server and have been in use since
> September 21st last year without any problems until yesterday when they
> suddenly stopped working. I replaced them with the exact same binaries that
> was used when the server was upgraded on September 21st (didn't help).
> They are also in use on two other servers where they are working prefectly
> (thank god for that) and they were compiled the same way as version 2.6.4 and
> 2.7.1 that were used before this version ( 2.8.5) and those versions worked
> without problems.
>
> Something happpend after a restart when I made changed?s to the parameters in
> main.cf (No changes whatsoever was made with the binaries used).
>
> Br
>
> Johan
> -----Ursprungligt meddelande-----
> Fr?n: owner-postfix-us...@postfix.org
> [mailto:owner-postfix-us...@postfix.org] F?r Wietse Venema
> Skickat: den 20 januari 2012 15:51
> Till: Postfix users
> ?mne: Re: SV: Strange Postfix error
>
> Johan Andersson:
> > Hi
> >
> > Thanks for your answer.
> >
> > I don't think that's the answer....because i haven't replaced anything
> > really (well, apart from replacing the binaries with a copy of the
> > ones that were already running).
>
> My reply was 100% about the consequences of replacing Postfix binaries of
> build X with Postfix binaries of build Y, where X does not equal Y.
>
> You now have a master daemon process of build X, executing binaries of build
> Y.
>
> Wietse
>
