OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: Too many open files error when sending to large list
From: Michael Tokarev (mjttls.msk.ru)
Date: Fri Aug 18 2000 - 12:54:41 CDT

Grr, something is very bad with your Solaris, your hardware, your
postfix binary or your vacation binary, or something else.
This just should never happen. I suggest you try to see what's
happened, using at least `ps' program, or, better, `lsof' (that
is available on Solaris) when it tries to deliver and failed.
Turning on debugging in postfix also will be very useful
(add -v parameter(s) to at least local service in master.cf).

Walter Obergehrer wrote:
>
> > On 17 Aug 2000 19:04:50 +0200, Walter Obergehrer
> > <wobergehrersz-testsysteme.de> wrote:
> ...
> > >3. the request is staying in the outqueue and is sent to (almost
> > all!) users
> > >again!
> >
> > Add an alias:
> > owner-foo: postmaster
> >
> > for your alias "foo".
> >
> > And add "owner_request_special=yes"
>
> Added, but I'm not sure if I understand, why this should help?
> The messages are resent to the recipients as if there is no problem. postfix
> just does not realize, that many of the mailing list recipients already got
> the mail in question and sends it again.
>
> > >I already tried to circumvent this problem in increasing the
> > limit of open
> > >files to 256 (ulimit) - no help.
> >
> > 256 is too few -- for 300 recipients !
> >
> > Go for at least 1024.
>
> Does it mean postfix needs one fdes per mail-address in a mailing list?
> If so, what do I have to do, when I want to serve 1024+n addresses? (1024 is
> the maximum for open fildescriptors per process - reported by 'sysdef -i')
>
> I've increased to 1024 but there is no change.
>
> #Again the log reports:
> ---
> Aug 18 18:25:32 enterprise postfix/local[15140]: warning: cannot open file
> /users/huber/.forward: Too many
> open files
> Aug 18 18:25:32 enterprise postfix/local[15140]: warning: 0128714D00: defer
> service failure
> Aug 18 18:25:32 enterprise postfix/local[15140]: 0128714D00:
> to=<hubersz-testsysteme.de>, relay=local, de
> lay=14, status=deferred (cannot append to file /var/mail/huber: unable to
> create lock file /var/mail/huber
> .lock: Too many open files)
> Aug 18 18:25:53 enterprise postfix/local[15140]: 0128714D00:
> to=<maierasz-testsysteme.de>, relay=local, delay=35, status=sent
> ("|/usr/bin/vacation -j maiera ")
> ...
> # but then: there seems to be a problem with a vacation program - but I
> cannot interpret the following lines - what is going on here?
> ...
> Aug 18 18:25:53 enterprise last message repeated 38 times <------------!!!
> Aug 18 18:25:53 enterprise postfix/local[15140]: fatal: pipe_comand: pipe:
> Too many open files
> Aug 18 18:25:54 enterprise postfix/qmgr[15110]: warning: mail_scan_any: got
> EOF; expected: string
> Aug 18 18:25:54 enterprise postfix/qmgr[15110]: warning: private/local:
> malformed response
> Aug 18 18:25:54 enterprise postfix/cleanup[15170]: warning: mail_scan_any:
> got EOF; expected: integer
> Aug 18 18:25:54 enterprise postfix/cleanup[15170]: warning: BA8DC14D0F:
> skipping further client input
> Aug 18 18:25:54 enterprise postfix/cleanup[15178]: warning: mail_scan_any:
> got EOF; expected: integer
> Aug 18 18:25:54 enterprise postfix/cleanup[15178]: warning: 0948114D10:
> skipping further client input
> Aug 18 18:25:54 enterprise postfix/cleanup[15181]: warning: mail_scan_any:
> got EOF; expected: integer
> Aug 18 18:25:54 enterprise postfix/cleanup[15181]: warning: E624214D17:
> skipping further client input
> #...many entries of this kind
> Aug 18 18:25:54 enterprise postfix/master[11945]: warning: process
> /opt/postfix/libexec/local pid 15140 ex
> it status 1
> Aug 18 18:25:54 enterprise postfix/master[11945]: warning:
> /opt/postfix/libexec/local: bad command startup
> -- throttling
> ??????????????
> ----
> /Walter