postfix-smtpd over inetd
hi list, today i have got a problem with the postfix-smtpd. on my machine there is pop-before-smtp-system running, writing the poppers-client-ip into a hash which is used by the postfix-smtpd. unfortunately this pop-before-smtp-mechanism sometimes was to slow and by this the first email often was rejected (latency of just 1-2 sec between pop and smtpd). now i tried to run the postfix-smtpd over the inetd, but using a shellskript for smtpd in the inetd-conf: #!/bin/bash /bin/sleep 2 /usr/sbin/sendmail -bs the bad news: after doing this NO rejecting was performed any longer ;-( why? i thougth the sendmail-command is a substitution by postfix and my script starts the same daemon as defined in the master.cf by postfix. does anybody out there know something about this? best regards, daniel -- GMX - Die Kommunikationsplattform im Internet. http://www.gmx.net
mailinglists@gmx.li wrote:
hi list,
today i have got a problem with the postfix-smtpd. on my machine there is pop-before-smtp-system running, writing the poppers-client-ip into a hash which is used by the postfix-smtpd. unfortunately this pop-before-smtp-mechanism sometimes was to slow and by this the first email often was rejected (latency of just 1-2 sec between pop and smtpd).
pop-befor-smtpd: # Daniel Roesen prefers this one; he feels that it does a better job # of being suitably prompt about noticing new logins, and that the # tuning that I chose is too likely to cause users to fail to be # able to send email. # # my $fi = File::Tail->new( # name => $logfile, # maxinterval => 2, # interval => 1, # adjustafter => 3, # resetafter => 30, # tail => -1, # ); try to addjust the interval, default is 5 secs ... that should be a better fix for you than playing around with inetd... -- intraDAT AG http://www.intradat.com Wilhelm-Leuschner-Strasse 7 Tel: +49 69-25629-0 D - 60329 Frankfurt am Main Fax: +49 69-25629-256 Junk mail is war. RFCs do not apply.
hi sven, thanks for your mail. i already tried to play around with this:
pop-befor-smtpd: # Daniel Roesen prefers this one; he feels that it does a better job # of being suitably prompt about noticing new logins, and that the # tuning that I chose is too likely to cause users to fail to be # able to send email. # # my $fi = File::Tail->new( # name => $logfile, # maxinterval => 2, # interval => 1, # adjustafter => 3, # resetafter => 30, # tail => -1, # );
my $fi = File::Tail->new( name => $logfile, maxinterval => 1, interval => 1, adjustafter => 3, tail => -1, ); i read the manpage of File::Tail and tried the entries you can see above. maybe this settings are a little bit heavy, but i thougth it was okay. the machine is justing doing mail... any changes you would recommend to me ? :-) best regards, daniel
hi sven, sorry for writing again..;-) but i think i did not mention that i played around with the File:.Tail stuff already BEFORE i got the problems with the too small latency between the pop-request and the smtpd-run. best regards, daniel
Daniel Quappe wrote:
hi sven,
sorry for writing again..;-) but i think i did not mention that i played around with the File:.Tail stuff already BEFORE i got the problems with the too small latency between the pop-request and the smtpd-run.
with the settings i posted a busy server runs very well .. its nearly never happened that somebody gets a reject... so i cannot help you more than say what i've done ;) whats about smtp auth? -- intraDAT AG http://www.intradat.com Wilhelm-Leuschner-Strasse 7 Tel: +49 69-25629-0 D - 60329 Frankfurt am Main Fax: +49 69-25629-256 Junk mail is war. RFCs do not apply.
hi sven,
hi sven,
sorry for writing again..;-) but i think i did not mention that i played around with the File:.Tail stuff already BEFORE i got the problems with the too small latency between the pop-request and the smtpd-run.
with the settings i posted a busy server runs very well .. its nearly never happened that somebody gets a reject... so i cannot help you more than say what i've done ;)
whats about smtp auth?
in general this is a good idea. unfortunately this mailserver is just a temporary machine just giving a customer a relay for all his emails over an isdn(!)- line...until he got a T1+-line (in the future, far away...) the corresponding mailserver is running a windows- software called "Internet Anywhere Mailserver" which can only authenticate by sending a pop-request ONE time before sending ALL mails. smtp-auth is not supported. that's it..;-) i think i have to circumvent the latency-problem by using the right combination of pop-daemon and smtpd-service. maybe it is sufficient to speed up writing the client-ip into the postfix-hash-table. do you know a good combination. i do not need to use postfix, but it is recommended. best regards, daniel
-- intraDAT AG http://www.intradat.com Wilhelm-Leuschner-Strasse 7 Tel: +49 69-25629-0 D - 60329 Frankfurt am Main Fax: +49 69-25629-256 Junk mail is war. RFCs do not apply.
-- To unsubscribe, e-mail: suse-security-unsubscribe@suse.com For additional commands, e-mail: suse-security-help@suse.com Security-related bug reports go to security@suse.de, not here
-- Daniel Quappe Mittwoch, der 6. März 2002 quappe@erster.de erster.de Gesellschaft fuer vernetzte Kommunikation mbH Friedrich-Ebert-Strasse 153 D-42117 Wuppertal Fon +49 (0)202 69 50 304 Fax +49 (0)202 69 50 299
#!/bin/bash
/bin/sleep 2 /usr/sbin/sendmail -bs
the bad news: after doing this NO rejecting was performed any longer ;-( why? i thougth the sendmail-command is a substitution by postfix and my script starts the same daemon as defined in the master.cf by postfix.
does anybody out there know something about this?
What about the exit-code of sendmail-call that is'nt transfered to that of the shell-script :O) exit $? Michael Appeldorn
hi michael,
What about the exit-code of sendmail-call that is'nt transfered to that of the shell-script :O)
exit $?
yes, you are right. i thought about it, but tried this creppy-code. i seemed to work (no zombies etc.). but do you think, that this could be the reason for the none-rejecting-problem? ;-) best regards, daniel
yes, you are right. i thought about it, but tried this creppy-code. i seemed to work (no zombies etc.). but do you think, that this could be the reason for the none-rejecting-problem? ;-)
you always return with no error, and so - why should there something will be rejected? Michael Appeldorn
participants (4)
-
Daniel Quappe -
mailinglists＠gmx.li
-
Michael Appeldorn -
Sven Michels