Problems with a Mail Server and SuSEfirewall2
Hi all ! I administer a small network of Windows and Linux boxes, which are connected to our ISP via SuSE 8.0 Pro Linux box which acts as router, firewall and mail server. The firewall is SuSEfirewall2, the mail server is sendmail 8.12.2, both pretty much out of the box, I just adapted the SuSEfirewall2- script to my needs (and had enough problems doing so !) A few months ago, our ISP had trouble with their own mail servers which led to mails not being sent at all. However, everything went smoothly after they installed another mail server, no problems since. However, I recently had to reboot my router box, which had been running continuously for months, and after the reboot the my router sent mails which were originally written and sent by users during the time of the ISP mail server problem, but had not been accepted and forwarded by the ISP at that time. I don´t understand this behaviour, especially since the regular e-mail traffic seems to be running without a hitch. I suspect that the firewall somehow blocks another sending of mails if the first one failed, but I don´t see why. Any help would be appreciated. Thanks. _____________________________________________ Dipl.-Ing. Jörg Leicher Ruhr-Universität Bochum Lehrstuhl fuer Energieanlagen und Energieprozesstechnik Gebaeude IB 3/133 D-44780 Bochum Tel: +49 (0) 234 / 32 - 27280 Fax: +49 (0) 234 / 32 - 14227 Email: joerg.leicher@leat.ruhr-uni-bochum.de URL: http://www.leat.ruhr-uni-bochum.de _____________________________________________
Hi, This doesn't look like a firewall problem. Your mailserver was not able to forward your mails to your ISP's mailserver and placed them in a queue instead. This queue was flushed on reboot. This is regular behavior. I'm not familiar with sendmail configuration, but it should be possible to flush the queue from time to time. Maybe it tried, but gave up after a specific amount of time or failed attempts. I'm currently working on a script running as a cron job which warns me when there are still messages in a specific queue (I use postfix). Regards, Holger Am Dienstag, 29. Juli 2003 09:10 schrieb remote:
Hi all !
I administer a small network of Windows and Linux boxes, which are connected to our ISP via SuSE 8.0 Pro Linux box which acts as router, firewall and mail server. The firewall is SuSEfirewall2, the mail server is sendmail 8.12.2, both pretty much out of the box, I just adapted the SuSEfirewall2- script to my needs (and had enough problems doing so !) A few months ago, our ISP had trouble with their own mail servers which led to mails not being sent at all. However, everything went smoothly after they installed another mail server, no problems since. However, I recently had to reboot my router box, which had been running continuously for months, and after the reboot the my router sent mails which were originally written and sent by users during the time of the ISP mail server problem, but had not been accepted and forwarded by the ISP at that time. I don´t understand this behaviour, especially since the regular e-mail traffic seems to be running without a hitch. I suspect that the firewall somehow blocks another sending of mails if the first one failed, but I don´t see why. Any help would be appreciated.
Thanks.
_____________________________________________
Dipl.-Ing. Jörg Leicher Ruhr-Universität Bochum Lehrstuhl fuer Energieanlagen und Energieprozesstechnik Gebaeude IB 3/133 D-44780 Bochum Tel: +49 (0) 234 / 32 - 27280 Fax: +49 (0) 234 / 32 - 14227 Email: joerg.leicher@leat.ruhr-uni-bochum.de URL: http://www.leat.ruhr-uni-bochum.de _____________________________________________
Hi On Tue, Jul 29, 2003 at 09:10:35AM +0200, remote wrote:
I don?t understand this behaviour, especially since the regular e-mail traffic seems to be running without a hitch.
Unless you mark a delivery agent as "expensive", sendmail will forward the mail immediately. Sometimes this is not possible and the mail is saved in a queue. Check the SENDMAIL_ARGS parameter in /etc/sysconfig/sendmail for something like "-q30m".
I suspect that the firewall somehow blocks another sending of mails if the first one failed, but I don?t see why.
No, it is not the firewall.
--
Stefan Tichy
Hi !
Unless you mark a delivery agent as "expensive", sendmail will forward the mail immediately. Sometimes this is not possible and the mail is saved in a queue. Check the SENDMAIL_ARGS parameter in /etc/sysconfig/sendmail for something like "-q30m".
You were right, "-q30m" ist given as an argument in /etc/sysconfig/sendmail. So, does this argument reroute outgoing mail if there is a problem with the ISP`s mail server ? If so, what can I do to flush the resulting queue from time to time, say once a week ? Another question : the things mentioned above only apply for outgoing mail, right ? But I also received mails that were months old when I rebooted my router recently. Is it for the same reason, or is there something else I should know/change in sendmail ? BTW, sorry for any rookie questions but I became admin somewhat out of the blue and have been learning by trial and error ever since, no background knowledge.:( Thanks for the help, Jörg
On Tue, 29 Jul 2003, remote wrote:
Sometimes this is not possible and the mail is saved in a queue. Check the SENDMAIL_ARGS parameter in /etc/sysconfig/sendmail for something like "-q30m".
You were right, "-q30m" ist given as an argument in /etc/sysconfig/sendmail. So, does this argument reroute outgoing mail if there is a problem with the ISP`s mail server ? If so, what can I do to flush the resulting queue from time to time, say once a week ?
This is just what "-q..." is for. The argument "-q30m" tells sendmail
to flush the queue every 30 minutes (e.g. look for mail, that have
been queued for some reason or another and retry to send any message
found in the queue).
So, if this argument was actually given to sendmail before the reboot,
then its strange, that you didn't see any try to send the mail.
sendmail (in the usual configuration) would have send a warning to the
sender if it still failed to send the message after a couple of hours
and would have given up to send the message after some days. But this
only happens while sendmail tries to resend the message, so if your
users didn't see any warning about problems to send mail, it's likely
that sendmail actually did not retry to send the mail. (Check your
logs!)
So maybe, due to some problem, that i'm unable to analyse from here,
the "-q30m" didn't make its way to sendmail's command line, although
it was given in /etc/sysconfig/sendmail?
--
Rolf Krahl
participants (4)
-
Holger Schletz
-
remote
-
Rolf Krahl
-
Stefan Andreas Tichy