postfix long lines lead to pflogsumm error message
On one of my machines my nightly pflogsumm report started to generate
lots of perl errors:
| Use of uninitialized value in length at /usr/local/bin/pflogsumm line 1472, <> line 1.
| Use of uninitialized value in substitution (s///) at /usr/local/bin/pflogsumm line 790, <> line 1.
| Use of uninitialized value in hash element at /usr/local/bin/pflogsumm line 792, <> line 1.
The offending log message looks like this:
May 14 01:54:26 ice postfix/smtp: A510E228BA: to=<[hidden email]>, orig_to=<[hidden email]>, relay=mx01.t-online.de[220.127.116.11]:25, delay=0.55, delays=0/0/0.19/0.36, dsn=5.7.0, status=bounced (host mx01.t-online.de[18.104.22.168] said: 550-5.7.0 Message considered as spam or virus, rejected 550-5.7.0 Message rejected because it was considered as spam. If you feel this 550-5.7.0 to be an error, please forward the wrong classified e-mail to our 550-5.7.0 abuse department at [hidden email] with all the header lines! 550-5.7.0 We will analyse the problem and solve it. We are sorry for any 550-5.7.0 inconvenience and thank you very much in advance for your support! 550-5.7.0 550-5.7.0 Die Annahme Ihrer Nachricht wurde abgelehnt, da sie als Spam 550-5.7.0 eingestuft wurde. Sollten Sie dies als Fehler ansehen, bitten wir 550-5.7.0 Sie darum, die E-Mail mit allen Kopfzeilen an [hidden email] 550-5.7.0 weiterzuleiten. Das Problem wird dann untersucht und geloest. 550-5.7.0 Wi
The problem appears to be that the log message gets so long that postfix
omits the closing paren around the message, and pflogsumm expects it.
The question is, should postfix always keep the ")" at the end of these
lines, or should pflogsumm be fixed not to expect it? I can produce a
patch for both things if a consensus is found about this ;)
It is too bad that the speed of light hasn't kept pace with the
changes in CPU speed and in network bandwidth. -- <[hidden email]>
Re: postfix long lines lead to pflogsumm error message
Stefan `Sec` Zehl:
> The question is, should postfix always keep the ")" at the end of these
> lines, or should pflogsumm be fixed not to expect it? I can produce a
> patch for both things if a consensus is found about this ;)