Wietse Venema: > Mats Luspa: > > Hello! > > > > Thank you for the exhausting explanation of the problem. > > > > Here you got the requested information about the system: > > root@outgoingmail-2:~# uname -a > > Linux outgoingmail-2 3.16.0-29-generic #39-Ubuntu SMP Mon Dec 15 > > 22:27:29 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
This looks like Ubuntu 14.10. After installing from ISO disk image in VirtualBox and catching up on all the updates since October 2014, this is what I see: $ uname -a Linux ubuntu1410 3.16.0-30-generic #40-Ubuntu SMP Mon Jan 12 22:06:37 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux On this system, Postfix 2.11.1 logging shows that the bounce service works as expected: Feb 14 14:33:21 ubuntu1410 postfix/smtp[1383]: 487714329E: to=<notex...@notexist.porcupine.org>, relay=192.168.254.1[192.168.254.1]:25, delay=0.15, delays=0.07/0.02/0.04/0.02, dsn=5.7.1, status=bounced (host 192.168.254.1[192.168.254.1] said: 554 5.7.1 <notex...@notexist.porcupine.org>: Relay access denied (in reply to RCPT TO command)) Feb 14 14:33:21 ubuntu1410 postfix/cleanup[1381]: 65769432A6: message-id=<20150214193321.65769432A6@ubuntu1410> Feb 14 14:33:21 ubuntu1410 postfix/qmgr[1211]: 65769432A6: from=<>, size=2254, nrcpt=1 (queue active) Feb 14 14:33:21 ubuntu1410 postfix/bounce[1384]: 487714329E: sender non-delivery notification: 65769432A6 Feb 14 14:33:21 ubuntu1410 postfix/qmgr[1211]: 487714329E: removed Feb 14 14:33:21 ubuntu1410 postfix/local[1385]: 65769432A6: to=<wie...@ubuntu1410.porcupine.org>, relay=local, delay=0.02, delays=0/0.01/0/0.01, dsn=2.0.0, status=sent (delivered to command: procmail -a "$EXTENSION") Feb 14 14:33:21 ubuntu1410 postfix/qmgr[1211]: 65769432A6: removed In conclusion, whatever the problem is, it is not in Postfix. My test shows that it works fine in a non-container environment on what should basically be the same kernel as what you use. Popular hype says that containers are just like the real thing, but they are in fact a complex combination of resource management tweaks to achieve isolation. All this may very well trigger bugs that don't trigger in a non-container environment. I suspect a kernel race condition. That is not something that I can or will program around in Postfix. Wietse