On Wed, Jan 27, 2021 at 03:01:43PM -0500, Alex wrote: > Some time ago I configured postfix to be multi-instance and now I'm > realizing some mail being generated locally by new crontab entries are > bouncing because apparently the main postfix instance doesn't know > where to send these emails (at least that's my guess).
No, that's not the issue. Rather, the issue is that the null client and smarthost (mta) instances have the same hostname, and you're forwarding to the smart on port 25. > Jan 27 14:21:02 xavier postfix/pickup[1247182]: B84BF4F69D: uid=1000 > from=<codec> I assume this is the default null-client local-submission instance. > Jan 27 14:21:02 xavier postfix/cleanup[1289165]: B84BF4F69D: > message-id=<[email protected]> > Jan 27 14:21:02 xavier postfix/qmgr[1247183]: B84BF4F69D: > from=<[email protected]>, size=1010, nrcpt=1 (queue active) > Jan 27 14:21:02 xavier postfix/smtp[1289167]: B84BF4F69D: > to=<[email protected]>, orig_to=<codec>, relay=none, delay=0.55, > delays=0.51/0.02/0.03/0, dsn=5.4.6, status=bounced (mail for > example.com loops back to myself) It is apparently configured with the same myhostname as the smarthost (mta) instance, which runs into loop detection issues. The simplest solution is to not forward to port 25. Configure the smarthost instance to accept mail from the null client on some other port and use that. > I'm not sure what information to provide to help troubleshoot this > without having to send the entire config, so hopefully the info below > is useful. The MX record for this domain should be using the > postfix-117 instance on this host for mail destined for my example.com > domain: > > # postmulti -l > - - y /etc/postfix > postfix-out mta y /etc/postfix-out > postfix-117 mta y /etc/postfix-117 > postfix-114 mta y /etc/postfix-114 > postfix-116 mta y /etc/postfix-116 > > I've set up the null-client instance as per the multi-instance readme. > However, mail for this example.com domain is relayed on to a remote > system from a transport entry in the postfix-out instance. The error appears to be in the null-client instance, not the (mta) instance. Configure the null client's relayhost to a destination port other than 25. Have the (mta) instance accept email from the null client there. This can be a loopback SMTP service (like a content filter but with no filtering). This is needed when the "mailhub" is on the same machine as the null client. -- Viktor.
