On 31 Aug 2020, at 13:12, Doug Denault wrote:
On Mon, 31 Aug 2020, Bill Cole wrote:
On 30 Aug 2020, at 20:24, Doug Denault wrote:
working system:
lighthouse:~> sockstat | egrep "postfix|master" | egrep ":[2\5]+"
postfix smtpd 98709 6 tcp4 *:25 *:*
postfix smtpd 98656 6 tcp4 *:25 *:*
postfix smtpd 98656 19 tcp4 127.0.0.1:53654
127.0.0.1:10023
postfix smtpd 98612 6 tcp4 *:25 *:*
root master 52014 12 tcp4 *:25 *:*
root master 52014 16 tcp4 *:587 *:*
The one in question
freeport:~> sockstat | egrep "postfix|master" | egrep ":[2\5]+"
root master 3938 13 tcp4 *:25 *:*
root master 3938 17 tcp4 *:587 *:*
That's absolutely normal. Postfix's 'master' process only spawns
smtpd processes as needed to handle SMTP sessions. As shown, it
listens on any ports that need persistent listeners, and hands off
connections to child processes.
Actual log entries and actual configuration are essential
information.
Thanks, I will look at the debugging link, perhaps I've not been
there.
That specific link explains how best to get useful support from this
list. It's also likely to be on your local machine, somewhere like
/usr/local/share/doc/postfix/DEBUG_README.
BTW I fully understand it is likely I've done something really dumb.
Not really. It is more likely to be a result of the changes in Postfix
*OR* in the surrounding environments between your old and new systems.
2.8->3.5 is most of a decade, and if you're also going FreeBSD 8->12
you've got a large collection of changes that could cause problems
inside and outside of Postfix, mostly outside. e.g. packet filter config
differences between your old machine's ipfw and whichever on-host
'firewall' you might be using on the new system.
In comparing the setup with a kerberos system, one running webmail and
a older version of the setup I am testing all work and I can not
see/find any differences.
That is why other eyes are useful. There has been substantial change
between 2.8 and 3.5, and we might see something relevant, if we were
looking at anything tangible.
I was not clear. The system receives email just fine. It will not
send. There are no log entries.
I find that surprising. Are you looking at /var/log/maillog? That is the
normal place to find log messages from Postfix. The default
/etc/syslog.conf on the base FreeBSD install since forever has included
a line like this:
mail.info /var/log/maillog
If you can stop and start postfix with nothing showing up in
/var/log/maillog, you have a deeper problem than Postfix.
The above output is from sockstat. I do not know what or if the linux
equivalent is.
The linux version of netstat has some of sockstat's features that are
lacking in FreeBSD's netstat, such as process correlation. Not really
relevant.
On all the working systems the processes listed as postfix are
present. They seem to be needed/used by the email clients to
communicate with postfix to send mail.
Not exactly. If a Postfix server is idle, it will have a master, qmgr,
and pickup process running and maybe a tlsmgr and an anvil process if it
has been active recently. All other processes are transient, as are
their sockets.
The sockstat output from a running system:
USER COMMAND PID FD PROTO LOCAL ADDRESS FOREIGN
ADDRESS
postfix smtp 7130 6 stream private/smtp
[ long list elided ...]
The postconf output is 1,000 lines.
Which is why the DEBUG_README section suggests using 'postconf -nf' to
just show the non-default settings.
I've done a full compare with the working system finding nothing I
will work on that and probably not bother anyone here further. I was
going for things to look at and seem to have gotten what's available
here. Thanks
The limit on the help you can get here is set by how much detail you can
provide about the problem. If you can't find where the log messages
which Postfix sends are going, it is going to be exceedingly difficult
for anyone to discern what exactly is failing to work.
--
Bill Cole
b...@scconsult.com or billc...@apache.org
(AKA @grumpybozo and many *@billmail.scconsult.com addresses)
Not For Hire (currently)
