Обсуждение: [ADMIN] pgdg96 repo with centos 7.x - server starts only on loopback

hi fellas

I'm using PostgreSQL 9.6 7 - x86_64 repo with centos 7.3.
Configuration pretty standard, including this among the lines:
...
listen_addresses = 'localhost, 192.168.2.200, 10.5.6.100'
...
and yet, after system reboot I get process listening only on:

tcp        0      0 127.0.0.1:15432         0.0.0.0:*
LISTEN      5373/postmaster
tcp6       0      0 ::1:15432               :::* LISTEN
5373/postmaster

all I have to do is to:
$ systemctl restart postgresql-9.6
to get expected results.

Is this not bizarre?
I peep into systemd's Unit and there:
...
[Unit]
Description=PostgreSQL 9.6 database server
After=syslog.target
After=network.target

seems ok, right?
What can be wrong?
many thank
L.

lejeczek <peljasz@yahoo.co.uk> writes:
> I'm using PostgreSQL 9.6 7 - x86_64 repo with centos 7.3.
> Configuration pretty standard, including this among the lines:
> ...
> listen_addresses = 'localhost, 192.168.2.200, 10.5.6.100'
> ...
> and yet, after system reboot I get process listening only on:

> tcp        0      0 127.0.0.1:15432         0.0.0.0:*
> LISTEN      5373/postmaster
> tcp6       0      0 ::1:15432               :::* LISTEN
> 5373/postmaster

> What can be wrong?

Looking into the postmaster's log would be informative, but I suspect
what is happening is that systemd is allowing the postmaster to start
before those non-default interfaces are up.  This is one of the ways
systemd sucks :-( ... it's actually quite hard to prevent that.
"After=network.target" does *not* do it, at least not without munging
the definition of that target.

I don't offhand recall the solution for this, but it's a systemd
question not a Postgres question.  Googling on that basis might help.

            regards, tom lane