> The existing behavior is kept by updating configuration on upgrade.
> Why it is not sufficient?
Why should it be necessary to update the configuration if neither the
behavior should change nor the meaning/default values of the settings in
the configuration files should have changed according to the upstream
documentation?
I still have this container, so here are the system information and the
configuration file:
cat /etc/*-release
PRETTY_NAME="Debian GNU/Linux 12 (bookworm)"
NAME="Debian GNU/Linux"
VERSION_ID="12"
VERSION="12 (bookworm)"
VERSION_CODENAME=bookworm
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"
grep cyrus_sasl_config_path /etc/postfix/main.cf && echo true || echo false
false
dpkg -l | grep postfix
ii postfix 3.7.11-0+deb12u1
i386 High-performance mail transport agent
ii postfix-pcre 3.7.11-0+deb12u1
i386 PCRE map support for Postfix
ii postfix-sqlite 3.7.11-0+deb12u1
i386 SQLite map support for Postfix
In trixie (actually this was not an dist-update but a new installation
to chance from i386 to amd64 and trying to use the old configuration), I
do need to set cyrus_sasl_config_path for it to work:
diff '--color=auto' -uBbwr /old/rootfs/etc/postfix/main.cf
/new/rootfs/etc/postfix/main.cf
--- /old/rootfs/etc/postfix/main.cf 2021-08-17 11:00:08.522464843 +0200
+++ /new/rootfs/etc/postfix/main.cf 2025-05-29 13:42:57.736365566 +0200
@@ -52,7 +54,8 @@
#smtpd_recipient_restrictions = permit_sasl_authenticated,
reject_unauth_destination
smtpd_helo_restrictions = permit_sasl_authenticated,
permit_mynetworks, reject_invalid_hostname, reject_unauth_pipelining,
reject_non_fqdn_hostname
# Postfix >= 2.3, ab Edgy
-#smtpd_sasl_path = smtpd
+smtpd_sasl_path = smtpd
+cyrus_sasl_config_path = /etc/postfix/sasl
#
#smtp_tls_note_starttls_offer = yes
#smtpd_tls_received_header = yes
dpkg --list | grep postfix
ii postfix 3.10.2-1
amd64 High-performance mail transport agent
ii postfix-pcre 3.10.2-1
amd64 PCRE map support for Postfix
I only reported this, because I assume that other people will also run
into this problem. If this is acceptable then this bug can be closed.