[Podman] Re: Can't maintain connection to container's listening port

12 Jan 2022

      On Wed, 2022-01-12 at 15:25 -0500, Daniel Walsh wrote:
...
Try the container in --privileged mode, to see if this is causing the
issue.
That didn't help. :(
...
If it still crashes, then I figure it is something with the user
namespace.
The container doesn't crash. It actually runs and I can see postgrey
running in the container. The problem is when I do a telnet test from
the host or from another server to the postgrey port I exposed, the
telnet connection doesn't stay up and instead drops immediately. If the
container is running rootfull, the telnet test is established and
doesn't disconnect.
...
Do you have an image I could try?
It's not uploaded anywhere. I'm turning some of my KVMs into containers
because I wanted to learn podman and get familiar with it. I've only
been doing container stuff in general since August of last year.

Hmmm...can I dump the Containerfile for each image here? They're not
big. The Rocky Linux 8 Dockerfile is copied directly from their docker
image page on dockerhub. 

There's some wrapping.

Here's the latest run command I used:

"CreateCommand": [
                "podman",
                "run",
                "-d",
                "--name",
                "postgrey",
                "--publish",
                "1.2.3.4:10023:10023",
                "--volume",
                "postgrey:/var/spool/postfix/postgrey:Z",
                "--privileged",
                "postgrey-v0.0.3"
            ],

The volume doesn't matter. I'm just preserving data from my KVM.

Below are the Containerfile, config file and Dockerfile.

--------
postgrey
--------

FROM local/rocky8-systemd

RUN dnf -y update && \
    dnf -y install epel-release && \
    dnf -y --nodocs install postgrey telnet && \
    dnf clean all && \
    systemctl enable postgrey

COPY postgrey /etc/sysconfig/

CMD ["/usr/sbin/init"]

--------------------
postgrey sysconfig file
--------------------

# Postgrey offers 2 listening types, --inet and --unix. As default, Fedora
# postgrey works under UNIX socket, but, changing to TCP socket on user's own
# is also available, for instance, let it work at 10023 port of localhost:
#   --inet=10023
# To be more detailed, there is another way if you still run it at localhost:
#   --inet=127.0.0.1:10023
#POSTGREY_TYPE="--unix=/var/spool/postfix/postgrey/socket"
POSTGREY_TYPE="--inet=127.0.0.1:10023"

# If postgrey works under UNIX socket way, PID file can be specified to
# custom location, note that no need to set this if postgrey is working
# under TCP socket way.
POSTGREY_PID="--pidfile=/var/run/postgrey.pid"

# Name of group which postgrey belongs, default is postgrey
POSTGREY_GROUP="--group=postgrey"

# Name of user which postgrey belongs, default is postgrey
POSTGREY_USER="--user=postgrey"

# DELAY
POSTGREY_DELAY="--delay=60"

# For more options can be used, please read manpage or execute `postgrey -h`.
# Custom options.
POSTGREY_OPTS=""

-------------
rocky linux 8
-------------

FROM rockylinux/rockylinux:latest

ENV container docker

RUN (cd /lib/systemd/system/sysinit.target.wants/; for i in *; do [ $i
== \
	systemd-tmpfiles-setup.service ] || rm -f $i; done); \
	rm -f /lib/systemd/system/multi-user.target.wants/*;\
	rm -f /etc/systemd/system/*.wants/*;\
	rm -f /lib/systemd/system/local-fs.target.wants/*; \
	rm -f /lib/systemd/system/sockets.target.wants/*udev*; \
	rm -f /lib/systemd/system/sockets.target.wants/*initctl*; \
	rm -f /lib/systemd/system/basic.target.wants/*;\
	rm -f /lib/systemd/system/anaconda.target.wants/*;

VOLUME [ "/sys/fs/cgroup" ]

CMD ["/usr/sbin/init"]

-- 
Ranbir