[Spacewalk-list] Cannot get Jabberd services running on Spacewalk-Proxy 2.5 running on CentOS 6 docker container

[Andy Warring]

Hi,

We run a Spacewalk 2.5 Master server and multiple Proxies, all running
CentOS 6.9 on a mix of VM-ware and AWS AMI’s.

These work very nicely but we have been investigating the possibility of
moving the Proxy servers into a Docker container.

We came across this on GitHub -
https://github.com/Varhoo/docker-spacewalk-proxy

The Host is running CentOS 7.4.1708 - 3.10.0-693.21.1.el7.x86_64
The container is running CentOS 6.9  (we originally tried a CentOS 7
container, but despite our efforts, we couldn’t get systemd running) ( we
followed these instructions -
https://stackoverflow.com/questions/28495341/start-a-service-in-docker-container-failed-with-error-failed-to-get-d-bus-conne
without success). So we looked at a CentOS 6 container which kind of
mirrored our existing Proxies.

Docker = Docker version 17.12.1-ce, build 7390fc6
SELinux = Host = Permissive, Container = Disabled.

So, we haven’t taken the Dockerfile like for like, as this is for Fedora
but we created our own Dockerfile to use CentOS 6, we install packages,
register to our Spacewalk server (via another proxy) and run the
configure-proxy.sh script.

All appears to be OK until we run:

[root at dockerproxy02 /]# rhn-proxy status
squid (pid  775) is running...
httpd (pid  794) is running...
router dead but subsys locked
sm dead but subsys locked
c2s dead but subsys locked
s2s dead but subsys locked

The Jabberd PID’s appear to be locked, a restart doesn’t help. If we
manually kill the PID’s we can then do a ‘service jabberd start’ but
immediately they do back to being ‘dead by subsys locked’

I have checked the host file for a valid FQDN, the jabberd config files
(c2s.xml, s2s.xml etc), cpu and memory usage of the container is very low.
I’ve compared the files from a working VM proxy to a container proxy and
the likes of the /etc/rhn/rhn.conf, /etc/squid/squid.conf appear to be
correct (the httpd and squid services do appear to be running)

On the container (we enabled rsyslog), /var/log/messages shows the jabberd
processes starting and no errors are being reported.

We are kind of stumped now, has anyone else got a working Spacewalk-Proxy
running in a docker container?

Is there a way of turning on some debug logs for Jabberd to try and
identify the issue?

Can a Proxy run with just squid and httpd (if we didn’t care about the push
actions), if we just wanted clients to register to Spacewalk and download
packages when required.

Would a newer version of Spacewalk / Spacewalk Proxy resolve this?

Thank you for any help/advice with this.

Andy Warring
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20180316/1c66b31e/attachment.htm>