Hello,

first of all: happy new year! ;-)

A bug report was submitted today [1], stating that, on CentOS 7 VMs, the
temporary enabling and disabling of SELinux that ElastiCluster does
(e.g., when enabling iptables firewall rules) renders `journald`
non-functional -- and possibly creates other problems too.

A workaround [2] posted on ServerFault is also reported to cause trouble
in the same issue report.

I do not know enough of SELinux myself to be able to take an informed
decision here.  As far as I can see there are these options:

1. Just disable SELinux altogether (`setenforce 0`) at the start of
   ElastiCluster playbooks.  This makes life simpler for anyone (well,
   makes *my* life simpler at least) but may be not what experienced
   CentOS/RHEL admins expect?  Also, is somebody replying on SELinux in
   production clusters built with ElastiCluster?

2. Try to use a workaround like `restorecon -r /` (assuming one exists
   that works reliably).  I have no idea what this workaround can be, though.

3. Try to do things correctly "the SELinux way". Last time I checked Red
   Hat's docs this involved rebooting the VM, which is not something we
   can do in the middle of an Ansible playbook. But maybe I read wrong?

Any opinions?

[1]: https://github.com/gc3-uzh-ch/elasticluster/issues/370
[2]: 
http://serverfault.com/questions/764687/systemd-journald-fails-to-start-on-centos-7

Ciao,
R

-- 
Riccardo Murri, Schwerzenbacherstrasse 2, CH-8606 Nänikon, Switzerland

-- 
You received this message because you are subscribed to the Google Groups 
"elasticluster" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to