Seems to be required in newer Amarisoft versions

This role makes a request to shacache to check if new amarisoft
binaries are available. If there are and if the license on the
machine allows it the binaries will be downloaded and decrypted
using the server's private key in /opt/private-key.
The binaries will then be installed in
/opt/amarisoft/vYYYY-MM-DD, to be used with ors-amarisoft
software release.

This will generate a key pair in /opt/private-key.
The public key will then be published in ansible report to
the slapos monitoring service running on the machine.
This private key will be used to decrypt encrypted
data uploaded on shacache.

We had several cases of playbook-tmp becoming empty when
machines were unplugged at the wrong time. This happenned
because data was still in the buffer when machines were
unplugged.

The solution is to do a sync and then atomically change
playbook-tmp

fixup for 203927d9

ansible will exit if there is a sha256sum mismatch, no
need to check rc code

When using IPv6 between gNB and MME on different
ORS, TCP throughput between phones with MTU 1500
and ORS is very bad.
Setting MTU to 1500 on lo (where the eNB / MME GTP
addresses are) fixes the issue.

See merge request nexedi/slapos.package!190

Since switch to Zope 4 HTML content of default Zope root page changed. Thus adjust accordingly.

extra-vars cannot be overrided directly so the "set_fact"
when extra playbook is ors.yml was not doing anything

This config was sending all logs to localhost:42185 in UDP but nothing
was listening on this port (it was supposed to be fluentd to aggregate
logs in Wendelin).

This was actually creating a weird scenario where qemu was temporarily
listening on UDP on localhost:42185 and receive the packet from
rsyslogd (which was transmitted to the guest machine and considered a
martian packet there)

Otherwise if playbook stops after removing ifupdown
then the ORS becomes unreachable because ors-image-backports
playbook might not get executed properly to configure
systemd-networkd