```
[213] SaltStack recommends using cmd.run together with onchanges, rather than cmd.wait
firewalld/init.sls:53
  cmd.wait:
```

Usually fix:

* `cmd.wait` + `watch`

To:

* `cmd.run` + `onchanges`

But this code doesn't use that construction, so leaving as-is.

* fix(debian10 iptables): install iptables from buster-backports

* fix(debian 10 iptables): Arch Linux test

Co-authored-by: Imran Iqbal <myii@users.noreply.github.com>

fixes #51

The current rich_rule macro is supporting to set if the connection
should be accepted or rejected or dropped but doesn't support setting
rate limiting in the 'accept' case. Add code for that.

Signed-off-by: Arnaud Patard <apatard@hupstream.com>

* Semi-automated using https://github.com/myii/ssf-formula/pull/287

* Semi-automated using https://github.com/myii/ssf-formula/pull/286

Close #44.

Configuration description taken from:

* https://firewalld.org/documentation/man-pages/firewalld.conf.html

While the default maps are mostly empty this sets the groundwork for
distribution specific defaults. The layout is based on the formula
template.

BREAKING CHANGE: `map.jinja` has been upgraded from using `pillar.get`
to `config.get`.

If the formula was ran on a system that reported an os_family that
wasn't one of "Debian", "RedHat", "Arch", "Suse" then the map.jinja
template would fail to render with "'NoneType' is not iterable.

This occurs because grains.filter_by will return None when it fails
match the grain to the input dictionary. The value is then blindly
passed into a dict.update() which causes the failure.

In this patch we ensure that the default values, as defined in
defaults.yaml, are always applied when grain matching fails.

* Semi-automated using https://github.com/myii/ssf-formula/pull/159

When specifying rich rules as a dictionary ipsets and services can be
specified as lists. They will be expanded out by the jinja template into
individual rich rules for the parent zone.

Now multiple sections ending with `services` can be defined for each in pillar
and all of them will get merged into one service block in the zone. The goal is
to keep backward compatibility while allowing different services to be defined
in different pillars. So basically have various parts of the pillar affecting
the firewall without need to define everything centrally. Helpful for the
exceptions to the rules.

```bash
firewalld-formula$ yamllint -s .
./pillar.example
  2:1       warning  missing document start "---"  (document-start)
  3:12      warning  truthy value should be one of [false, true]  (truthy)
  12:13     warning  truthy value should be one of [false, true]  (truthy)
  15:89     error    line too long (108 > 88 characters)  (line-length)
  19:13     warning  truthy value should be one of [false, true]  (truthy)
  30:89     error    line too long (363 > 88 characters)  (line-length)
  96:89     error    line too long (170 > 88 characters)  (line-length)
  108:15    error    wrong indentation: expected 12 but found 14  (indentation)
  115:13    error    empty value in block mapping  (empty-values)
  116:2     error    syntax error: found character '%' that cannot start any token
  152:89    error    line too long (112 > 88 characters)  (line-length)

./firewalld/defaults.yaml
  3:1       warning  missing document start "---"  (document-start)
```

```bash
Examining firewalld/map.jinja of type state
[209] Jinja comment should have spaces before and after: {# comment #}
firewalld/map.jinja:4
{## Start with  defaults from defaults.yaml ##}

[209] Jinja comment should have spaces before and after: {# comment #}
firewalld/map.jinja:7
{##

[201] Trailing whitespace
firewalld/map.jinja:16
  }, grain='os_family', merge=salt['pillar.get']('firewalld:lookup'))

[209] Jinja comment should have spaces before and after: {# comment #}
firewalld/map.jinja:19
{## Merge the flavor_map to the default settings ##}

[209] Jinja comment should have spaces before and after: {# comment #}
firewalld/map.jinja:22
{## Merge in salt:lookup pillar ##}
```

Firewalld does not parse the name attribute.

Log message:
firewalld[1999]: ERROR: Failed to load zone file 'public.xml':
PARSE_ERROR: icmp-block-inversion: Unexpected attribute name

update kitchen travis
fix ipsets.sls

See https://github.com/saltstack-formulas/firewalld-formula/pull/21#pullrequestreview-146958098

See https://github.com/saltstack-formulas/firewalld-formula/pull/21#pullrequestreview-146958098

I wanted to be able to add an ipset as a source in the zone without using a rich rule.  I believe this change accomplishes that.  Tested and working on CentOS 7 (salt master and minion).

human readable