incorrect access to other_config record of the open_vswitch table in configure_ovs_dpdk() and configure_ovs_hw_offload()

[Bayani Carbone]

In charm-layer-ovn/lib/charms/ovn_charm.py, access to the other_config record of the open_vswitch table is not done using the correct syntax with regards to the format of the data structure.
This affects the methods ``configure_ovs_dpdk`` and ``configure_ovs_hw_offload``.

Currently the following code is used to access the record (this is for the hw_offload case but it's similar for the ovs_dpdk case):

    def configure_ovs_hw_offload(self):
        """Configure hardware offload specific bits in Open vSwitch.

        :returns: Whether something changed
        :rtype: bool
        """
        something_changed = False
        opvs = ch_ovsdb.SimpleOVSDB('ovs-vsctl').open_vswitch
        other_config_fmt = 'other_config:{}'
        for row in opvs:
            for k, v in (('hw-offload', 'true'),
                         ('max-idle', '30000'),
                         ):
                if row.get(other_config_fmt.format(k)) != v:
                    something_changed = True
                    if v:
                        opvs.set('.', other_config_fmt.format(k), v)
                    else:
                        opvs.remove('.', 'other_config', k)
        return something_changed

``row.get(other_config_fmt.format(k))`` assumes that the data structure format is for example:
{'other_config:hw-offload': 'true', 'other_config:max-idle': '30000'}

However, the data structure format is in fact a nested dictionary like this:
{'other_config': {'hw-offload': 'true', 'max-idle': '30000'}}

Thus to access a value in the nested dictionary the following should be used:
``row.get('other_config').get(k)``

This bug does not break anything however, because the condition ``row.get(other_config_fmt.format(k)) != v`` always evaluates to true and since we currently set default values for all options ``opvs.set('.', other_config_fmt.format(k), v)`` is executed in all cases.

However, if we want to use the ``else`` block in the future this will need to be fixed.

For the configure_ovs_dpdk() case a fix is in progress in the following PR:
https://github.com/openstack-charmers/charm-layer-ovn/pull/41

Note: hw_offload configuration test cases will also need to be adapted.

Debug output:
./hooks/config-changed
tmux kill-session -t ovn-chassis-dpdk/2 # or, equivalently, CTRL+a d

4. CTRL+a is tmux prefix.

More help and info is available in the online documentation:
https://discourse.jujucharms.com/t/debugging-charm-hooks

root@node07:/var/lib/juju/agents/unit-ovn-chassis-dpdk-2/charm# vi lib/charms/ovn_charm.py
root@node07:/var/lib/juju/agents/unit-ovn-chassis-dpdk-2/charm# ./hooks/config-changed
none
> /var/lib/juju/agents/unit-ovn-chassis-dpdk-2/charm/lib/charms/ovn_charm.py(518)configure_ovs_dpdk()
-> if row.get('other_config').get(k) != v:
(Pdb) row
{'_uuid': UUID('85e8f98e-2ac3-42c5-811c-1220a28b5948'), 'bridges': UUID('bbc43e5e-79e0-4f9b-a717-337b196154e2'), 'cur_cfg': 28, 'datapath_types': ['netdev', 'system'], 'datapaths': {}, 'db_version': '8.2.0', 'dpdk_initialized': True, 'dpdk_version': 'DPDK 19.11.3', 'external_ids': {'hostname': 'node07.maas', 'ovn-encap-ip': '172.27.12.45', 'ovn-encap-type': 'geneve', 'ovn-remote': 'ssl:172.27.12.206:6642,ssl:172.27.12.198:6642,ssl:172.27.12.193:6642', 'rundir': '/var/run/openvswitch', 'system-id': 'node07.maas'}, 'iface_types': ['dpdk', 'dpdkr', 'dpdkvhostuser', 'dpdkvhostuserclient', 'erspan', 'geneve', 'gre', 'internal', 'ip6erspan', 'ip6gre', 'lisp', 'patch', 'stt', 'system', 'tap', 'vxlan'], 'manager_options': UUID('c909af8e-6046-4394-a84c-9c56eaaafc61'), 'next_cfg': 28, 'other_config': {'dpdk-init': 'true', 'dpdk-lcore-mask': '0x01', 'dpdk-socket-mem': '1024', 'pmd-cpu-mask': '0x9'}, 'ovs_version': '2.13.1', 'ssl': UUID('cef62424-0ed7-484a-b665-893e22a0fb83'), 'statistics': {}, 'system_type': 'ubuntu', 'system_version': '20.04'}
(Pdb) row.get('other_config')
{'dpdk-init': 'true', 'dpdk-lcore-mask': '0x01', 'dpdk-socket-mem': '1024', 'pmd-cpu-mask': '0x9'}
(Pdb) row.get('other_config').get('pmd-cpu-mask')
'0x9'
(Pdb) row.get('other_config:pmd-cpu-mask')
(Pdb)

[Alex Kavanagh (ajkavanagh)]

Triaged: Yes, definitely bug in the code as on master on date of comment.

Schedule to fix ASAP from the equivalent charm-helpers code that Bayani is currently working on.

Thanks for raising the bug, Bayani!

[Frode Nordahl (fnordahl)]

Thank you for the bug and triage. While there is no doubt this is a bug, I'm not quite convinced about the currently triaged importance. The age of the bug and absence of merged fix is a testament to that.

A fix is underway and will be picked up as part of ongoing work scheduled to complete soon.