Easier Kea sabotage (demo only)
Ok, sabotage is maybe not the best word, but now that I have your attention, here's what the problem is:
we want to kill and later revive Kea server that is working in HA pair, so the HA problems demonstration is easier.
During the demo I did the following:
- docker-compose exec agent-kea-ha2 /bin/bash
- edit /etc/supervisor.conf. In [kea-dhcp4] section I modified autorestart = false
- kill 1 (supervisor process). This restarted the container (and kicked me out in the process)
- I was now able to get in again (docker-compose ...) and then kill kea: killall kea-dhcp4 <observe how Stork detects HA failures, wait a bit>
- start kea with kea-dhcp4 -c /etc/kea/kea-dhcp4.conf
The absolute minimum we could do is this:
- tweak docker recipes to build the containers with autorestart disabled
- maybe tweak the DHCP traffic generator, to make it easier to start/stop the service.
Alternatively, we should describe somewhere how non-developer can stop and start Kea service in the container. I don't know, maybe it would be easier to figure some easy way to stop/start the whole container altogether?
In any case there should be some wiki update regarding how to do it.