[epic] Support causing openvswitch failures to test failure conditions
Tests like HA, HPC, and Multipath require the ability to test that they work when the network conditions are less than ideal
In a simple 2 node cluster for example, this would mean disconnecting one node from it's ports
only openQA knows about which tests are running in parralel, os-autoinst knows about which ports are connected to which vswitch
This issue requires a lockapi extensions so that a test can instruct openQA/os-autoinst to disconnect a test VM from it's network, and then reconnect it.
#4 Updated by okurz over 1 year ago
- Status changed from New to Rejected
- Assignee set to okurz
openSUSE/SLE in the meantime has some tests for "flaky network" but not using openvswitch. The problem here is that while it sounds nice it would need better understanding of test writers what openvswitch would support or our wrapper to control its state and this is not necessarily making test reviewers on failures more easy to understand. I don't think this specific implementation suggestion is helpful for further work. Of course I would like it if someone comes back to the topic from a specific test need and then uses whatever is deemed as the best way of implementing.
See for the reason of tracker change: http://mailman.suse.de/mailman/private/qa-sle/2020-October/002722.html