| Jon Hall | b548801 | 2017-06-21 14:08:36 -0700 | [diff] [blame] | 1 | This test is designed to verify that an ONOS cluster behaves correctly when some |
| 2 | ONOS nodes are stopped. Currently, we will stop nodes so that each raft partition will |
| 3 | lose a member, but we make sure that there is always a majority of nodes |
| 4 | available in each partition. |
| 5 | |
| 6 | As written, the test only supports an ONOS cluster of 3,5, or 7 nodes. |
| 7 | This is because the test doesn't apply to a single node cluster, ONOS clusters |
| 8 | should be deployed in odd numbers, and the partition generation and node |
| 9 | killing scheme used doesn't give the same properties for clusters of more |
| 10 | than 7 nodes. Namely, each partition won't have exactly one node killed. |
| 11 | |
| 12 | This test will repeat killing a single node then recovering multiple times. |
| 13 | |
| 14 | The gerneral structure for the test: |
| 15 | - Startup |
| 16 | - Assign switches |
| 17 | - Verify ONOS state and functionality |
| 18 | - Device mastership |
| 19 | - Intents |
| 20 | - Leadership election |
| 21 | - Distributed Primitives |
| 22 | - Stop some ONOS nodes |
| 23 | - Verify ONOS state and functionality |
| 24 | - Restart nodes |
| 25 | - Verify ONOS state and functionality |
| 26 | - Dataplane failures |
| 27 | - link down and up |
| 28 | - switch down and up |