[TRNSPRTPCE-86] Node disconnect tests in test_topology.py fail when trying to delete node in the openroadm-network Created: 12/Feb/19 Updated: 03/Sep/21 Resolved: 29/Apr/19 |
|
| Status: | Verified |
| Project: | transportpce |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Medium |
| Reporter: | Jonas Mårtensson | Assignee: | Christophe BETOULE |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | 0 minutes | ||
| Time Spent: | 3 days | ||
| Original Estimate: | Not Specified | ||
| Description |
|
It seems that when deleting a node in the topology-netconf, it is also deleted in the openroadm-network. Therefore, when later trying to delete it in the openroadm-network it does not exist anymore and a 404 is returned. |
| Comments |
| Comment by Ahmed Triki [ 13/Feb/19 ] |
|
ojnasjonas Can you please check if the new code resolve the problem? |
| Comment by Jonas Mårtensson [ 13/Feb/19 ] |
|
The patch does not solve the issue since nodes are still deleted in the openroadm-network after being deleted in the topology-netconf. Note that the problem does not always happen since it may take some time before nodes are really deleted but if you introduce a few seconds delay before deleting nodes in the openroadm-network you will see that the nodes does not exist anymore causing the test to fail with a 404. More tests fail with this patch since nodes are not deleted in the clli-network after the failure to delete nodes in the openroadm-network. This causes some later getClliNetwork tests to fail. |
| Comment by Guillaume Lambert [ 14/Feb/19 ] |
|
OK, so Ahmed maybe the first action to take is to modify the topology test suite to add this timer and expose the problem and then try to investigate with colleagues what is wrong in the current workflow before proposing a fix ( should the 2 actions be merged or should we fix the java code ? ) |
| Comment by Ahmed Triki [ 18/Feb/19 ] |
|
As you can see in the patch 4, I have added a timer to be sure that the node disconnection process is well taken into account, so I get the bug as it is described by Jonas. When nodes are disconnected at the topology-netconf layer, they will be automatically disconnected at the openroadm-network as well. Therefore, they will not be found by ODL and the test fails. So I delete the part that concerns the node disconnection at the openroam-network. However, at the end of the test suite when all the nodes are disconnected, the roadm-to-roadm links are still exist (failure of the last test). We will check with the colleagues if this is normal! |
| Comment by Olivier Renais [ 14/Mar/19 ] |
|
The fact that the node is automatically removed from network topollogy when it disapear from the netconf topology fits with expected behavior. So I guess we should modify or remove the test to fit with this.expected behavior. for what concerns link, we expect something different. The link state shall change, but the link shall not disappear, because the link is built in a two srtep process. Discovery (leveraging lldp), and consolidation, checking consistency with data that describe a link and are not known by the device (SRLG, PMD,...). So we shall not test that the link does or does not exist. Note that they are currently user stories adressing the process of OMS information consolidation (TRNSPRTPCE-80) |