Refine handle Intra-domain link up/down

[congwang09]

As discussed in the AHM, we should discuss on how we handle the RED l2vpn in usecase 1:
https://sdx-docs--58.org.readthedocs.build/en/58/network-use-cases.html#use-case-1-intra-domain-link-up-down

[italovalcy]

The RED L2VPN in Use Case 1 can be described as follows:

Suppose you have a L2VPN created between endpoints on different domains (inter-domain L2VPN) -- for Use Case 1, it was created from Ampath3 port 50 to Sax01 port 50. Suppose that one domain had multiple failures on intra-domain links, for instance on Ampath domain we had failures on links Ampath1--Ampath3 and Ampath1--Ampath2. The current implemented logic for handling link failures determine that: we do not handle intra-domain link failures, since it is responsibility of the OXP to find paths for those cases. However, for this particular use case, we may a situation where the breakdown was established using a path that is not available anymore and the OXP alone cannot recovery.

The observed behavior was: the L2VPN service was impacted (i.e., no connectivity anymore from Ampath3 port 50 to Sax01 port 50 -- permanent failure); the L2VPN status continue to display as UP.

From what we discussed during the All Hands Meeting, the following behavior/changes should be implemented:

• The SDX-LC should query the OXPO in regular intervals to check the status of the per-domain L2VPNs (i.e., individual breakdowns) and update the SDX L2VPN accordingly. For example: if one L2VPN was created on SDX-Controller and that L2VPN resulted on the creation of 3 child L2PVN / breakdowns, one for each domain, any failure on any of the breakdowns will be reflected on a failure on the "parent" L2VPN (i.e, status = down).
• On L2VPN status change (inter-domain L2VPNs only! TODO: what about L2VPNs that could be created outside of the OXP? for instance the green l2vpn on use case 1 above), SDX-Controller should try to find another path if possible to reprovision the L2PVN. If another PATH is found, SDX-Controller should request the removal of the previous breakdowns and send requests to create new ones.

@jab1982 please let me know if the blueprint above seems reasonable for you or please feel free to augment/change the suggested behavior.

[jab1982]

Agreed with both suggestions. Please proceed with it.

[YufengXin]

Is there an end-to-end test case for this? If so, could you report back the test result?

[italovalcy]

Is there an end-to-end test case for this? If so, could you report back the test result?

There isn't, @YufengXin. Please feel free to create one. We demonstrated this during last AHM.

[italovalcy]

Hi everybody, since the new requirements were updated and reviewed by the team, I will change the ownership back to Yufeng so he can assign someone to start implementing those requirements.

[jab1982]

@YufengXin , this issue is critical to our operation. Can you work on this issue please? If a deeper discussion is needed, let's schedule time to go through the details.